EPA/600/R-04/080
August 2004
Analysis of Mercury in
Vermont and New Hampshire Lakes:
Evaluation of the
Regional Mercury Cycling Model
By
Christopher D. Knightes and Robert B. Ambrose, Jr.
Ecosystems Research Division
Athens, GA 30605
National Exposure Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
-------
Evaluating R-MCMfor 91 VT/NH Lakes
NOTICE
The U.S. Environmental Protection Agency (EPA), through its Office of Research and
Development (ORD), partially funded and managed the research described herein under
cooperative agreement CR 825495-01 with the Vermont Department of Environmental
Conservation (VT DEC) under the Regional Environmental Monitoring and Assessment
Program (REMAP). This work has been subjected to the Agency's peer and
administrative review, and has been approved for publication as an EPA document.
Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
-------
Evaluating R-MCMfor 91 VT/NH Lakes
ABSTRACT
An evaluation of the Regional Mercury Cycling Model (R-MCM, a steady-state fate and
transport model used to simulate mercury concentrations in lakes) is presented based on
its application to a series of 91 lakes in Vermont and New Hampshire. Visual and
statistical analyses are presented in an effort to investigate both the behavior of the model
as well as the model's ability to predict the observed mercury concentrations in the water
column, sediments and fish tissue. The sensitivity of the model to certain parameters and
processes was also evaluated. A comparison of model trends to the observed trends was
made. These investigations provide further insight into the complications and challenges
that surround modeling the fate and transport of mercury within a given water body, and
understanding the exposure concentrations of mercury in the surrounding ecosystem via
mercury bioaccumulation in the aquatic food web (e.g., fish) and its transfer to
piscivorous wildlife and humans.
-------
Evaluating R-MCMfor 91 VT/NH Lakes
FOREWORD
The National Exposure Research Laboratory Ecosystems Research Division
(ERD) in Athens, Georgia, conducts process, modeling, and field research to assess the
exposure risks of humans and ecosystems to both chemical and non-chemical stressors.
This research provides data, modeling, tools, and technical support to EPA Program and
Regional Offices, state and local governments, and other customers, enabling
achievement of Agency and ORD strategic goals for the protection of human health and
the environment.
ERD research includes studies of the behavior of contaminants, nutrients, and
biota in environmental systems, and the development of mathematical models to assess
the response of aquatic systems, watersheds, and landscapes to stresses from natural and
anthropogenic sources. ERD field and laboratory studies support process research, model
development, testing and validation, and the characterization of variability and prediction
uncertainty.
Leading-edge computational technologies are developed to integrate core science
research results into multi-media (air, surface water, ground water, soil, sediment, biota),
multi-stressor, and multi-scale (organism, population, community, ecosystem; field site,
watershed, regional, national, global) modeling systems that provide predictive
capabilities for complex environmental exposure scenarios face by the Agency.
Exposure models are distributed and supported via the EPA Center for Exposure
Assessment Modeling (CEAM) (www.epa.gov/athens/ceampubl), the Watershed and
Water Quality Model Technical Support Center (www.epa. gov/athens/wwqtsc), and
through access to Internet tools (www.epa.gov/athens/onsite).
This research project is a component of the ERD mercury research program,
which seeks to better understand the environmental cycling of the major speciated forms
of mercury, especially the characteristics that induce mercury methylation in ecosystems
and the pathways of exposure. In this project, the Regional Mercury Cycling Model was
applied to a large set of lakes in Vermont and New Hampshire. The goals were to better
understand mercury transformation and bioaccumulation processes and to evaluate our
present ability to predict mercury fate. Knowledge and data gained in this evaluation will
be used to develop or improve mercury analysis capabilities in existing models used by
EPA in various regulatory programs.
Rosemarie C. Russo, Ph.D.
Director
Ecosystems Research Division
Athens, Georgia
-------
Evaluating R-MCMfor 91 VT/NH Lakes
ACKNOWLEDGEMENT
This modeling study would not have been possible without the comprehensive regional
mercury database created by the Vermont Agency of Natural Resources and documented
in a companion report. Neil Kamman served as principal investigator for this project, and
we gratefully acknowledge his advice and insight. Dr. Eric Miller provided insight on
mercury deposition over the Vermont and New Hampshire region, along with personal
communications, which we greatly appreciate. Dr. Reed Harris, a senior engineer at
Tetra Tech, is the principal author of the Regional Mercury Cycling Model, which we
applied in this study. Dr. Harris provided useful suggestions and comments throughout
this study, which we appreciate. We also acknowledge the faithful modeling work
performed by Adrienne Harris, a summer intern from Spelman College, now at Duke
University. Her work is incorporated into Chapter 3 and Sections 6.3 and 6.4.
-------
Evaluating R-MCMfor 91 VT/NH Lakes
TABLE OF CONTENTS
TABLE OF CONTENTS i
LIST OF TABLES Hi
LIST OF FIGURES viii
EXECUTIVE SUMMARY. xii
1 INTRODUCTION /
2 METHODS 6
2.1 Experimental Dataset 6
2.2 Regional Mercury Cycling Model (R-MCM) 7
2.3 Application of the Model 9
2.3.1 Tierl 11
2.3.2 Tier 2 12
2.3.3 Tier 3 12
2.3.4 Tier 4 13
2.3.5 TierS 13
3 EVALUATION OF R-MCM: DEFAULT MODEL RESULTS AND TRENDS.... 14
3.1 Default Model Output 14
3.2 Trends in the Default Run Data Output 15
3.2.1 Lake Area (Lake Size) 16
3.2.2 Epilimnion DOC (Trophic Status) 17
3.2.3 Epilimnion pH (Acidity) 18
3.2.4 Lake Stratification 18
4 EVALUATION OF R-MCM: ENTIRE LAKE DATA SET 19
4.1 General Visual Inspection 19
4.2 Error Sum of Squares Analysis 23
4.3 Summary 24
5 EVALUATION OF R-MCM: LAKE CHARACTERISTICS 26
5.1 Visual Analysis 26
5.1.1 Acidity 26
5.1.2 Stratification 27
5.1.3 Lake Size 30
5.1.4 Trophic Status 31
5.1.5 Summary of Visual Analysis 34
5.2 Statistical Evaluation of Model Successes and Inadequacies 35
5.2.7 Chi-Square Goodness of Fit. 36
5.2.2 T-Test on the Mean of the Residuals 40
-------
Evaluating R-MCMfor 91 VT/NH Lakes
5.3 Model Performance Statistics 47
5.3.1 Maximum Error 48
5.3.2 Root Mean Square Error 49
5.3.3 Coefficient of Determination 49
5.3.4 Modeling Efficiency 52
5.3.5 Coefficient of Residual Mass 53
5.3.6 Model Performance Statistics Summary 54
5.4 Summary 56
6 MODEL SENSITIVITY AND SYSTEM EVALUATION. 58
6.1 Evaluation of Loss Rates: Effect of Photoreduction and Particle Settling 58
6.1.1 Settling Velocity 59
6.1.2 Photodegradation 59
6.1.3 Settling Velocity and Photodegradation 59
6.1.4 Summary of Evaluation of Loss Rates 60
6.2 Sensitivity Evaluation of Hypolimnion Surface Area 60
6.2.1 Visual Analysis and Maximum and Absolute Changes 61
6.2.2 Non-dimensional Model Sensitivity Analysis 64
6.2.3 Re-evaluating Hypolimnion Area Sensitivity by Keeping Constant Volumetric
Flow Rate (Adjusting dwith V) 72
6.2.4 Investigation of Mechanism Causing Increase in Mercury Concentration with
Decreasing Hypolimnion Surface Area 74
6.2.5 Hypothetical Lake Evaluation of the Change in Simulated Mercury Species
Concentrations as a Function of Hypolimnion Area: A Simple Mathematical
Thought Experiment 77
6.2.6 Summary 82
6.3 Comparison of Model Behavior and Observed Data 83
6.3.1 Trends in ObservedData 84
6.3.2 Percent Methylmercury for Observed and Predicted Data 86
6.3.3 Summary 90
6.4 Watershed Influences and Loading 94
6.4.1 Investigation of the Watershed Element Mercury Outflow Parameters 97
7 CONCLUSIONS 103
8 REFERENCES 110
11
-------
Evaluating R-MCMfor 91 VT/NH Lakes
LIST OF TABLES
Table 2.1. Lake Categories and Characteristics.
Table 2.2. Lake Names and Input Characteristics for Default Set-Up.
Table 2.3. Tiers 1 through 5 and their Associated Default Input Values and Lake
Characteristics.
Table 2.4. Parameter Updates for Tier 1.
Table 2.5. Parameter Updates for Tier 3.
Table 2.6. Parameter Updates for Tier 4.
Table 2.7. Parameter Updates for Tier 5.
Table 3.1. Combinations of Characteristics for Default Runs of R-MCM.
Table 3.2. Predicted Results for Combination of Default Lake Characteristics.
Table 3.3. Summary of Results for the Combination of Default Lakes Run by R-MCM.
Table 3.4. Examples of Literature Published Ranges for Mercury Concentrations in
Different Media in the Environment.
Table 5.1. Model Performance Statistics for Default Scenario: %2, Reference %2 at 90%
Confidence, Number of Observations.
Table 5.2. Model Performance Statistics for Tier 1 Scenario: Chi-Square Value,
Reference Chi-Square Value at 90% Confidence, number of observations
Table 5.3. Model Performance Statistics for Tier 2 Scenario: %2, Reference %2 at 90%
Confidence, Number of Observations.
Table 5.4. Model Performance Statistics for Tier 3 Scenario: %2, Reference %2 at 90%
Confidence, Number of Observations.
Table 5.5. Model Performance Statistics for Tier 4 Scenario: %2, Reference %2 at 90%
Confidence, Number of Observations.
Table 5.6. Model Performance Statistics for Tier 5 Scenario: %2, Reference %2 at 90%
Confidence, Number of Observations.
Table 5.7. Model Performance Statistics for Default Scenario: T-Test Value, Reference
Value at 90% Confidence, Number of Observations, Mean Residual, and Standard
Deviation.
in
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.8. Model Performance Statistics for Tier 1 Scenario: T-Test Value, Reference
Value at 90% Confidence, Number of Observations, Mean Residual, and Standard
Deviation.
Table 5.9. Model Performance Statistics for Tier 2 Scenario: T-Test Value, Reference
Value at 90% Confidence, Number of Observations, Mean Residual, and Standard
Deviation.
Table 5.10. Model Performance Statistics for Tier 3 Scenario: T-Test Value, Reference
Value at 90% Confidence, Number of Observations, Mean Residual, and Standard
Deviation.
Table 5.11. Model Performance Statistics for Tier 4 Scenario: T-Test Value, Reference
Value at 90% Confidence, Number of Observations, Mean Residual, and Standard
Deviation.
Table 5.12. Model Performance Statistics for Tier 5 Scenario: T-Test Value, Reference
Value at 90% Confidence, Number of Observations, Mean Residual, and Standard
Deviation.
Table 5.13. Model Performance Statistics for Default Scenario: Maximum Error and
Root Mean Square Error.
Table 5.14. Model Performance Statistics for Default Scenario: Coefficient of
Determination, Modeling Efficiency, and Coefficient of Residual Mass.
Table 5.15. Model Performance Statistics for Tier 1 Scenario: Maximum Error and Root
Mean Square Error.
Table 5.16. Model Performance Statistics for Tier 1 Scenario: Coefficient of
Determination, Modeling Efficiency, and Coefficient of Residual Mass.
Table 5.17. Model Performance Statistics for Tier 2 Scenario: Maximum Error and Root
Mean Square Error.
Table 5.18. Model Performance Statistics for Tier 2 Scenario: Coefficient of
Determination, Modeling Efficiency, and Coefficient of Residual Mass.
Table 5.19. Model Performance Statistics for Tier 3 Scenario: Maximum Error and Root
Mean Square Error.
Table 5.20. Model Performance Statistics for Tier 3 Scenario: Coefficient of
Determination, Modeling Efficiency, and Coefficient of Residual Mass.
Table 5.21. Model Performance Statistics for Tier 4 Scenario: Maximum Error and Root
Mean Square Error.
IV
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.22. Model Performance Statistics for Tier 4 Scenario: Coefficient of
Determination, Modeling Efficiency, and Coefficient of Residual Mass.
Table 5.23. Model Performance Statistics for Tier 5 Scenario: Maximum Error and Root
Mean Square Error.
Table 5.24. Model Performance Statistics for Tier 5 Scenario: Coefficient of
Determination, Modeling Efficiency, and Coefficient of Residual Mass.
Table 5.25. Formulas for Statistical Evaluation Parameters.
Table 6.1. Hypolimnion Surface Area Sensitivity Analysis Results.
Table 6.2. Polynomial Linear Regression Results, Coefficients, Standard Errors,
Adjusted R2, and F Significance.
Table 6.3. Hypolimnion Area Sensitivity Evaluation. Hypothetical Default Lake
Conditions and Results: Changing R and DH, Constant Q.
Table 6.4. Hypolimnion Area Sensitivity Evaluation. Hypothetical Default Lake
Conditions and Results: Changing R, Constant Q.
Table 6.5. Results for Mathematical Analysis of Simple, Stratified Lake System
Table 6.6. Comparison of Summary Statistics for Percent Methylmercury: Predicted and
Observed Results.
Table 6.7. Minima for Error Sum of Squares and Associated Standard Deviation and
Their Associated RlUp and R2Up Values.
Table A-l. Observed Results from the VT/NH REMAP Study.
Table A-2. Predicted Results for the Default Run.
Table A-3. Predicted Results for Tier 1.
Table A-4. Predicted Results for Tier 2.
Table A-5. Predicted Results for Tier 3.
Table A-6. Predicted Results for Tier 4.
Table A-7. Predicted Results for Tier 5.
Table A-8. Predicted Epilimnetic Methylmercury Concentrations for the Hypolimnion
Surface Area Sensitivity Analysis.
Table A-9. Predicted Epilimnetic Total Mercury Concentrations for the Hypolimnion
Surface Area Sensitivity Analysis.
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-10. Predicted Hypolimnetic Methylmercury Concentrations for the
Hypolimnion Surface Area Sensitivity Analysis.
Table A-ll. Predicted Hypolimnetic Total Mercury Concentrations for the Hypolimnion
Surface Area Sensitivity Analysis.
Table A-12. Predicted Sediment Methylmercury Concentrations for the Hypolimnion
Surface Area Sensitivity Analysis.
Table A-13. Predicted Sediment Total Mercury Concentrations for the Hypolimnion
Surface Area Sensitivity Analysis.
Table A-14. Predicted Fish Mercury Concentrations for the Hypolimnion Surface Area
Sensitivity Analysis.
Table A-15. Epilimnetic Methylmercury Concentration Sensitivity to Change in
Hypolimnion Surface Area/Epilimnion Epilimnion Area.
Table A-16. Epilimnetic Total Mercury Concentration Sensitivity to Change in
Hypolimnion Surface Area/Epilimnion Epilimnion Area.
Table A-17. Hypolimnetic Methylmercury Concentration Sensitivity to Change in
Hypolimnion Surface Area/Epilimnion Epilimnion Area.
Table A-18. Hypolimnetic Total Mercury Concentration Sensitivity to Change in
Hypolimnion Surface Area/Epilimnion Epilimnion Area.
Table A-19. Sediment Methylmercury Concentration Sensitivity to Change in
Hypolimnion Surface Area/Epilimnion Epilimnion Area.
Table A-20. Sediment Total Mercury Concentration Sensitivity to Change in
Hypolimnion Surface Area/Epilimnion Epilimnion Area.
Table A-21. Fish Tissue Mercury Concentration Sensitivity to Change in Hypolimnion
Surface Area/Epilimnion Epilimnion Area.
Table A-22. Epilimnion MeHg and HgT Concentrations for Range of RlUp and R2Up
Values.
Table A-23. Hypolimnion MeHg and HgT Concentrations for Range of RlUp and R2Up
Values.
Table A-24. Sediment MeHg and HgT Concentrations for Range of RlUp and R2Up
Values.
VI
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-25. Fish Tissue Hg Concentrations for Range of RlUp and R2Up Values.
vn
-------
Evaluating R-MCMfor 91 VT/NH Lakes
LIST OF FIGURES
Figure 3.1. Default Modeling Outputs for All Combinations of Drainage Lakes
(Epilimnion and Hypolimnion Mercury Concentrations).
Figure 3.2. Default Modeling Outputs for All Combination of Drainage Lakes (Sediment
and Fish Mercury Concentrations).
Figure 4.1. Predicted Cone, vs Measured Cone, of Lake Variables for Default Run.
Dashed line is y = x.
Figure 4.2. Predicted Cone, vs Measured Cone, of Lake Variables for Tier 1 Run.
Dashed line is y = x.
Figure 4.3. Predicted Cone, vs Measured Cone, of Lake Variables for Tier 2 Run.
Dashed line is y = x.
Figure 4.4. Predicted Cone, vs Measured Cone, of Lake Variables for Tier 3 Run.
Dashed line is y = x.
Figure 4.5. Predicted Cone, vs Measured Cone, of Lake Variables for Tier 4 Run.
Dashed line is y = x.
Figure 4.6. Predicted Cone, vs Measured Cone, of Lake Variables for Tier 5 Run.
Dashed line is y = x.
Figure 4.7. Error Sum of Squares for Runs and Variables.
Figure 5.1. Default Results. Lakes Separated by Acidity, o: Acidic, x: Circumneutral, +:
Alkaline.
Figure 5.2. Default Results. Lakes Separated by Stratification: o: Well Mixed, x:
Stratification.
Figure 5.3. Default Results. Lakes Separated by Lake Size: o: Small, x: Medium.
Figure 5.4. Default Results. Lakes Separated by Trophic Status: o: Oligotrophic, x:
Mesotrophic, +: Eutrophic, *:Dystrophic.
Figure 5.5. Tier 5 Results. Lakes Separated by Acidity, o: Acidic, x: Circumneutral, +:
Alkaline.
Figure 5.6. Tier 5 Results. Lakes Separated by Stratification: o: Well Mixed, x:
Stratification.
Figure 5.7. Tier 5 Results. Lakes Separated by Lake Size: o: Small, x: Medium.
Vlll
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Figure 5.8. Tier 5 Results. Lakes Separated by Trophic Status: o: Oligotrophic, x:
Mesotrophic, +: Eutrophic, *:Dystrophic.
Figure 6.1. Effects of Settling Velocity on Mercury Concentrations.
Figure 6.2. Effects of Photoreduction on Mercury Concentrations.
Figure 6.3. Combined Effects of Photoreduction and Settling Velocity on Mercury
Concentrations.
Figure 6.4. Predicted vs. Observed Epilimnion and Hypolimnion Mercury
Concentrations for the Hypolimnion Area Sensitivity Runs.
Figure 6.5. Predicted vs. Observed Sediment and Fish Tissue Mercury Concentrations
for the Hypolimnion Area Sensitivity Runs.
Figure 6.6. Hypolimnion Area Sensitivity Analysis. Predicted Concentrations vs.
Hypolimnion to Epilimnion Surface Area Ratio. Each lake is connected by
dashed lines.
Figure 6.7. Hypolimnion Area Sensitivity Analysis. Response Surface and Predicted
Concentrations for Epilimnetic Methylmercury Concentrations.
Figure 6.8. Hypolimnion Area Sensitivity Analysis. Response Surface and Predicted
Concentrations for Epilimnetic Total Mercury Concentrations.
Figure 6.9. Hypolimnion Area Sensitivity Analysis. Response Surface and Predicted
Concentrations for Hypolimnetic Methylmercury Concentrations.
Figure 6.10. Hypolimnion Area Sensitivity Analysis. Response Surface and Predicted
Concentrations for Hypolimnetic Total Mercury Concentrations.
Figure 6.11. Hypolimnion Area Sensitivity Analysis. Response Surface and Predicted
Concentrations for Sediment Methylmercury Concentrations.
Figure 6.12. Hypolimnion Area Sensitivity Analysis. Response Surface and Predicted
Concentrations for Sediment Total Mercury Concentrations.
Figure 6.13. Hypolimnion Area Sensitivity Analysis. Response Surface and Predicted
Concentrations for Fish Tissue Mercury Concentrations.
Figure 6.14. Hypolimnion Area Sensitivity Analysis. Delineation of Regions of Specific
Percent of Model Sensitivity. Epilimnetic Methylmercury.
Figure 6.15. Hypolimnion Area Sensitivity Analysis. Delineation of Regions of Specific
Percent of Model Sensitivity. Epilimnetic Total Mercury.
Figure 6.16. Hypolimnion Area Sensitivity Analysis. Delineation of Regions of Specific
Percent of Model Sensitivity. Hypolimnetic Methylmercury.
IX
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Figure 6.17. Hypolimnion Area Sensitivity Analysis. Delineation of Regions of Specific
Percent of Model Sensitivity. Hypolimnetic Total Mercury.
Figure 6.18. Hypolimnion Area Sensitivity Analysis. Delineation of Regions of Specific
Percent of Model Sensitivity. Sediment Methylmercury.
Figure 6.19. Hypolimnion Area Sensitivity Analysis. Delineation of Regions of Specific
Percent of Model Sensitivity. Sediment Total Mercury.
Figure 6.20. Hypolimnion Area Sensitivity Analysis. Delineation of Regions of Specific
Percent of Model Sensitivity. Fish Tissue Mercury.
Figure 6.21. Hypolimnion Area Sensitivity Evaluation. Default Model Runs with
Changes in R.
Figure 6.22. Hypolimnion Area Sensitivity Evaluation. Default Model Runs with
Changes in Mean Hypolimnion Depth.
Figure 6.23. Hypolimnion Area Sensitivity Evaluation. Default Model Runs with Wider
Range Variation in R, and Well-Mixed Models with Dimensions Similar to R =
0.95 and R = 0.
Figure 6.24. Hypolimnion Area Sensitivity Evaluation. Output from Simple
Mathematical Formulation of Arbitrary Lake System with Structure of Default
Model.
Figure 6.25. Observed Mercury Concentrations versus the Default Level Classifications
for the VT and NH Lakes Dataset (Epilimnion and Hypolimnion Mercury
Concentrations).
Figure 6.26. Observed Mercury Concentrations versus the Default Level Classifications
for the VT and NH Lakes Dataset (Sediment and Fish Mercury Concentrations).
Figure 6.27. Percent Methylmercury in Total Mercury in the Hypolimnion, Epilimnion
and Sediments.
Figure 6.28. Predicted (y-axis) versus Observed (x-axis) Epilimnion Methylmercury
Concentrations for Different Combinations of RlUp and R2Up. Default/Baseline
case is RlUp =0.1, R2Up = 0.1.
Figure 6.29. Predicted (y-axis) versus Observed (x-axis) Epilimnion Total Mercury
Concentrations for Different Combinations of RlUp and R2Up. Default/Baseline
case is RlUp =0.1, R2Up = 0.1.
Figure 6.30. Predicted (y-axis) versus Observed (x-axis) Hypolimnion Methylmercury
Concentrations for Different Combinations of RlUp and R2Up. Default/Baseline
case is RlUp =0.1, R2Up = 0.1.
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Figure 6.31. Predicted (y-axis) versus Observed (x-axis) Hypolimnion Total Mercury
Concentrations for Different Combinations of RlUp and R2Up. Default/Baseline
case is RlUp =0.1, R2Up = 0.1.
Figure 6.32. Predicted (y-axis) versus Observed (x-axis) Fish Tissue Mercury
Concentrations for Different Combinations of RlUp and R2Up. Default/Baseline
case is RlUp =0.1, R2Up = 0.1.
Figure 6.33. Predicted (y-axis) versus Observed (x-axis) Sediment Total Mercury
Concentrations for Different Combinations of RlUp and R2Up. Default/Baseline
case is RlUp =0.1, R2Up = 0.1.
Figure 6.34. Surface interpolation plots of the sum of squares values for all combinations
of the RlUp and R2Up values explored in Figures 6.24 to 6.29.
Figure 6.35. Surface interpolation plots of the estimated standard deviations for all
combinations of the RlUp and R2Up values explored in Figures 6.24 to 6.29.
XI
-------
Evaluating R-MCMfor 91 VT/NH Lakes
EXECUTIVE SUMMARY
Mercury is recognized as an important environmental and ecological contaminant
because of its neurotoxicity. Because of its inherent environmental risk, it is important to
understand mercury fate and transport mechanisms, as well as to be able to predict its
exposure concentrations within a given environment or ecosystem. This report is in
partial fulfillment of Task # 15529, Mercury Fate and Transport in Watersheds, under
Goal 8, GPRA Objective 8.3, and GPRA Sub-Objective 8.3.1. The goal of this research
is to provide the scientific information and technical data needed to reduce uncertainties
limiting the Agency's ability to assess and manage mercury and methylmercury risks. To
that end, the details outlined in this report constitute the beginnings of an investigation of
the impacts of mercury atmospheric deposition and subsequent lake and watershed fate
and transport processes on the resulting mercury exposures of fish and piscivorous
wildlife of New England lakes. A developmental model was used and evaluated for its
ability to adequately capture and simulate the governing fate and transport processes of
mercury in a watershed and associated water bodies. The ultimate goal in this project is
to improve the assessment tools needed to successfully manage mercury exposure and
risk. The starting point, however, was to evaluate currently available models and to
understand their strengths and weaknesses. Based on this evaluation, ways to improve
the model system can be investigated, and then, if necessary, used to develop a new and
improved model.
In this effort, we first organized and synthesized for modeling purposes the
observed mercury concentration data and lake characteristics collected and presented in
the companion report, "Biogeochemistry of Mercury in Vermont and New Hampshire
xn
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Lakes: An Assessment of Mercury in Waters, Sediments and Biota of Vermont and New
Hampshire Lakes," by the VT DEC, 2003. The model chosen for the study was the
Regional Mercury Cycling Model (R-MCM). The R-MCM was run for a series of
scenarios and parameter refinements. The predicted results were then compared with
observed results.
From our evaluation, we have determined that the R-MCM is not currently at a
level where it can be directly applied to a new region or series of lakes without some
amount of calibration. The R-MCM was found to capture specific trends observed in the
general mercury literature, for example, the trend of decreasing fish tissue mercury
concentrations with increasing pH. Using various visual and statistical analyses, we
found that R-MCM did not provide predictability that was better than simply using the
mean of observed values. Through a rigorous separation technique of evaluating only
specific lake characteristics, we concluded that there were no types of lake that the R-
MCM simulated particularly well. The default level input parameterization was found to
have the largest amount of random scatter for the data points. By specifying precise
values for the default level characteristics, the scatter was greatly reduced. This suggests
that the greatest improvement in the R-MCM's predictive capability might be gained by
gathering precise data on more general lake characteristics such as: pH, size, trophic
status, and stratification.
Further comparison of the R-MCM simulated versus observed mercury
concentrations revealed these important results:
Total mercury concentrations were generally under-predicted,
Percent methylmercury in the epilimnion was generally under-predicted,
Xlll
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Percent methylmercury in the sediment was generally over-predicted, and
Increasing the outflow of methylmercury from the watershed improved the
model's predictive ability.
These results taken in concert, suggest that the major problem with the R-MCM may lie
more in the inaccuracies of modeling mercury loading to the water body and less in the
modeling of the fate and transport of mercury within the water body. Clearly, if loading
is not adequately modeled, then there is no hope for accurately predicting mercury
concentrations within the watershed-lake system. Our results suggest that it is important
to accurately model both the mercury cycle in watersheds and the loading from these
watersheds into the water body of concern.
xiv
-------
Evaluating R-MCMfor 91 VT/NH Lakes
1 INTRODUCTION
Mercury has long been recognized as an important environmental pollutant by the
USEPA because of its neurotoxicity (USEPA, 1997). The primary pathway of human
exposure to mercury is via consumption of fish tissue contaminated with mercury
(USEPA, 1997). Fish and piscivorous birds are also exposed to mercury contamination
and are the primary ecosystem receptors of concern. Because of mercury's properties of
appreciable bioaccumulation and biomagnification, the concentrations of mercury within
wildlife are greater in the higher trophic levels (such as game fish and birds). High
mercury concentrations in fish have been detected in remote lakes far from industrial
sources, suggesting that atmospheric transport and deposition of mercury is a significant
source of mercury to these ecosystems. Mercury enters the global pool of mercury in the
atmosphere from both natural and anthropogenic sources. The mercury can then travel
large distances until it transfers from the atmosphere to terrestrial and aquatic ecosystems
via wet and dry deposition. Once mercury enters a watershed, it can undergo
transformations and reactions controlled by both chemical and biological mechanisms to
form methylmercury. The methylmercury then accumulates within the food web,
increasing in concentration as it works its way up the trophic levels. Therefore, it is not
only important to understand the general fate and transport of total mercury within the
watershed and associated water bodies, but also to understand the transformations of
mercury between oxidation states and molecular structures.
The Regional Mercury Cycling Model (R-MCM) was developed by Tetra Tech,
Inc. in an effort to use process modeling and mass balances to predict mercury
concentrations in the water column, the sediments and fish within a given lake under
1
-------
Evaluating R-MCMfor 91 VT/NH Lakes
various loading scenarios. The R-MCM was originally developed with funding from the
Electric Power Research Institute (EPRI) and the Wisconsin Department of Natural
Resources for application to a set of seven, oligotrophic Wisconsin seepage lakes. The
R-MCM (version I.Ob, Tetra Tech, 1996) used in the analyses presented in this report
came out of that work with model enhancements that include: photochemical reduction of
methylmercury and Hg(II) to elemental mercury; the use of runoff coefficients for Hg(II)
and methylmercury to describe the fraction of mercury deposited on wetlands and
uplands that results in loading to the water body; and an updated approach to mercury
dynamics in the food web. Details on the R-MCM are described more fully in Section
2.2.
The R-MCM was originally developed for and calibrated to a set of seven,
oligotrophic seepage lakes in Wisconsin (mid-west United States). The model was then
applied to a dataset of 21 Wisconsin lakes that spanned a wider range of acidities (pH 5 -
8) and dissolved organic carbon concentrations (DOCs) ranging from 3 to 21 mg/L. The
model was next applied to a clear, acidic lake in Florida and Lake 240, a small
circumneutral pH Canadian Shield lake with a pH of 6.8 and DOC of 5 mg /L. The
Wisconsin lakes did not have significant stream inflows, but it was believed that the
elevated DOC levels may have reflected inputs from nearby wetland areas. Lake 240
was the first application of the R-MCM to a lake with significant surface stream inflows.
Watershed export of Hg(II) was based on runoff coefficients derived from the work of St.
Louis et al. (1994, 1996). Most recently, the model was applied to 24 lakes in
Kejimkujik Park, Nova Scotia, with pHs of 4.3 - 5.9 and DOC of 2.3 - 15.3 mg /L. That
application was the first test of the model calibration of watershed export coefficients.
-------
Evaluating R-MCMfor 91 VT/NH Lakes
The Kejimkujik model application was calibrated most effectively by assuming
that there was no trapping of mercury in the upstream lakes, that is, by assuming that all
mercury deposited on the upstream watersheds is subsequently transported downstream.
Additionally, the R-MCM was found to predict a linear relationship of increasing
concentration of methylmercury in surface waters with increasing DOC. For 21 out of 24
lakes, the predominant source of methylmercury was in-situ methylation.
The purpose of the current report is to present our recent work using and
evaluating the R-MCM to model mercury concentrations in lakes in the Northeastern
United States, specifically lakes in Vermont (VT) and New Hampshire (NH). This is the
first time that the R-MCM has been used in the New England Region of the United States
and will therefore evaluate how well the R-MCM can be transported to a different region
of a country. The New England, Wisconsin, and Canadian lakes are on similar latitudes
and in similarly temperate zones, but they are in different topographies and experience
different climates. The New England application of the R-MCM was performed without
recalibration to evaluate if the model can be directly applied to a new region, as well as to
perform an evaluation of what level of parameter refinement (but not parameter
calibration) would provide the best results. Additionally, this work will provide insight
into the differences among the fate and transport processes governing mercury cycling in
this new lake system, and it will allow for some level of mechanistic description of the
governing mercury processes.
The data set used in this investigation is the largest available to our knowledge to
which the R-MCM has been applied. The dataset covers a wide range of physical and
chemical lake characteristics. The range of characteristics in these lakes include: small
-------
Evaluating R-MCMfor 91 VT/NH Lakes
and medium sized lakes (mean 73 ha, range 8 - 669 ha); stratified and well-mixed lakes;
acidic, circumneutral, and alkaline lakes (mean pH 6.65; range 4.60 - 7.97); trophic
levels consisting of oligotrophic, mesotrophic, eutrophic and dystrophic; and a range of
dissolved organic carbon (DOC) concentrations (mean 4.3 mg/L, range 0.35 - 10.9
mg/L). The experimental data set is described in Section 2.1.
The model was initially evaluated for the simplest, most general and qualitative
approach using the default parameterization written in the R-MCM code. In the next
evaluation, a more quantitative method was employed to refine the model
parameterization by using measured values for parameters describing the actual lakes in
our study. Our goal was to evaluate not only the success of the R-MCM, but to see which
parameters are most important for model improvement. The parameter refinement levels
we developed were designated as "Tiers;" five tiers were used. These tiers and the
default level parameterization are described in Section 2.3.
The analyses developed in this project are presented in different chapters. First,
the R-MCM was evaluated for patterns and trends, as described in Chapter 3. Next, the
R-MCM was evaluated for its ability to predict the observed concentrations of the VT
and NH data set as described in Chapter 4. Next, evaluation of the R-MCM was further
broken down into groups of lakes, based on their lake characteristics, as described in
Chapter 5. The model was then investigated for means on how best to improve model
prediction. To this end, the sensitivity of the model to a few key processes and
parameters was investigated in Chapter 6. Within Chapter 6, the large data set available
was used in conjunction with the R-MCM evaluation to make mechanistic inferences on
-------
Evaluating R-MCMfor 91 VT/NH Lakes
mercury processes and cycling. Conclusions from our study and evalustions are
presented in Chapter 7.
-------
Evaluating R-MCMfor 91 VT/NH Lakes
2 METHODS
2.1 Experimental Dataset
The dataset used in this paper comes from a sampling program that occurred
during 1998 through 2000, when individual Vermont and New Hampshire lakes were
investigated through the Vermont and New Hampshire Environmental Monitoring and
Assessment Program (EMAP). Lakes were selected using an algorithm to ensure random
selection of sampling units, given specific constraints based on lake size and lake to
watershed area ratio (Kamman et al., 2004). The number of lakes used in this study
resulted in 91 lakes. These 91 lakes spanned a complete range of trophic states and
acidities.
During this study, total mercury (HgT) and methylmercury (MeHg)
concentrations were measured in the epilimnion, hypolimnion (if present), and sediment.
Epilimnion samples were taken as subsurface (-0.2 m) grabs. Hypolimnion samples
were collected at one-meter above the sediment-water interface. Sediment samples were
taken from the deepest hole in the lake using a gravity corer.
Fish tissue concentrations were measured using yellow perch (Percaflovescens)
as a standard. Yellow perch was chosen because of its use in previous studies of fish-Hg,
its use in developing fish consumption advisories, its ubiquity and ease of capture, and
because of its occupation of different trophic positions depending on size. Yellow perch
were not present in all of our study lakes, however. A roster of all study lakes known to
contain yellow perch was consulted, resulting in only 47 of our 91 study lakes having
perch (and therefore could have fish tissue observations). Five perch were collected from
-------
Evaluating R-MCMfor 91 VT/NH Lakes
each study lake with perch, and a two-inch section of fillet was taken for sampling.
Length (in mm) and weight (in g) were measured for each fish, and scales and/or otolith1
were retained for fish age determination. Because yellow perch tissue HgT is known to
vary with fish size and age, an ANCOVA was used to assess the influence of length and
age on tissue HgT concentrations, and to estimate length and age-corrected fillet HgT
concentrations. Using the ANCOVA results, the five measured fish concentrations at
each sampling were converted into concentrations normalized to an age of 4.6 yrs (mean
age of perch in all lakes). This conversion allowed for lake cross-comparison and
comparison to model-predicted fish concentrations.
In addition to the mercury concentrations from the sampling effort, the physical
and chemical characteristics of the lakes were used to parameterize the model. More
details on these parameters and the specific values used are presented in the following
section. More details on the sampling techniques are provided elsewhere (Kamman and
Estabrook, 1998a; Kamman and Estabrook, 1998b; Kamman et al., 2004).
2.2 Regional Mercury Cycling Model (R-MCM)
The R-MCM was developed with funding from the Electric Power Research
Institute (EPRI) and the Wisconsin Department of Natural Resources (DNR). R-MCM is
a steady-state, process-driven model that predicts the cycling and fate of total mercury, as
well as MeHg, Hg(0), and Hg(II) for a given lake. The model was developed and
calibrated for 7 oligotrophic seepage lakes in Wisconsin. Features have since been added
1 An otolith is a structure within the inner ear of fishes formed from alternating layers of high and low-
density calcium carbonate. Hard parts of the fish are frequently used as a method for aging fishes
(assigning an appropriate age to a given fish). Fish scales have historically been used (and do not require
sacrificing the fish), but otoliths are internal hard parts that continue to form annuli or increments despite
times of stress or food deprivation (Murphy and Willis, 1996).
-------
Evaluating R-MCMfor 91 VT/NH Lakes
to the model to accommodate a larger range of lake conditions and to predict mercury
export in a watershed. The R-MCM has been used for other regions, such as the lakes in
the Kejimkujik Park of Nova Scotia (Tetra Tech, Inc., 2002).
The R-MCM models and couples the fate and transport of MeHg, Hg(0) and
Hg(II) in the epilimnion, hypolimnion and sediments. Using a steady-state
approximation, the concentrations of mercury species are calculated incorporating:
atmospheric loadings of mercury through dry and wet deposition; loading from watershed
sources; the specific hydrology characteristics of the lake; the chemical and physical
parameters describing the lake; the biomass and particle characteristics of the lake; the
predatory and prey fish species and their characteristics within the lake; and the
equilibrium, partitioning, and thermodynamic constants for the various mercury species
within the lake. Once the mercury concentrations have been calculated within the lake, a
dynamic fish growth module is run to calculate the mercury concentrations in fish tissue
for the fish species of interest.
The model has two levels with which the user can parameterize each lake. On the
first level, a blank database is created for the given suite of lakes. Then, each lake is
entered using its default parameterization. In this step, the type of lake is described in a
qualitative sense. The lake characteristics and the options available are outlined in Table
2.1. In addition to these lake categories and characteristics, the R-MCM also has
categories for fish populations. Specifically, the user can enter the fish to be modeled as
predators and prey. Using the R-MCM Beta Version I.Ob, the available fish species are:
lake trout, northern pike, muskellunge, finescale dace, largemouth bass, smallmouth bass,
bluegill, walleye, and yellow perch. For the VT/NH study, perch were used as both
-------
Evaluating R-MCMfor 91 VT/NH Lakes
predator and prey species. The model can then be run with the default parameterization
as defined using the qualitative characteristics for the lakes and the input fish species
selection.
As a second level of parameterization option, the user can go into the database
using Microsoft Access 2.0 or use the "Edit" function in the R-MCM interface. To
update the database using the Edit function, the user needs to open each lake individually,
open the input edit form, and then edit each parameter value individually. Since the most
user-friendly and efficient parameterization method is to use the simple, default user
interface, clearly, it would be useful to understand how successful the R-MCM is by
using only this level of input, as well as to know how much improvement is gained when
specific parameters are updated/customized.
2.3 Application of the Model
As described previously, the model inputs were entered for the series of lakes
initially using the default, qualitative inputs ("default scenario"). Then, the model
parameterization was refined by using lake-specific and region-specific inputs. A total of
six scenarios were created with different levels of parameter refinement (default scenario
plus Tiers 1 through 5). Each Tier used an increased level of input parameter refinement.
Tier numbers increased with increasing refinement. All refinements from one tier were
carried into the next tier, resulting in an effective "building block" design. The initial
level of user interface was the qualitative input of lake characteristics (see Table 2.1).
The R-MCM program assigns default values to the model parameters depending on the
user-selected lake characteristics in Table 2.1. R-MCM creates a Microsoft Access 2.0
database from the user input and model default parameters for each lake. The
-------
Evaluating R-MCMfor 91 VT/NH Lakes
characteristics and the lake names used for the default scenario in this study are presented
in Table 2.2.
Before any simulations were performed, a quality assurance and quality control
program was created to make sure that all data were entered and read appropriately. This
QA/QC was performed by creating a Microsoft Access 2.0 query, which extracted the
assigned numeric codes for each category for each lake. This file was exported into a
Microsoft Excel 2000 file. A Boolean comparison was made to check for errors. If any
errors were found, the given lake was deleted and re-entered with the correct input data.
The query was then re-run to verify that all parameters were input correctly. After all
input data parameters were found to be correct, the default scenario was run and the
output data for epilimnion MeHg and HgT, hypolimnion MeHg and HgT, fish tissue
mercury concentration, and sediment mercury concentrations were exported into a
Microsoft Excel 2000 file as well as into a Matlab data file.
After the default scenario was run, the default database was updated as described
earlier to reflect the refined input parameters through a series of Tiers. The values of
default parameters that were updated through the Tier refinements are listed in Table 2.3.
The query capability of Microsoft Access 2.0 was used once again to ensure data quality.
First, a database file consisting of the appropriate input parameters was created in the
Project Database Access 2000 file. Because the R-MCM uses Access 2.0, the Access
2000 file could not be directly imported for use in the R-MCM. Therefore, an
intermediate step was necessary. The data file was first imported into Microsoft Excel.
The file was then saved as a Microsoft Excel 5.0 file, and imported into the Microsoft
Access 2.0 R-MCM database. From there, an SQL update query was written and run to
10
-------
Evaluating R-MCMfor 91 VT/NH Lakes
update the R-MCM project database. The query was designed to match the lake numbers
of the database file with the lake numbers of the master file (R-MCM input parameter
file), which correspond to each lake name. The master file was used as the link between
the database, which uses lake numbers, and the REMAP project database, which uses
lake names. Using SQL programs, the R-MCM project database was updated with the
lake-specific or region-specific data. This method of updating was found to be efficient
and successful. Manual checks were performed to verify that the updates were correct.
The types of parameter updates are presented in the following sections as Tiers 1 through
5.
2.3.1 Tierl
The first series of refinements focused on updating those parameters most
associated with the characteristics of the default categories. Those parameters were:
pH,
Lake Surface Area,
Hypolimnion Surface Area,
Hydrologic Residence Time,
Epilimnion DOC, and
Hypolimnion DOC.
There were no data for hypolimnion surface area available, so this was modeled as a
default value of one-third of the lake surface area. (Evaluation of the sensitivity of the
model to hypolimnion surface area is detailed in Section 6.2.) Update values for these
refinements are presented in Table 2.4.
11
-------
Evaluating R-MCMfor 91 VT/NH Lakes
2.3.2 Tier 2
The second series of refinements focused on updating the mercury loading
according to regional specifications. These updates included the precipitation rate, the
concentration of Hg(II) in precipitation, and the reactive gaseous mercury (ROM)
deposition rate. The exact updates and values were:
Hg(II) concentration in precipitation, default 10 ug/m3 updated to regional
estimate of 9 ug/m3,
Precipitation rate changed from default of 0.8 m/yr to regional estimate of
1.0 m/yr, and
ROM updated to 4.5 ug/m3 (estimated as half of Hg(II) wet deposition2).
2.3.3 TierS
The third series of refinements updated the mercury loading according to lake-
specific estimates. These values came from research performed and modeling done by
Dr. Eric Miller (Miller, 2002; Miller, personal communication, 2003). The inputs updated
were:
Hg(II) concentration in precipitation,
Precipitation rate, and
ROM (modeled as 1 % of the total mercury vapor concentration).
2 The estimation of RGM deposition as half the Hg(II) wet deposition is a rough approximation used within
R-MCM. The default parameterization within R-MCM was originally written assigning an RGM of 3.5
ug/m2/yr and wet Hg(II) deposition at 9 ug/m2/yr. The default RGM was, therefore, approximately 40% of
wet deposition. This was rounded to 50%, as a rough estimator, because there is no general approximation
for the VT/NH region without using site-specific data. Lake-specific RGM values, based on Eric Miller's
work, were subsequently found to be nearer to 100% of the wet Hg(II) deposition. These latter values are
incorporated into the Tier 3 refinement.
12
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Values corresponding to the specific lakes of our study are presented in Table 2.5. Since
there were some lakes that did not have update values; these lakes were left with their
default parameter values.
2.3.4 Tier 4
The fourth series of refinements updated the watershed characteristics. These
values came from lake-specific data collected during the VT/NH REMAP study. The
updated inputs were:
Total catchment ratio area to lake area,
Fraction of total catchment covered by wetlands, and
Fraction of total catchment covered by lakes.
Values corresponding to our specific study lakes are presented in Table 2.6. There were
some lakes that did not have values for these parameters, so the parameters for those
lakes were left as the default values.
2.3.5 TierS
The fifth series of refinements focused on updating specific lake characteristics:
Mean thickness of the epilimnion, and
Mean thickness of the hypolimnion (if present).
Values corresponding to our study lakes are presented in Table 2.7.
13
-------
Evaluating R-MCMfor 91 VT/NH Lakes
3 EVALUATION OF R-MCM: DEFAULT MODEL RESULTS AND
TRENDS
R-MCM is a complex, process-based model that has almost 300 different
parameters describing the mechanisms governing mercury fate, transport, and
concentrations in a watershed. The default input allows for choosing different lake
categories from those listed in Table 2.1 and described in Section 2.2. Before addressing
the VT and NH data set, it was deemed useful to run the R-MCM on a default level and
study/evaluate its predicted mercury concentrations. By investigating the predicted
output of the default runs, some insight could be gained regarding the impacts of the
various fate and transport mechanisms simulated in the model. Trends in the different
predicted mercury species concentrations within the various media could also be
examined.
3.1 Default Model Output
An R-MCM input parameter database was established for a series of 72 "default
parameter" lakes. The lakes were chosen to represent an array of all combinations of the
following characteristics: acidity, stratification, size and trophic state. All lakes in this
evaluation were drainage lakes with no summer hypoxia, with perch modeled as both the
predator and prey fish species. All lakes in the default model run were subjected to
identical precipitation rates and atmospheric mercury loadings as defined in the default
program (see Table 2.3 and discussion in Section 2.3, specifically Section 2.3.2). The
default run combination of lakes is presented in Table 3.1. These lakes were simulated
14
-------
Evaluating R-MCMfor 91 VT/NH Lakes
using R-MCM to predict Epilimnion MeHg and HgT, Hypolimnion MeHg and HgT,
Sediment MeHg and HgT, and fish tissue mercury concentrations.
The predicted values for the range of default lakes (simply numbered from 1 to
72) are presented in Table 3.2. Summary information regarding these predictions is
presented in Table 3.3. To put the predicted mercury concentrations into context relative
to observed measurements of mercury concentration, Table 3.4 presents published values
for observed mercury concentrations in different media, including ranges and means.
The larger concentrations of the observed data were associated with water bodies
impacted by direct mercury loading sources, which the default set-up runs were not
parameterized to handle. The data presented in the first row of Table 3.4 (Driscoll et al.,
1994) were for lakes in the Adirondack region of New York and, therefore, are believed
to be more representative of lakes impacted predominately by atmospheric deposition.
The observed results of Driscoll's work generally agree well with the range of predicted
concentrations in the water and in fish tissue. This results suggest that the R-MCM is
equipped to handle the general trends and ranges of mercury concentrations in the
northeast region of the US.
3.2 Trends in the Default Run Data Output
After tabulating the default R-MCM run results, the predicted mercury
concentrations were plotted against the lake characteristics specified in Table 2.1. These
slots are presented in Figures 3.1 and 3.2. These figures provide insight into the range of
values that a given mercury species concentration might have in a specific media
depending on specific lake characteristics. These figures also reveal the output trends
that the R-MCM produces given the inherent fate and transport processes and
15
-------
Evaluating R-MCMfor 91 VT/NH Lakes
mechanisms incorporated in the model. Keeping in mind that these concentration output
trends are not directly associated with observations or any specific water body but rather
that they depict only possible behavior in the hypothetical default lakes. It is important to
mentally separate these model predictions from actual observed results.
From these figures, it is very clear that for some lake characteristics and for
mercury species in some media there is quite a wide range of possible predicted
concentration values. The magnitude of the ranges of these predicted values point
towards the most important parameters that affect the mercury prediction and the value of
these predictions.
3.2.1 Lake Area (Lake Size)
The predictions of mercury species concentrations against lake area are consistent
across the different media. An inverse correlation between mercury species
concentration and lake size was exhibited. As the size of the lake increased, there was a
corresponding decrease in mercury species concentration. Given a uniform mercury
loading from both the atmosphere and the watershed across the different default set-up
water bodies, there appears to be a consistent dilution effect as a lake increases in size.
The scatter within any one given size lake (e.g., all small lakes) is large, however. This is
important because it is not safe to assume that a small lake will have more mercury in its
different media than a large lake. Indeed, these results suggest that the other
characteristics describing the lake can so greatly affect the mercury concentration
prediction that it is quite possible for a large lake to have a greater concentration than a
small lake. However, if all other lake characteristics are similar, then it is likely that the
larger lake will have lower mercury species concentrations than the small lake.
16
-------
Evaluating R-MCMfor 91 VT/NH Lakes
3.2.2 Epilimnion DOC (Trophic Status)
The trophic status of a lake is another characteristic that the user enters. The
model uses trophic status to assign a default dissolved organic carbon (DOC)
concentration. In Figures 3.1 and 3.2 in order of increasing DOC concentration (left to
right), the trophic status for each lake mercury species concentrations range plot
corresponds to oligotrophic, mesotrophic, eutrophic, and dystrophic, respectively. There
was not a consistent pattern of predicted mercury species concentration versus trophic
status. There was a positive correlation between epilimnion methylmercury and
epilimnion DOC. For epilimnion total mercury, there was not much difference between
the oligotrophic and mesotrophic lakes, but there wais an increase for the dystrophic
lakes and a decrease for the eutrophic lakes. There were also apparent negative
correlations between sediment mercury concentrations (both MeHg and HgT) and fish
mercury (both prey and predator) and DOC. For fish and sediment, the predicted
mercury species concentrations associated with eutrophic lakes spanned a tighter range
(less scatter/smaller standard deviation) than the other trophic status lakes, and generally
had lower concentrations. The dystrophic lakes had a larger predicted species
concentration range than the eutrophic lakes, while the oligotrophic and mesotrophic
lakes had similar predicted mercury concentration values and ranges. The hypolimnion
mercury species concentration predictions were generally scattered and did not appear to
demonstrate any appreciable correlation. As with the lake area (Section 3.2.1),
interpretation of the correlation of mercury species concentrations with epilimnion DOC
(trophic status) must be tempered because of the large amount of scatter in the predicted
17
-------
Evaluating R-MCMfor 91 VT/NH Lakes
values. Even when a correlation was demonstrated, the amount of scatter tended to
dominate over the subtleties across the other lake characteristics.
3.2.3 Epilimnion pH (Acidity)
For all of the default run mercury species concentration predictions, there was a
general inverse correlation with increasing epilimnion pH. That is, as the pH increased
(acidity decreased) there was a decrease in mercury species concentration. The
correlation was most pronounced for the sediment and fish mercury concentrations, but
was also clearly evident for the epilimnion mercury species concentrations. The
correlation was much less noticeable for the hypolimnion mercury species
concentrations. There was an appreciable scatter among the data points, which must be
noted. The range of predictions for sediment and fish mercury species concentrations
was quite small, however, for alkaline lakes. Having higher mercury concentrations in
acidic lakes versus lower concentrations in alkaline lakes is a well-established
phenomenon in natural systems. This is not a phenomenon imposed on the model system
directly, but rather simply is a result of the chemistry and governing fate processes that
affect mercury concentrations as a function of pH.
3.2.4 Lake Stratification
There was no appreciable correlation observed between simulated mercury
species concentrations and lake stratification (i.e., well mixed versus stratified). There
was a large amount of scatter across the range of default lakes, so it was difficult to make
any conclusions on the impact of stratification.
18
-------
Evaluating R-MCMfor 91 VT/NH Lakes
4 EVALUATION OF R-MCM: ENTIRE LAKE DATA SET
4.1 General Visual Inspection
The modeling results for the default run plus the five Tiers are presented in
Figures 4.1 through 4.6. In these figures, the results of predicted mercury species
concentrations are plotted versus measured concentrations for: epilimnion MeHg,
epilimnion HgT, hypolimnion MeHg, hypolimnion HgT, fish tissue HgT and sediment
HgT. A scan of the figures reveals that some demonstrate a large amount of scatter as
well as appreciable bias. In Figure 4.1, the initial run with default settings, there seems to
be a general over-prediction (predicted values are greater than observed values) of
epilimnion MeHg concentrations with some exceptions. There is a great amount of
scatter in the epilimnion MeHg and HgT and fish HgT concentrations. The hypolimnion
HgT concentrations were predicted to fall within a much narrower range than the
measured values, indicating poor prediction ability for the observed ranges of
hypolimnion concentrations. The default model runs generally predicted mercury species
concentrations that were generally much less than the observed concentrations (under-
prediction) for the hypolimnetic waters and sediments.
The first tier of refinement involved updating those parameters most directly
associated with those defined by the default lake characteristics (as shown in Table 2.1).
For example, if the lake is designated acidic, the default lake pH is assigned to be 5.3.
Similarly, if the lake is designated circumneutral, the model assigns a default pH of 6.5,
and likewise a pH of 8.0 for designated alkaline lakes. In Tier 1, the observed pHs, as
measured in and reported from the VT/NH REMAP study, were input to replace the
19
-------
Evaluating R-MCMfor 91 VT/NH Lakes
model default values. This was done for all the variables/parameters listed in Table 2.3
for Tier 1, producing the input data set shown in Table 2.4.
The results of the predicted versus observed mercury species concentrations for
Tier 1 are presented in Figure 4.2. The most noticeable differences between the default
run and the Tier 1 results are the decreased predicted concentrations for epilimnion
MeHg and HgT and fish HgT. There was a large amount of scatter in the default run
epilimnion concentrations that was not seen in the Tier 1 epilimnion concentrations. For
some of the data, predominantly for lakes associated with lower observed concentrations,
there appeared to be an improvement in the model. However, there remain large errors
for the middle to high observed concentrations. Additionally, there were several data
points predicted as zero or near-zero concentrations where non-zero concentrations were
observed. In both the Tier 1 epilimnion MeHg and HgT, several of the higher observed
concentrations were appreciably under-predicted. The predicted hypolimnion
concentrations were not as dramatically different for the Tier 1 case compared to the
default case. This suggests that the refinements of Tier 1 do not greatly affect the
model's predictive capacity for hypolimnion concentrations. The fish concentrations
were generally reduced in the Tier 1 compared to the default, except for a couple of
predicted values that actually increased, and have larger residual errors than in the default
case. The sediment concentrations also did not change appreciably from the default to
the Tier 1 scenario.
The results for Tier 2 and Tier 3 are presented in Figures 4.3 and 4.4. These Tiers
were parameter refinements associated with the atmospheric input of mercury to the
system. Tier 3 was a region-specific adjustment and Tier 4 was a lake-specific
20
-------
Evaluating R-MCMfor 91 VT/NH Lakes
adjustment. There were no appreciable differences between the model outputs of Tiers 2
and 3 and those of Tier 1. This suggests that these parameter refinements did not have a
dramatic impact on the modeling results, although there does seem to be some minor
adjustments in some predicted values. This is not to say that these results are not
important, but rather that there are not dramatic differences between the model outputs
for the refined atmospheric loads versus the default values. Therefore, minor adjustments
did not cause noticeable shifts in the results. However, it is possible that if the model
were to be applied to a region with dramatically different mercury depositions, then this
refinement could be critical.
The Tier 4 scenario involved refining the watershed parameters for all the lakes.
There was a wide range of parameter values and changes for the simulated lakes, so it
was expected that the watershed input might have a dramatic impact on the model results.
However, there were no major changes in the model results, but there were some subtle
differences. Specifically, there was a general increase in the epilimnion predicted
concentrations for both MeHg and HgT. In fact, in Mitchell Lake, the predicted
epilimnion HgT jumped from 0.930 ng/L (Tier 3) to 5.93 ng/L (Tier 4) [predicted
epilimnion MeHg was 0.080 ng/L (Tier 3) and 0.420 ng/L)]. The observed values for
Mitchell Lake are 0.264 ng/L (MeHg) and 4.115 ng/L (HgT). This data point is now seen
in Figure 4.5 above the y=x line, a big improvement from its previous Tier 3, Figure 4.4
under-prediction. This movement was caused by the fact that the total catchment are to
lake area ratio for Mitchell Lake is 165.14, an order of magnitude larger than the default
value of 10. This points to the importance of watershed property influences when
modeling mercury in water bodies. Another general facet of the Tier 4 results was that
21
-------
Evaluating R-MCMfor 91 VT/NH Lakes
some of the near-zero predictions were increased. These increases were not enough to
bring the data points all the way up to the y = x line, but represented a significant
improvement from the previous Tiers (i.e., residual errors were reduced). This result also
shows that the model may be sensitive to the structure of the watershed, and that the
watershed characteristics may have an important impact on accurately modeling mercury
fate and characteristics in lakes.
The final Tier, Tier 5, involved refining the epilimnion and hypolimnion depths.
The original R-MCM was designed for the mid-west lakes of Wisconsin. From the
values used as defaults in the original model, it appears that those lakes were relatively
shallow compared to the deeper lakes of the mountainous regions of Vermont and New
Hampshire. It was therefore believed that these two depths could impact the model
results. When our study lakes were updated with their estimated actual epilimnion and
hypolimnion depths, model results did show an additional subtle increase in the
epilimnion mercury species concentrations, but there was not as dramatic an impact on
the other simulated mercury species concentrations as might have been expected.
A visual review of Figures 4.2 through 4.6 does not reveal many differences
across the Tiers, but rather that the most dramatic difference was from the default run to
Tier 1. This suggests that the changes in the parameter refinement from Tier 2 through
Tier 5 generally produced subtle, but not dramatic, changes in predictive ability. Clearly,
the most dramatic change in model output during the parameter refinement process
occurred from the default run to the Tier 1 run. This suggests that user input update of
the lake characteristics parameters is the place to start for R-MCM application. It also
suggests that a higher level of accuracy in these lake characteristics parameters, and in
22
-------
Evaluating R-MCMfor 91 VT/NH Lakes
understanding the mercury fate and transport processes that respond to these variables
may result in a better predictive capability.
4.2 Error Sum of Squares Analysis
From an inspection of the predicted versus observed concentration plots for the
default scenario, there is a wide range of scatter. This scatter seems to decrease (i.e., the
data points are closer together) when the default run outputs are compared to the
predicted versus observed concentration plots for the Tiered scenarios. However, this
apparent decrease in scatter is also associated with a decrease in the value of the
predicted concentrations themselves. That is, the range of prediction is both compressed
and lowered. Specifically, some of the predicted species concentrations were even
lowered below the y = x line. Therefore, it is unclear if the overall predictability of the
model improved much, if at all during parameter refinement. A quantitative method for
assessing improvement in model predictability can be performed by calculating the error
sum of squares (also known as the residual sum of squares) for each species-media
combination simulation. This is done by taking the sum of the squared differences
between the predicted and measured species concentration values, as described in EQN
4-1 (see for example, Box et al., 1978; Neter and Wasserman, 1974),
,.-^. (EQN 4-1)
1=1
where S, is the error sum of squares, Oi is the observed species concentration in lake i
and Pi is the predicted species concentration in lake i in the same medium.
The results of the error sum of squares for all runs and observations are presented
in Figure 4.7 in graph form to more easily visualize the changes. The y-axis is plotted on
23
-------
Evaluating R-MCMfor 91 VT/NH Lakes
a log-scale so that all variables can be plotted on the same figure. This figure shows that
there were really only small changes in predictability in going from the default case
through the five Tiers. There were some improvements for certain species, while for
others the results actually got worse. For example, for epilimnion HgT, the default error
sum of squares value was 291.3, which decreased to 228.4 by Tier 5 (22% decrease).
The fish concentration error sum of squares had a similar 22% decrease. However, the
epilimnion MeHg sum of squares increased 28%, the hypolimnion HgT and MeHg
increased 24% and 21%, respectively, and the sediment concentration sum of squares
increased 18%. From this analysis, it seems that the model, as a whole, did not improve
significantly with the progression of parameter refinements. If the model user were
solely interested in the fish tissue mercury concentration, then the user would probably
have the best success with a Tier 5 setup. However, the likely decreased success for
other species-media combinations should make one hesitant. If nothing else, this analysis
gives one pause to wonder why the model would get better in some cases and worse in
others. It seems that some fate and transport processes or sources may not be completely
taken into account.
4.3 Summary
The analyses presented here show that the greatest changes in the behavior of the
predicted versus observed mercury species concentration-media plots result from using
the measured values of lake characteristics parameters in place of the default parameter
assignment (i.e., the Tier 1 refinement). There were additionally only more subtle
changes achieved in the results within the Tier refinements themselves (i.e., Tier 1
through Tier 5). In general, all the tiers produced an under-prediction of mercury species
24
-------
Evaluating R-MCMfor 91 VT/NH Lakes
concentrations for most of the lakes. This suggested that the mercury loading needed to
be investigated. But Tier 2 and 3 involved refinement of the atmospheric deposition of
mercury, there were not many changes in the model results for Tier 2 and 3. This
suggested that the default deposition values were on par with the region-specific and
lake-specific values of our study area. Refinement in Tier 4 showed that the watershed
structure can dramatically impact the R-MCM modeling capability. This was
dramatically seen for Mitchell Lake, which had a significant improvement in modeling
results from the default run through Tier 3 and Tier 4. Tier 5 did not produce any
dramatic changes. An error sum of squares analysis of all the model run results showed
that there was some improvement in the success of the model across all the tiers for some
mercury species-media combinations but there was also a decline in performance for
others. Overall, this analysis indicated that there was really not much difference,
statistically, in the success of the model across the tiers.
Because there was not an appreciable difference in R-MCM performance across
the Tiers, and it would take a significant amount of time and effort to evaluate each and
every Tier, we decided that our time could be most effectively and efficiently spent by
only investigating and comparing the results from the default run with those of Tier 5.
The results of our investigation are provided in Section 5.O..
25
-------
Evaluating R-MCMfor 91 VT/NH Lakes
5 EVALUATION OF R-MCM: LAKE CHARACTERISTICS
5.1 Visual Analysis
After the visual inspection and the analysis of sum of squares error of our model
outputs, we concluded that the model did not seem to be providing a good representation
of mercury species concentrations in all the lakes of the study. Our next step then was to
separate the lakes into groups and to see if there were any patterns between lake types
and classifications. For Tier 5 and for the default run, modeled versus observed
concentration data for the lakes were plotted using different symbols corresponding to
each lake's characteristics (e.g., for acidities these are acidic, circumneutral, and
alkaline). The results for the default run are presented in Figures 5.1 through 5.4 and for
Tier 5 run are presented in Figures 5.5 through 5.8. The first thing that was noticed in
this approach was that there was not a specific lake type that was clustered in any one
region in any of the plots. This suggested that the errors are not simply related to a
specific lake type, but rather there may be more general, confounding errors or process
omissions that are not associated with one specific lake type or characteristic. There are
some observed patterns, however, that are discussed in the following sections.
5.1.1 Acidity
Lake acidity was the first lake characteristic reviewed. For both the default (Figure
5.1) and Tier 5 (Figure 5.5) results, the alkaline lakes seemed to cluster more closely
around the y=x line for epilimnion MeHg and HgT, hypolimnion MeHg and HgT and
fish concentrations. For the alkaline lakes, the model grossly under-predicted sediment
concentrations. For the acidic and circumneutral lakes, the model predicted a much
26
-------
Evaluating R-MCMfor 91 VT/NH Lakes
greater scatter than for the alkaline lakes. In general, the scatter for the Tier 5 scenario
was much less than the scatter for the default case and the predicted values were of
smaller magnitude, resulting in a downward shift in the predicted results. From this
simple visual analysis, it seemed that, generally speaking, the model did better predicting
concentrations for the alkaline lakes than it did for the acidic or circumneutral lakes. This
is an interesting result because the model was originally designed for acidic lakes, and
not for alkaline lakes. There is a general belief that acidic lakes are more susceptible to
increased mercury concentrations, so it is possible that other processes are confounding
the modeling in these possibly more complicated lakes. Alkaline lakes may just be
generally simpler to model, and therefore, there is a better predictive capability for them.
This suggests that the model may be capturing the general processes occurring in the
lakes, but the processes in lakes with lower pHs, and therefore more complicated mercury
chemistry, cannot be as easily modeled successfully.
5.1.2 Stratification
Looking next at the stratification levels of the lakes, the picture wais less clear.
For the default run (Figure 5.2), the well-mixed lakes appear to be generally over-
estimated for the epilimnion MeHg and HgT, though there is an appreciable amount of
scatter, with a few lakes even having appreciable under-predictions. For the epilimnion
HgT, the well-mixed lakes had larger over-prediction residual errors than for the
epilimnion MeHg, but the epilimnion HgT had more under-predicted lakes than the
epilimnion MeHg. For the epilimnion HgT, there is a row of greatly over-predicted,
well-mixed lakes around 5 ng/L, but below these the lakes are generally closer to the y=x
27
-------
Evaluating R-MCMfor 91 VT/NH Lakes
line. For the epilimnion MeHg, the well-mixed lakes are clustered closer to the y=x line,
with one lake far below the y=x line and far to the right of the major cluster.
The default, stratified lakes generally seemed to have less scatter for both the
epilimnion MeHg and HgT concentrations. For example, the epilimnion MeHg default,
stratified lake concentrations are predominately clustered tightly around the y=x line,
suggesting a better predictability for the default model for stratified lakes compared to
well-mixed lakes. However, there remained one far outlier to the right of the cluster and
far below the y=x line. Indeed, this far-right outlier lake had the largest residual error and
observed epilimnion MeHg concentration. The default, stratified lakes epilimnion HgT
concentrations had a larger scatter than the default, stratified lakes epilimnion MeHg
concentrations, but there is similarity between the epilimnion MeHg and HgT
concentrations for these lakes because the HgT concentrations are also tightly clustered
around the y=x line for many of the lakes. A few stratified lakes form a row of
predictions at approximately 4 ng/L, and there is an outlier lake to the far right well,
below the y=x line. No comparison information could be gleaned from the hypolimnion
concentration predictions, because only stratified lakes have hypolimnion data.
For the default model fish concentration values, almost all of the well-mixed lakes
were over-predicted, with some dramatically large residuals, while the stratified lakes
were generally well-predicted, but with some scatter. For the sediment concentrations,
there was not much of a visual distinction between the stratification types, and most
results were under-predicted. Generally, for the default case, the visual analysis suggests
that the model performed better for the stratified lakes than for the well-mixed lakes.
28
-------
Evaluating R-MCMfor 91 VT/NH Lakes
In the Tier 5 stratification runs for stratification (Figure 5.6), the stratified lakes
were clumped together with appreciable under-prediction of epilimnion MeHg. In
general, the model predicted values near zero, despite the range of measured
concentrations. Even with the parameters refined to the Tier 5 level, it seems that the
model could not adequately account for the total amount of MeHg in the epilimnion.
This is readily seen in the extreme case of the lake where the largest measured
concentration of epilimnion MeHg was predicted as having a near-zero value. The well-
mixed lakes covered a wider predictive range, but possessed a large amount of scatter in
both over-prediction and under-prediction, with a tendency toward in under-prediction.
For the epilimnion HgT, the stratified lakes were found to generally be under-predicted.
Again, the model could not capture the range of measured values with its predictions.
This result was similar for the well-mixed lakes, except that at low measured
concentrations the model did have some over-predictions, but as the measured
concentrations increased the predicted values did not correspondingly increase. For fish
concentrations, the Tier 5 results showed a smaller range of predictive results for the
stratified lakes, which clustered well around the low measured concentrations, but did not
increase with increased measured concentrations. This shows a lack in the necessary
predictive range for these scenarios. For the well-mixed lakes, the fish concentrations
have a wide range of scatter, with larger over-predictions than under-predictions, but both
are present. There was not much of a distinction between the well-mixed and stratified
predictions for the Tier 5 run sediment concentrations, and both had a strong tendency for
under-prediction. The analysis of the Tier 5 results for lake stratification showed a
different result from the default case. Here it seems that the well-mixed lakes have a
29
-------
Evaluating R-MCMfor 91 VT/NH Lakes
better range of prediction and that the stratified lakes have a tendency for under-
prediction. This is opposite to the default case.
5.1.3 Lake Size
The next series of model run separations was based on lake size. For the default
run (Figure 5.3), the model had a narrow predictive range for medium lakes with respect
to epilimnion MeHg and HgT, fish HgT, and sediment HgT concentrations, while the
range for these variables for the small lakes was appreciably greater. There was not
much difference for the small and medium lakes in the default and Tier 5 runs for the
hypolimnion MeHg or HgT. For the default run results, the medium lakes were clustered
relatively close to the y=x line for epilimnion MeHg and HgT and fish HgT
concentrations except for a few outliers, while there was a larger pattern of scatter for
these concentrations in the small lakes. The default sediment mercury concentrations
were generally under-predicted in both the small and medium lakes, with a clustering of
the results for the medium lakes, and a wider range of scatter for the small lakes.
For the Tier 5 case (Figure 5.7), the model did about the same for both small and
medium lakes for epilimnion HgT, hypolimnion MeHg and HgT, fish HgT, and sediment
HgT concentrations. The epilimnion MeHg was better predicted for small lakes, but
there was still the strong tendency towards under-prediction for both lake sizes. The
medium lakes had the greatest bias towards under-prediction. From this analysis, it
seemed that for the default case, the model did better for small lakes, however, when all
the data were incorporated, the model did about equally well for the small or medium
lakes. That is, there was no specific bias towards success or failure with lake size.
30
-------
Evaluating R-MCMfor 91 VT/NH Lakes
5.1.4 Trophic Status
The final model runs separation was based on trophic status. For both the default
(Figure 5.4) and Tier 5 cases (Figure 5.8), this separation did not lead to a clear pattern.
For the default runs, the epilimnion MeHg in the oligotrophic lakes formed a tight cluster
about the y = x line. This suggests that the epilimnion MeHg was modeled relatively well
for the oligotrophic lakes. A similar pattern was seen in the default oligotrophic lakes for
epilimnion HgT, but there was a bit more scatter with some lakes having large errors.
Still there was a relatively good fit for these data. The default oligotrophic hypolimnion
MeHg concentrations were generally over-predicted, however, while the default
oligotrophic hypolimnion HgT concentrations were under-predicted. In fact, the most
extreme outliers for the hypolimnion HgT were oligotrophic lakes. The default run
oligotrophic fish concentrations spanned a wide range from under-prediction to over-
prediction. Specifically, the range is nearly from the most under-predicted to the most
over-predicted fish concentration, suggesting a weak predictive power for oligotrophic
lakes. Default, oligotrophic lake sediment concentrations were scattered about the y = x
line, and were the only type of lake to have over-predictions.
For default mesotrophic lakes, the epilimnion MeHg concentrations formed a
tight cluster around the y = x line, but also had a string of over-predictions. The data
were also scattered about the y = x line for the epilimnion HgT and hypolimnion MeHg.
Hypolimnion HgT and sediment HgT concentrations were under-predicted within a
narrow range of prediction values across the wider range of measured values. The fish
mercury concentrations were mostly clustered about the y = x line but had a few dat
points appreciably over-predicted, a pattern similar to the default epilimnion MeHg.
31
-------
Evaluating R-MCMfor 91 VT/NH Lakes
There are not that many eutrophic lakes in the study, and fewer eutrophic lakes
with fish concentrations, so it was difficult to draw too many conclusions. For the data
that were available, the default epilimnion MeHg predictions for eutrophic lakes were
scattered about the y = x line. The results for the default epilimnion HgT and sediment
mercury concentrations were predominately under-predicted.
For the default dystrophic lakes, there is a cluster about the y = x line for the
epilimnion MeHg, but there are also two under-predicted outliers as well as a string of
data points that are over-predictions. The default dystrophic epilimnion HgT data have
results near the y = x line, with one data point appreciably under-predicted, but the results
are predominately over-predicted. The data points predicted to have the highest
epilimnion HgT, approximately 4 to 5 ng/L, were all dystrophic lakes. The default
dystrophic lakes were scattered for the hypolimnion concentrations, as well as for the fish
and sediment HgT concentrations.
For the Tier 5 scenario, the results were a little different. For the oligotrophic
lakes, there is a downward shift in the predicted epilimnion MeHg concentrations
yielding a preponderance of under-predicted results. For the epilimnion HgT, the
oligotrophic lakes remained scattered about the y = x line, and the farthest outlier for the
oligotrophic lakes was nearer the y = x line than for the default run. The oligotrophic
hypolimnion MeHg concentrations were clustered tightly around the y = x line, while the
hypolimnion HgT concentrations were still under-predicted with some severe results.
The fish mercury concentrations for the oligotrophic lakes were better, with less scatter
but a greater bias towards under-prediction. Oligotrophic sediment HgT concentrations
were scattered with under-predictions similar to the default case.
32
-------
Evaluating R-MCMfor 91 VT/NH Lakes
For Tier 5 mesotrophic lakes, the epilimnion MeHg concentrations no longer had
the string of over-predictions that the default case had. These concentrations decreased;
some were nearer to the y = x line while others were closer to zero, effectively being
farther from the y = x line. For the Tier 5 mesotrophic epilimnion HgT concentrations,
the predicted values also decreased. The scatter remained similar to the default case, with
an increased bias towards under-prediction. Hypolimnion MeHg concentrations were
scattered closely about the y = x line, while the hypolimnion HgT concentrations were all
under-predicted. The mesotrophic fish HgT concentrations covered a greater range of
predicted values than those in the default case. The sediment HgT concentrations were
all under-predicted.
Tier 5 eutrophic data were all under-predicted except for fish HgT concentrations
where there were only three data points, all were over-predictions. The epilimnion and
hypolimnion HgT all were under-predictions with little variability over the wide range of
measured variables. Epilimnion and hypolimnion MeHg concentrations were also were
under-predicted, but are scattered close to the y = x line with only a few outliers.
Tier 5 dystrophic lake data did not have the large, over-predicted values that the
default dystrophic lake data exhibited. All the dystrophic lake epilimnion MeHg
concentrations were predicted to be near zero, even though the measured values had a
wide range. The epilimnion HgT concentrations were generally under-predicted, but the
predicted values spanned a wide range and there was a cluster of data about the y = x
line. Dystrophic lake hypolimnion MeHg and HgT concentrations both displayed a
narrow predicted range with a wide measured range. This suggests little predictive
capability for these lakes for these variables. For the dystrophic lakes, the fish HgT
33
-------
Evaluating R-MCMfor 91 VT/NH Lakes
concentrations were generally under-predicted but with a good amount of scatter. The
sediment HgT concentrations were all under-predicted.
5.1.5 Summary of Visual Analysis
Even after separating the predicted and measured data points by lake
characteristic and visually analyzing the results, it was difficult to clearly discern general
patterns. However, going through each lake variable by each lake characteristic, some
general inferences could be made regarding the model capabilities.
Relative to lake acidity, the model seemed to predict better for alkaline lakes
than it did for acidic or circumneutral lakes. This may be due to the more
complicated mercury chemistry and fate processes present in non-alkaline
lakes.
Relative to lake stratification, opposite results were found for the default runs
versus the Tier 5 runs. For the default scenario, the model was found to be
better at predicting concentrations in stratified lakes, while for the Tier 5
scenario, the model was found to be better at predicting concentrations in well
mixed lakes. This result may point towards how pattern refinement can
increase model prediction for different types of lakes, and that the governing
fate processes and parameters must be well understood to adequately model
these complex systems.
Relative to lake size, there was more scatter in the predictions for the default
scenario, medium lakes compared to the default scenario small lakes, whereas
the Tier 5 results did not display any distinction in model predictive ability as
a function of lake size.
34
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Relative to trophic status, there was a great deal of scatter in the predicted
data, so that neither the default nor the Tier 5 scenarios showed a clear pattern.
In some instances, such as in the epilimnion MeHg and HgT concentrations,
the default runs for oligotrophic lakes indicated a relatively strong predictive
capability, but this performance was not demonstrated for the hypolimnion,
fish tissue, or sediment mercury concentration predictions. Therefore, the
visual analysis did not provide much useful information on the model
capability relative to the trophic status of lakes.
5.2 Statistical Evaluation of Model Successes and Inadequacies
Following the visual analysis, a statistical approach was used to further test the
qualitative observations. A comprehensive residual sum of squares analysis suggested
that there was not an appreciable improvement in the model performance due to the data
input refinement when viewed as a whole. However, it may be important to sort through
when the model does well and when it does not. During the visual analysis, the
characteristics of the lakes were considered. Similarly, for the statistical approach,
analyses based on the lake characteristics were also performed. A series of different
statistical tests was used as described in standard textbooks and in the modeling literature.
First, a Chi-Square goodness of fit analysis was performed (Tables 5.1 -5.6). Next, a t-
test on the residuals was performed (Tables 5.7-5.12). Lastly, a set of model
performance statistics, including maximum error, root mean square error, coefficient of
determination, modeling efficiency and coefficient of residual mass were used (Tables
35
-------
Evaluating R-MCMfor 91 VT/NH Lakes
5.13 - 5.24). The equations used for these statistics are presented in the following
sections and summarized in Table 5.25.
5.2.1 Chi-Square Goodness of Fit
A standard method of estimating how well a model fits observed data is the use a
"Chi-Square Goodness of Fit" test. This test compares the measured values with the
modeled values and estimates if there is an acceptable amount of error according to a
certain preset level of confidence. Specifically, the chi-square statistic is given by
, (EQN5-1)
Where P, and 0, are the predicted and observed values, respectively for each data point i
and the statistic being tested is
P(x2
-------
Evaluating R-MCMfor 91 VT/NH Lakes
at the 90% confidence level. These chi-square analyses were performed and tabulated for
the six scenarios (default and Tiers 1 - 5) in Tables 5.1 - 5.6. All results are presented
in the tables for completeness, but for brevity and clarity only the default and Tier 5
scenarios are specifically reviewed in the following text.
5.2.1.1 Default Scenario
For epilimnion methylmercury, the model seemed to fit the data quite well for all
types of lakes (i.e., the model passed the chi-square goodness-of-fit test). Hypolimnion
methylmercury was fit well for only alkaline and oligotrophic lakes. Epilimnion total
mercury was modeled well only for alkaline and medium lakes, while hypolimnion total
mercury never passed the chi-square test. The fish tissue mercury concentrations were
modeled well for acidic, alkaline, medium, stratified, and dystrophic lakes. The model
also seemed to fit all sediment data well.
The goodness-of-fit test suggests that the model could be used to model
epilimnion methylmercury and sediment mercury concentrations for all types of lake with
a reasonable error. However, because neither small nor medium lakes were modeled well
for hypolimnion mercury concentrations, the model cannot be used to successfully
simulate these species, regardless of acidity or trophic status. Similarly, the epilimnion
total mercury concentration cannot be predicted because the model did not fit either well-
for well mixed or stratified lakes. Fish concentrations could only be predicted well for
acidic or alkaline, medium, stratified, dystrophic lakes. Based on general observation, it
seems that the default case performed best across the various mercury species
concentrations for alkaline, medium, stratified lakes.
37
-------
Evaluating R-MCMfor 91 VT/NH Lakes
5.2.1.2 TierS
The Tier 5 scenario had some similar successes and inadequacies as the default
case. For example, the epilimnion methylmercury predictions again were fit well for all
types of lakes. For epilimnion total mercury, the Tier 5 scenario fit the alkaline lakes
well, similar to the default case, but did not fit medium lakes, unlike the default case.
Additionally, the Tier 5 epilimnion total mercury concentration was simulated well for
small, stratified, oligotrophic and mesotrophic lakes (unlike the default). Hypolimnion
methylmercury was modeled well for alkaline lakes, similar to the default case, but also
for oligotrophic lakes. Hypolimnion total mercury was never modeled well, just like the
default case. For the fish tissue mercury concentrations, the Tier 5 case passed the chi-
square test for all lakes except for the eutrophic and well-mixed lakes, unlike the default
case, which failed for small, oligotrophic and mesotrophic lakes. The Tier 5 scenario
passed the chi-square test for all sediment concentrations, similar to the default case.
In summary, as with the default case, the Tier 5 model was successful in modeling
epilimnion methylmercury and sediment mercury concentrations for all types of lakes.
Also like the default case, the model never passed the goodness-of-fit test for the
hypolimnion total mercury concentrations. Because the Tier 5 model could not
successfully simulate small or medium lakes for hypolimnion methylmercury, the model
cannot be reliably used to predict this species behavior for any type of lake (regardless of
acidity or trophic status). However, there was an improvement in predicting epilimnion
total mercury concentration, with the model passing the chi-square test for alkaline,
small, stratified, oligotrophic and mesotrophic lakes for this variable. From a general
38
-------
Evaluating R-MCMfor 91 VT/NH Lakes
inspection of the model results across all the mercury species, the model did best for
alkaline, small, stratified, oligotrophic lakes.
5.2.1.3 Chi-Square Test Summary
The Chi-Square Goodness of Fit test demonstrated the following:
For the default case, the model showed good predictability for
Epilimnion MeHg (all lakes)
Epilimnion HgT (alkaline and medium lakes)
Hypolimnion MeHg (alkaline and oligotrophic lakes)
Hypolimnion HgT (no lakes - never predicted well)
Fish Tissue Hg (acidic and alkaline, medium, stratified, and dystrophic
lakes)
Sediment HgT (all lakes).
For the Tier 5 case, the model showed good predictability for
Epilimnion MeHg (all lakes)
Epilimnion HgT (alkaline, small, stratified, and oligotrophic and
mesotrophic lakes)
Hypolimnion MeHg (alkaline and oligotrophic lakes)
Hypolimnion HgT (no lakes - never predicted well)
Fish Tissue Hg (all lakes except eutrophic and well-mixed)
Sediment HgT (all lakes).
The results of the chi-square test showed that Epilimnion MeHg and sediment HgT could
be modeled for any type of lake in both scenarios, while Hypolimnion HgT could never
39
-------
Evaluating R-MCMfor 91 VT/NH Lakes
be modeled. Excluding Hypolimnion HgT, alkaline lakes were the only type of lake for
which each lake mercury species concentration predictions passed the chi-square test.
5.2.2 T-Test on the Mean of the Residuals
The paired t-test is a standard method for evaluating the deviation of predicted
values from observed values. If the model fits the observed data perfectly, then the
predicted value would exactly equal the observed value for all observations. Similarly,
the model is acceptable if the deviation of the error residual (difference between the
predicted and observed) is within an accepted region of confidence. A statistical method
for evaluating this is by calculating the residual error and estimating if the mean error is
not significantly different from zero. This is accomplished using the t-test, with the t-
statistics as shown in EQN 5-3:
(EQN 5-3)
where n is the total number of observations,
d =
1=1
(EQN 5-4)
where P, is the predicted value and Oi is the corresponding observed value,
1 , (EQN 5-5)
n-1 n-1
where
di=Pi-Oi (EQN 5-6)
and the statistic being tested is
40
-------
Evaluating R-MCMfor 91 VT/NH Lakes
P(t
-------
Evaluating R-MCMfor 91 VT/NH Lakes
circumneutral lakes suggested that the bias was not significant, while the bias for alkaline
lakes was significant. There was also a positive bias for small and well mixed lakes
(over-prediction) and a negative bias for medium, stratified and oligotrophic lakes
(under-prediction).
5.2.2.7.2 Epilimnion Total Mercury
For the epilimnion total mercury concentrations, the mean residual error was
found to pass the t-test with 90% confidence for alkaline, medium, stratified, and
oligotrophic lakes. There was a positive bias for acidic and circumneutral lakes, with a
stronger bias for acidic lakes than circumneutral. Small, well-mixed, mesotrophic lakes
also exhibited a positive bias.
5.2.2.1.3 Hypolimnion Methylmercury
For hypolimnion methylmercury concentrations, the model passed the t-test at
90% confidence for all lakes except well-mixed and mesotrophic lakes. For these lakes,
there was a positive prediction bias.
5.2.2.1.4 Hypolimnion Total Mercury
For hypolimnion total mercury concentrations, the t-test was passed only for
eutrophic and dystrophic lakes. However, the standard deviations for these types of lakes
were 10.38 and 24.50, respectively. Therefore, there was a high level of scatter, which
resulted in the passing of the t-test (there was no statistical reason to believe that the
mean was different from zero). All lakes had a negative mean residual, demonstrating a
negative bias (under-prediction) for all hypolimnion total mercury concentration
predictions. Additionally, the standard deviations and means were large. This test
42
-------
Evaluating R-MCMfor 91 VT/NH Lakes
demonstrated the great difficulty of the model in predicting the hypolimnion total
mercury concentrations, and especially demonstrated the negative bias of these
predictions.
5.2.2.7.5 Fish Tissue Mercury Concentration
The t-test at 90% confidence for the fish tissue mercury concentrations revealed
that acidic, alkaline, medium, stratified, eutrophic and dystrophic lakes all had mean error
residuals that were not statistically different than zero. There was positive prediction bias
for the circumneutral, small, well-mixed, oligotrophic and mesotrophic lakes. The
oligotrophic and mesotrophic lakes generally had similar levels of positive bias. No lakes
were found to have a significant negative bias.
5.2.2.1.6 Sediment Total Mercury Concentration
For the sediment mercury concentrations, only oligotrophic lakes had a mean
error residual that was not statistically different from zero. All other lakes had a negative
predictive bias (under-prediction).
5.2.2.2 Tier 5 Scenario
5.2.2.2.7 Epilimnion Methylmercury.
For the epilimnion methylmercury concentration predictions using the default
scenario, the t-test at 90% confidence showed that only the dystrophic lakes had a mean
error residual that was not significantly different than zero. However, the dystrophic
lakes had a large standard deviation compared to the estimated mean. This suggests that
there was a significant scatter for this lake type, but that the scatter was relatively equally
43
-------
Evaluating R-MCMfor 91 VT/NH Lakes
distributed on either side of the mean (positive and negative). All other types of lakes
had a negative prediction bias, suggesting that the Tier 5 scenario significantly under-
predicted them..
5.2.2.2.2 Epilimnion Total Mercury
Only the oligotrophic lakes passed the t-test at 90% confidence. All other lake
types had a negative bias. The model was found to under-predict all these other lake
types.
5.2.2.2.5 Hypolimnion Methylmercury
For the hypolimnion methylmercury concentration predictions, the t-test revealed
that the mean error residuals were not statistically different from zero for acidic, alkaline,
mesotrophic, eutrophic, and dystrophic lakes. For the other types of lakes, the model had
a negative bias resulting in under-prediction.
5.2.2.2.4 Hypolimnion Total Mercury
For hypolimnion total mercury concentration predictions, the mean residuals of
the error were found to not be significantly different from zero for eutrophic and
dystrophic lakes. However, for these two lake types, the standard deviations of the mean
were quite large, suggesting that passing the t-test was due to the large amount of scatter,
and not precision. The other lake types all had large negative means, demonstrating a
significant negative bias and under-prediction for hypolimnion total mercury
concentration.
5.2.2.2.5 Fish Tissue Mercury Concentration
44
-------
Evaluating R-MCMfor 91 VT/NH Lakes
For the fish tissue mercury concentrations, the mean residuals of the error for
circumneutral, small, well-mixed, oligotrophic, mesotrophic, and eutrophic lakes were
found to not be statistically significantly different from zero. The remaining lake
characteristics produced significant, negative means suggesting the model's under-
predictive bias for these conditions.
5.2.2.2.6 Sediment Total Mercury Concentration
For sediment mercury concentrations, the mean error residuals for oligotrophic
lakes were found to not be statistically significantly different from zero. All other lake
types had statistically significant negative means, suggesting the model's bias towards
under-prediction for these lake types and characteristics.
5.2.2.3 t-test Summary
The t-test analyses combined with the mean residual errors produced the
following general information:
For the default case,
Epilimnion MeHg: Predictions for alkaline, medium, stratified and
oligotrophic lakes had a significant negative prediction bias, while the
small and well-mixed lakes had a significant positive bias.
Epilimnion HgT: Acidic and circumneutral, small, well-mixed and
mesotrophic lakes had a positive prediction bias.
Hypolimnion MeHg: Well-mixed and mesotrophic lakes had a positive
prediction bias.
45
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Hypolimnion HgT: means and standard deviations were large and all
means were negative, demonstrating the negative prediction bias for all
predictions.
Fish Tissue Hg: Circumneutral, small, well-mixed, oligotrophic and
mesotrophic lakes had positive prediction bias.
Sediment HgT: All lakes and conditions except the oligotrophic lakes had
a negative prediction bias.
For the Tier 5 scenario, the model showed good predictability for
Epilimnion MeHg: Even though all lake types had a negative prediction
bias. Dystrophic lakes did not have a residual mean error statistically
different from zero, but still exhibited negative prediction bias for this
species.
Epilimnion HgT: All lake types exhibited a negative prediction bias.
Oligotrophic lakes did not have a residual mean error statistically different
from zero, but still showed a negative prediction bias.
Hypolimnion MeHg: Acidic and alkaline, mesotrophic, eutrophic, and
dystrophic lakes all had mean residual errors that were not statistically
different from zero. All other lake types and characteristics exhibited a
negative prediction bias. Mesotrophic lakes displayed mean residual
errors that were not statistically different from zero, and were the only
lakes that had even a small positive mean residual error.
46
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Hypolimnion HgT: mean residual errors and standard deviations were
large and all residual means were negative demonstrating the negative
prediction bias for all prediction for this species.
Fish Tissue Hg: Circumneutral, small, well-mixed, oligotrophic,
mesotrophic, and eutrophic lakes had mean residual errors that were not
statistically significantly different from zero. All other lake types and
characteristics had negative prediction bias.
Sediment HgT: All lake types and characteristics except oligotrophic had
residual mean errors statistically significant from zero, and had negative
predictive bias.
5.3 Model Performance Statistics
In addition to the standard t-test and chi-square goodness-of-fit test, there has
been other research in the literature addressing the issue of model evaluation. In some of
these works, additional metrics have been introduced. Specifically, a series of statistics
have been suggested by League and Green (1991) that include maximum error (ME), root
mean square error (RMSE), coefficient of determination (CD), modeling efficiency (EF),
and coefficient of residual mass (CRM). As League and Green point out, ME, RMSE,
and CD range from zero to infinity, EF is less than or equal to 1, and CRM can be any
value. The optimal values for model fit are ME = 0.0, RMSE = 0.0, CD = 1.0, EF = 1.0,
and CRM = 0.0.
47
-------
Evaluating R-MCMfor 91 VT/NH Lakes
5.3.1 Maximum Error
The maximum error (ME) statistic given by EQN 5-8
ME = Max\Pt - 0, "_t (EQN 5-8)
indicates the largest deviation from model fit for a series of data points. The ME is
obviously correlated with the magnitude of the observed data, which must be taken into
account. The ME for the hypolimnion total mercury concentration predictions is quite
large, and was the largest ME from all the measurements. The epilimnion total mercury
ME was next in magnitude, approximately an order of magnitude smaller than that for the
hypolimnion total mercury concentration. The epilimnetic methylmercury, hypolimnetic
methylmercury, fish tissue mercury and sediment mercury concentrations all had similar
maximum errors, about an order of magnitude less than the epilimnetic total mercury
concentration maximum error. The magnitude of these errors is more indicative of the
magnitude of the concentrations; however it also indicates that the model has increasing
error as the measurement value increases.
The maximum error can be used, however, across the same variable to compare
errors. For example, for the epilimnion methylmercury concentrations, the circumneutral
lakes had the largest error, followed by acidic lakes for both the default and Tier 5
scenarios. Circumneutral lakes had the largest maximum error for all variables in the
default case and for all variables except hypolimnion concentrations for the Tier 5 runs.
Alkaline lakes had the smallest maximum errors of all the lakes in the default and Tier 5
runs except for the default fish mercury concentrations.
The small lakes had greater maximum errors for all variables except for
epilimnion methylmercury concentration, where the medium lakes had larger maximum
48
-------
Evaluating R-MCM for 91 VT/NH Lakes
errors. The maximum errors were similar for the different lake stratification types and
neither medium nor stratified lakes for either the default or Tier 5 scenario came out as
consistently having the larger maximum error. The largest maximum error was not
consistent across lake trophic status and measured variable, either.
Lastly, the greatest maximum error was calculated for the hypolimnion total
mercury concentration of a circumneutral, small, stratified, oligotrophic lake; the value
was 28.31.
5.3.2 Root Mean Square Error
The root mean square error (RMSE) is the sum of the squares of the residuals
normalized to the mean observed value and the number of observations expressed as a
percent (it is multiplied by 100) as defined in EQN 5-9,
100
RMSE = \ -^ ^ (EQN 5-9)
O
where P, is the predicted value, 0, is the corresponding observed value, O is the mean of
the observed values, and n is the number of observations. Large values for the RMSE are
seen across all the default and Tier 5 model runs. The values are all around 100%, with
the largest values for hypolimnion total mercury concentrations.
5.3.3 Coefficient of Determination
The coefficient of determination (CD) is a measure of how the variance of the
observed data compares to the variance in the predicted data. CD is defined in EQN 5-
10,
49
-------
Evaluating R-MCMfor 91 VT/NH Lakes
(EQN5-10)
v '
1=1
where P, is the predicted value, 0; is the corresponding observed value, and O is the
mean of the observed values. In effect, if the model is perfect, then CD = 1.0. However,
the CD is based on the variances of the predicted and observed values about the mean
observed value. So, if the predicted values have a similar variability about the observed
mean as the observed values do, then the CD will be near unity even if the two variances
are large. If both the predicted and observed values have small variability, but different
means (regardless of whether the predictions are biased high or low), then the
denominator will be large, and CD will approach zero. The closer to zero CD gets, the
bigger the bias. If the observations have a larger variability than the corresponding
predicted values, then the CD will increase, indicating that the predictive capability of the
model is decreasing. In this latter case, the model essentially cannot capture the
fluctuations in the observations.
The CDs for both the default and the Tier 5 scenarios generally had similar
ranges. The larger values of CD were for the hypolimnion methylmercury
concentrations, with values of 14.17 and 5.03 for the default acidic and small lakes,
respectively, and 4.05 and 3.86 for the Tier 5 acidic and small lakes, respectively. The
reduction in these generally large CDs suggests that the model improved from the default
to the Tier 5 scenario for predicting hypolimnion methylmercury concentrations for small
and acidic lakes. The CDs for both the default and Tier 5 hypolimnion methylmercury
concentrations are still high, however, demonstrating that the model cannot adequately
simulate the range of observed hypolimnion methylmercury concentrations.
50
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Medium lakes also had relatively high CD values for epilimnion methylmercury
and epilimnion total mercury concentrations (4.73 and 2.65, default; 2.09 and 1.38, Tier
5, respectively). For the Tier 5 runs, small lakes had a CD of 1.94 for epilimnion total
mercury concentration. Stratified lakes had CDs of 2.50 and 1.31 for default and Tier 5
scenarios, respectively for epilimnion total mercury concentration. Tier 5 well-mixed
and stratified lakes had CDs of 2.09 and 1.48, respectively, for epilimnion total mercury
concentration. These CD values greater than unity also demonstrate how the model is
unable to capture the range for these observed concentrations.
The CDs for fish mercury concentration were small (0.13) for the Tier 5 acidic
lakes, suggesting a predictive bias. Similar low CD values were seen for circumneutral
lakes for fish mercury concentration (0.12, default; 016, Tier 5), for alkaline lakes for
sediment mercury concentration (0.14, default; 0.15, Tier 5), for small lakes for fish
mercury concentration for the default scenario (0.10), for medium lakes for sediment
mercury concentration for the Tier 5 scenario (0.13), and for well-mixed lakes for fish
mercury concentration (0.06, default; 0.09, Tier 5).
For fish mercury concentrations as a function of trophic status characteristic, low
CDs were also found. For the default scenario, the CD ranged from 0.00 to 0.36 and
from 0.22 to 0.89 for the Tier 5 scenario. The fish mercury concentration CD for
oligotrophic Tier 5 lakes had a value of 0.89, which is near unity. The lowest fish
mercury concentration CDs were for mesotrophic and eutrophic lakes in each scenario
(0.08 and 0.00, default; 0.09 and 0.03, Tier 5, respectively).
51
-------
Evaluating R-MCMfor 91 VT/NH Lakes
5.3.4 Modeling Efficiency
Modeling efficiency (EF) is another statistic to evaluate how well a model relates
predictions to observed data as defined in EQN 5-11,
(EQN 5-11)
where P, is the predicted value, Oi is the corresponding observed value, and O is the
mean of the observed values. EF can be negative, but its maximum value is one. As
Mayer and Butler discuss, EF is essentially an overall indicator for goodness of fit. If EF
is negative, then the model cannot be recommended, with preferable values close to one
indicating a "near-perfect" model (Mayer and Butler, 1993). If EF is negative, then the
model predicted values are worse than using the observed mean as a predicted value.
Our use of modeling efficiency as a gauge of model success revealed an
appreciable amount of negative numbers (indicating model inefficiency). Our most
noticeable result was that for sediment mercury concentration, all the EFs were negative,
for all Tiers and all lake characteristics. Not only were all these EFs negative, but many
were large negative numbers. The EFs closest to one for sediment mercury
concentrations were -1.79 for the Tier 5 small lakes and -1.12 for the default acidic lakes.
The EFs farthest from unity for sediment mercury concentrations were -6.41 for Tier 5
medium lakes and -5.95 for default alkaline lakes.
The EFs for fish mercury concentration were also nearly all negative, except for
the default medium lakes that were very near zero at 0.01 and for the Tier 5 stratified
lakes at 0.08. The EFs for fish mercury concentration had some appreciable negative
52
-------
Evaluating R-MCMfor 91 VT/NH Lakes
values. Specifically, for the default lakes, EFs were calculated to be -9.63, -12.07, -
362.50, and -1.79 for oligotrophic, mesotrophic, eutrophic and dystrophic lakes,
respectively. Similarly, the corresponding Tier 5 fish mercury concentration values were
-0.12, -12.31, -30.21, and -3.45. The best EF results were for oligotrophic lakes in the
Tier 5 scenario, and dystrophic lakes in the default scenario.
Hypolimnion methylmercury concentrations had better EF values. For acidic
lakes, for example, EF was 0.93 for the default scenario and 0.75 for the Tier 5 scenario.
However, the hypolimnion total mercury concentration EFs were all negative, except for
one positive value of 0.32 for the default small lakes.
5.3.5 Coefficient of Residual Mass
The coefficient of residual mass (CRM) is defined by EQN 5-12,
n n
CRM = -^ ^ (EQN 5-12)
where P, is the predicted value and Oi is the corresponding observed value. Similar to the
modeling efficiency, the CRM can take negative values. The maximum value of CRM is
one. Ideally, CRM would be equal to zero. If the observed values are greater than the
predicted values, then CRM will approach unity; if the predicted values are greater than
the observed, then the CRM will become negative.
The instance where CRM was nearest to zero in this study was for the epilimnion
total mercury concentration in the Tier 5 oligotrophic lakes, with a CRM equal to 0.18.
For the default case, there were two CRMs near zero, -0.01 for the fish mercyrt
53
-------
Evaluating R-MCMfor 91 VT/NH Lakes
concentration in stratified lakes, and 0.01 for the epilimnion methylmercury
concentration in eutrophic lakes.
The default scenario produced an appreciable number of negative CRMs. This
suggests that the model may be generally over-predicting, because the sum of the
predicted concentrations is greater than the sum of the observed values. For the default
scenario, eutrophic lake fish tissue mercury concentrations, there was a particularly large,
negative CRM of -3.45. Conversely, for the Tier 5 scenario, there were only a few
negative CRMs (fish tissue mercury concentrations in circumneutral lakes, CRM = -0.16;
in small lakes, -0.38; in well-mixed lakes, -0.76; in mesotrophic lakes, -0.46; in eutrophic
lakes, -0.95; and for the hypolimnion methylmercury concentration in mesotrophic lakes,
-0.23). Generally speaking, the CRMs for the Tier 5 runs were closer to unity, suggesting
that the model under-predicts for the Tier 5 scenario.
5.3.6 Model Performance Statistics Summary
The model performance statistics presented in the previous text sections provided
a deeper insight into the model's predictive capabilities. A summary of key points
derived from each of the statistics follows:
Maximum Error
Circumneutral lakes had the largest maximum error of the lake acidity
characteristics for epilimnion, fish and sediment total mercury
concentrations.
Alkaline lakes had the smallest maximum errors for all lakes except for
the default fish tissue mercury concentration, that had a CRM near the
acidic lakes maximum error.
54
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Small lakes had greater maximum errors for all concentrations except
epilimnion MeHg concentrations.
All lake stratification and trophic status characteristics produces similar
maximum errors for all lake mercury species concentrations.
The greatest maximum error was for Round Pond; a circumneutral, small,
stratified, oligotrophic lake, with a maximum error value of 28.31.
Root Mean Square Error normalized to mean observed value
All values were near 100%, suggesting that the predicted error is roughly
the size of the predicted value.
Coefficient of Determination
There was a wide range of coefficient of determinations: near zero, near
unity, and greater than unity.
It was difficult to discern any clear pattern because for any given lake
characteristic, the CDs were positive, negative, or near unity.
Modeling Efficiency
Many negative modeling efficiency values calculated, suggesting that the
model is inefficient, and that using the mean value of the observations
would provide greater predictability than using the model.
Especially large negative model efficiency values were found for Tier 5
medium lake sediment HgT concentration and default alkaline lake
sediment HgT concentrations.
Hypolimnion MeHg concentrations had some model efficiency values
near unity, e.g., default and Tier 5 acidic lakes.
55
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Coefficient of Residual Mass
The closest CRMs to zero for the Tier 5 was for epilimnion HgT
concentrations in oligotrophic lakes, CRM = 0.18.
The closest CRMs to zero for the default case were -0.01 for fish mercury
concentrations in stratified lakes and 0.01 for epilimnion MeHg
concentrations in eutrophic lakes.
Generally, a large number of negative coefficients of residual mass were
observed.
5.4 Summary
Using the different statistical techniques presented in this section permitted a
more thorough analysis of the behavior of the model and its predictive capabilities. Only
alkaline lakes generally passed the chi-square goodness of fit test. This suggests that the
complex behavior of mercury in acidic, and even circumneutral, lakes may be
confounding predictive capacity. However, when the t-test was performed, it was found
that alkaline lakes have a tendency to result in a negative bias. That is, the mean residual
error was indeed statistically different from zero, and that there was a general tendency
for the model to actually under-predict in alkaline lakes. There was a lack of any clear
patterns within the various performance statistics. Therefore, it is difficult to make any
strong conclusions regarding model performance for any particular type of lake.
Alkaline lakes were also found to have smaller maximum errors than the other
lake acidity types, but as the chi-square test illustrated, alkaline lakes tended to be under-
predicted. Modeling efficiency and the coefficient of residual mass analyses suggested
that the variability in the model predictions is so great that the user would be more
56
-------
Evaluating R-MCMfor 91 VT/NH Lakes
successful if the mean of the observed values were used for the specific lake in question.
Therefore, for this set of northeast US lakes, the model did not have enough predictive
power to add useful information to the modeler. Clearly, this is not a very satisfactory
result. For the Tier 5 scenario, there was a pattern of negative residuals for the lake
concentrations, again suggesting the tendency for model under-prediction.
Upon finding no clear pattern from this statistical parameter analysis based on a
separation of lake characteristics, a different approach was taken to find ways of
improving the model's predictive power. One result that was quite clear from the
statistical analysis was the overwhelming amount of negative bias (under-predictability)
demonstrated in the model. This implies that the total amount of mercury in the system is
not being adequately taken into account, and this leads us to the topic of discussion in the
next chapter.
57
-------
Evaluating R-MCMfor 91 VT/NH Lakes
6 MODEL SENSITIVITY AND SYSTEM EVALUATION
The next step in our evaluation delved deeper into the model itself and into an
evaluation of the observed data once it became clear that the model had difficulty
simulating this set of lakes as a whole. Particularly, it became clear that one overlying
trend was the model's inability to predict the total mercury in the system. This was most
evident in the hypolimnion, but also obvious in the epilimnion and sediment. The R-
MCM had a strong tendency to under-predict these concentrations. Two different
processes could produce this modeling result. First, the estimated loss rates could be
dominating the system, resulting in lower concentrations. Second, the mercury input
terms could be incomplete; there could even be a source not currently modeled. To better
understand the reasons underlying the observed bias, we investigated the model
sensitivity to certain key parameters, i.e., their impact on processes that affect the total
mass of mercury in the system, and the relationship of certain parameters to the observed
concentrations.
6.1 Evaluation of Loss Rates: Effect of Photoreduction and
Particle Settling
In this section of the report, we investigate the possibility of bias in the simulated
internal mercury loss rates. To this end, a quick and simple sensitivity study was
performed to evaluate the impact of photoreduction and particle settling velocity on the
predicted total mercury concentrations in the epilimnion, hypolimnion and sediment.
Essentially, the rates governing these two mercury loss processes were decreased to see if
there would be significant increases in the predicted total mercury concentrations.
58
-------
Evaluating R-MCMfor 91 VT/NH Lakes
6.1.1 Settling Velocity
The model uses a default particle settling velocity of 0.5 m/s. Two additional
settling velocities, 1.0 and 10 m/s, were modeled to examine what changes would occur
in the predicted mercury species concentrations. The effects of these settling velocities
on epilimnion, hypolimnion and sediment total mercury concentrations are shown in
Figure 6.1. These higher settling velocities had little impact on the epilimnion total
mercury concentration. For the hypolimnion, there was a general increase in most of the
mercury species concentrations, but not in a way that would produce any overall relevant
improvement in the model performance. There was little change in the sediment mercury
concentrations.
6.1.2 Photodegradation
For photodegradation, the model default photodegradation rate results were
compared to those with no photodegradation at all. Photoreduction is a mechanism that
can result in loss of mercury from the water column via dissolved species reduction to
elemental mercury that can then enter the air via evasion. The effects of
photodegradation rate on simulated epilimnion, hypolimnion and sediment total mercury
concentrations are shown in Figure 6.2. Removing photodegradation completely did not
result in any appreciable predicted concentration changes from the model default case
values.
6.1.3 Settling Velocity and Photodegradation
For completeness, a further analysis was performed to evaluate if there is any
difference in model output when both photodegradation and settling velocity are changed.
59
-------
Evaluating R-MCMfor 91 VT/NH Lakes
The combined effects of changing both settling velocity and photodegradation rate on
epilimnion, hypolimnion and sediment total mercury concentrations are shown in Figure
6.3. Again, no appreciable difference in model performance was found through this
analysis.
6.1.4 Summary of Evaluation of Loss Rates
From the results of our loss rate analysis, it was clear that these two factors are
not those mostly causing under-prediction of the concentrations of mercury in the lakes.
Therefore, additional investigation was necessary to elucidate those mechanisms that
could alleviate these under-predictions.
6.2 Sensitivity Evaluation of Hypolimnion Surface Area
One of the Tier 1 parameter changes was the hypolimnion area, which is generally
defined as a fraction of the lake surface (epilimnion) area (see Tier 1 parameters in Table
2.3). The model parameters of pH, lake size, residence time and water column DOC all
came from lake-specific data, gathered in the VT/NH REMAP study (see Section 2.2).
However, hypolimnion surface area was not immediately available for the lakes in this
study. An assumption was therefore made to estimate it. For the default set-up, the R-
MCM estimated the hypolimnion surface area as a given fraction of the epilimnion
surface area. The epilimnion surface area (assumed identical to the parameter "lake
size") was a measured parameter available in the VT/NH REMAP study data set. The
hypolimnion surface area was estimated as one-third of the lake surface (epilimnion
surface) area. The purpose here is to present a simple sensitivity analysis on how the
60
-------
Evaluating R-MCMfor 91 VT/NH Lakes
model reacts to different approximations for the hypolimnion surface area in terms of the
ratio between the hypolimnion surface area and the epilimnion surface area, R:
_ Hypolimnion Surface Area
Epilimnion Surface Area
The sensitivity analysis was performed by repeatedly running the R-MCM using
all the Tier 1 scenario parameter values except with different hypolimnion surface areas
based on different choices of R. Because the hypolimnion surface area is only relevant
for lakes with stratification, all well-mixed lakes were removed from the analysis. The
ratios (R values) chosen were: 1/2, 1/3, 1/4, and 1/5, where 1/3 was the default base case.
The R-MCM was used to predict seven mercury species concentrations: epilimnetic
methylmercury (EPI_MeHg), epilimnetic total mercury (EPI_HgT), hypolimnion
methylmercury (HYP_MeHg), hypolimnetic total mercury (HYP_HgT), sediment
methylmercury (Sed_MeHg), sediment total mercury (Sed_HgT), and fish tissue mercury
(Fish).
6.2.1 Visual Analysis and Maximum and Absolute Changes
The predicted concentrations were plotted versus the observed results for each of
these sensitivity runs in Figures 6.4 and 6.5. The actual concentrations predicted are also
presented in the appendix in Tables A-8 through A-14. From visual inspection of the
figures, it is apparent that there can be quite a change in the predicted concentrations of
mercury species in some lakes due to simply changing the hypolimnion surface area.
There was a noted general tendency for the predicted mercury concentrations to increase
as the hypolimnion surface area decreased (as R decreased). It also seemed that the
methylmercury concentration had relatively larger changes than the associated changes in
61
-------
Evaluating R-MCMfor 91 VT/NH Lakes
total mercury concentration. Finally, the sediment mercury concentrations did not seem
to have a large change, whereas the associated fish tissue mercury concentrations seemed
to be more affected by the changing hypolimnion surface area.
To gain a better understanding of the overall trends in the predicted
concentrations, the average and maximum change for each concentration for each lake
was calculated as a function of the area ratio from 1/5 to 1/2. Each was calculated both
as a percent change and as an absolute change. The absolute change was simply the
difference between the predicted mercury species concentrations at the extremes of the
sensitivity analysis, i.e. for R = 1/2 and R = 1/5.
Absolute Change = C1/5 - C1/2 (EQN 6-2)
Similarly, the percent change is the absolute change divided by the predicted variable
concentration for R = 1/2. This portion of the equation is an effective relative change.
This term is then multiplied by 100 to create a percentage result.
C -C
Percent Change = 100 -^ v (EQN 6-3)
^-1/2
The results from these calculations are presented in Table 6.1.
The data in Table 6.1 show that the hypolimnetic total mercury had the largest
average and maximum absolute change. This is most likely a reflection of the higher
predicted values for the hypolimnetic total mercury concentrations. This is confirmed by
reviewing the percent changes for total mercury concentration. Specifically, the percent
change for hypolimnetic total mercury concentration (58%) was not the greatest percent
change of all the predicted concentrations. For the percent change, the largest value is
seen for sediment methylmercury concentration, with a maximum percent change of
62
-------
Evaluating R-MCMfor 91 VT/NH Lakes
100%. The other predicted concentrations had maximum percent changes ranging from
32% to 67%, falling into a relatively narrow range including that for the hypolimnetic
total mercury concentration.
The hypolimnetic methylmercury concentrations had a greater percent change
than the hypolimnetic total mercury concentrations even though the hypolimnetic
methylmercury concentration had the lesser absolute change. Therefore, the
hypolimnetic methylmercury concentration is more sensitive to R than the total mercury
concentration. There was a similar result with the epilimnetic methylmercury and total
mercury concentrations.
Sediment methylmercury concentration had the smallest predicted values and
absolute change (both average and maximum), followed closely by the sediment total
mercury concentration. This is similar to the hypolimnion concentration analysis in that
the magnitude of the absolute change was on par with the magnitude of the predicted
concentration itself. However, unlike the sediment methylmercury concentration, the
sediment total mercury concentration also had the smallest percent average change. For
the maximum change, although the absolute change was smallest for sediment
methylmercury concentration and followed closely by the total mercury concentration,
but epilimnion total mercury concentration had the smallest maximum percent change.
The sediment methylmercury concentration had the highest maximum percent change of
all predicted concentrations. This may be due to the order of magnitude smaller values
for the predicted sediment concentrations. Thus, an apparently slight absolute value
change in sediment mercury concentration is in fact a large percent change. On average,
the sediment mercury concentration was less sensitive to the hypolimnion surface area,
63
-------
Evaluating R-MCMfor 91 VT/NH Lakes
although one must remember that small changes in absolute value result in large
percentage increases.
The fish tissue mercury species concentrations were not as sensitive as the
hypolimnion species concentrations, but were more sensitive than the sediment
concentrations. Based on our statistical metrics, the fish concentrations were found to be
sensitive to the changes in the hypolimnion surface area. The average percent change in
fish mercury concentration of approximately 40% shows that changes in this one lake
parameter can result in relatively important effects on the modeling results. In Figure
6.5, there are only a few lakes that span relatively wide ranges of sediment and fish
mercury species concentrations. However, this figure does show how changes in
hypolimnion surface area can change a lake's specific model prediction result from one
of under-prediction to one of over-prediction for these two important modeling endpoints.
6.2.2 Non-dimensional Model Sensitivity Analysis
The visual analysis of the absolute and relative changes in Section 6.2.1 showed
that the order of magnitude of a predicted mercury species concentration affected the
calculated sensitivity of that prediction. Therefore, it is appropriate to use a non-
dimensional metric to analyze model sensitivity. Such a parameter for model sensitivity
is A, defined as the relative change in a predicted mercury species concentration divided
by the relative change in the parameter undergoing sensitivity analysis, then multiplied
by 100 to change this relative fractional change to a percentage relative change. This is
identical to the sensitivity analysis approach used in the Mercury Report to Congress,
Section 6.4 (EPA, 1997).
64
-------
Evaluating R-MCMfor 91 VT/NH Lakes
(C -C
^SR '-'i/:
V Cl/3 J
(EQN 6-4)
{ 1/3
where:
= the sensitivity of model output C to parameter R [percent]
C = model output [predicted mercury species concentration]
R = parameter being varied for sensitivity analysis (EQN 6-1)
Ci/3 = model predicted output value for base case (i.e., R = 1/3)
CSR = model predicted output value of changed parameter, R
1/3 = sensitivity parameter value for base case (i.e., R = 1/3)
RS = current model parameter value in the sensitivity simulation
EQN 6-4 defines the percent change in modeled species concentration with respect to a
fractional change from the hypolimnion surface area/epilimnion surface area ratio of one-
third. The sensitivity simulations produced concentration predictions for R = 1/2, 1/4 and
1/5 relative to the base case R = 1/3. The extent of model output sensitivity is directly
related to the value of A. A large A value equates to a greater sensitivity of the model to
R. An arbitrary system of A value ranges was defined to classify this sensitivity. The
sensitivity simulation results were then grouped into our four sensitivity level categories
as follows:
SENSITIVITY A RANGE
Extra Strong >99%, <-99%
Strong 50% to 99%, -99% to -50%
Moderate 25% to 49%, -49% to -25%
Weak -25% to 25%
65
-------
Evaluating R-MCMfor 91 VT/NH Lakes
The model sensitivity simulation results for this additional analysis are also
presented in Table 6.1. The results show that there is a range of sensitivities to R for all
of the predicted mercury species concentrations. On one extreme, it is clear that there are
no predicted concentrations for which the model has an extra strong sensitivity to R. At
the other end of the scale, the epilimnion and sediment total mercury concentrations have
only a weak sensitivity to changes in R for this model. This analysis also shows that the
methylmercury concentrations in the epilimnion, hypolimnion and sediment all have a
moderate to strong sensitivity to R. The averages of all the As showed a strong sensitivity
for the epilimnion, hypolimnion, and sediment methylmercury concentrations. The
model also demonstrated that predictions for fish tissue mercury concentration had a
strong sensitivity to R for all average As.
The results in Table 6.1 and the data presented in the appendix demonstrate that
the extent of model sensitivity is dependent on what region of the parameter space is
being evaluated. That is, the level of sensitivity was different for each value of R used to
calculate A. To gain a better understanding of the sensitivity of the model predictions,
the predictions were plotted against the different hypolimnion areas. These results are
presented in Figure 6.6 using the epilimnetic total mercury predicted concentrations as an
example. In this plot, there are a series of lines connecting four data points. Each line
represents a specific lake in the modeling, each with its own, specified characteristics.
Effectively, the characteristics represent an input vector that R-MCM uses to predict the
output variables. For a given series of points plotted in Figure 6.6 (those connected by
the line), the parameter vector is identical for all other parameters except for R. The data
points represent the predicted concentration for each of the four R. Because of the large
66
-------
Evaluating R-MCMfor 91 VT/NH Lakes
variability in the types of lakes, the lines connecting the points can cross each other. The
data points and lines plotted in Figure 6.6 show that the predicted concentration generally
decreases with increasing R. Upon visual inspection, it seems that there may be a
correlation between the magnitude of the slope and the magnitude of the predicted
concentration. Because the slope of the line corresponds to the amount of model
sensitivity to the parameter R, it is difficult to make a clear-cut evaluation of the
sensitivity over this wide range of lake characteristics. The slopes can be averaged (as
done in Section 6.2.1) to get a general sense of the model sensitivity, but this will not
capture the variability possible in parameter sensitivity.
To account for the variability of model sensitivity when predicting mercury
species concentration, a response surface of two variables was generated using a second
order polynomial and least squares regression. Specifically, the predicted mercury
species concentrations were approximated using the following polynomial:
C(/?,C1/3) = & +ftR + /32R2 + /?3C1/3 + /?4C1/32 +&/?C1/3 (EQN6-5)
where j3o-s are the fitted coefficients, R is the hypolimnion to epilimnion surface area
ratio, and C;/j is the predicted concentration for R = 1/3 (i.e., the base case predicted
concentration). The results of these linear regressions are plotted in Figures 6.7 - 6.13.
These figures are three-dimensional plots of the predicted mercury species concentration
for R = 1/3 versus R versus the predicted mercury species concentration at R. These plots
show how well the response surface fits the data, as well as the shape of the response
surface in general. The fitted coefficients, their standard errors, the adjusted R-square
value, and the F significance of the regression for each mercury species concentration are
67
-------
Evaluating R-MCMfor 91 VT/NH Lakes
presented in Table 6.2. Figures 6.7 - 6.13 and Table 6.2 together provide a clearer
picture of the model sensitivity to R.
The change in sensitivity due to the predicted concentration value itself and the
change due to the hypolimnion to epilimnion surface area ratio are effectively separated
through this technique. The value and standard error for each coefficient (ft) represent
the significance of each estimated coefficient and the influence of each coefficient on the
shape of the response surface. The adjusted R2 is one measure of the goodness-of-fit of
9 _
the model the higher the adjusted R , the better the model accuracy. The F value
indicates the significance level of the model accuracy (i.e., an F value of 0.01 indicates a
9 _
99% confidence level in the fit). The adjusted R and F significance values show that the
polynomial fit is quite strong. The worst F significance and adjusted R2 values were for
the hypolimnion total mercury concentration prediction at 8.2E-58 and 0.82, respectively,
demonstrating an excellent fit.
A reliable approximation of the parameter space by an analytical equation (such
as EQN 6-5) permits various mathematical manipulations to yield better insights into the
model sensitivity. For example, the instantaneous slope at any given point on the surface
can be estimated by taking the first derivative of the response surface equation. This
yields an exact and instantaneous sensitivity measure at each point within the parameter
space. The derivative of EQN 6-5 is given in EQN 6-6.
^- = j81+2j82R + j85Cl/3 (EQN 6-6)
oR
This derivative relationship describes the instantaneous change in predicted mercury
species concentration with respect to a change in R as a function of R, assuming that Ci/3
is held constant. Derivative values can be translated into a metric similar to A, so that a
68
-------
Evaluating R-MCMfor 91 VT/NH Lakes
non-dimensional metric value can be calculated as a percentage similar to EQN 6-4. The
sensitivity metric, DSM, is defined in EQN 6-7.
~J/"i D
.DSM = 100% (EQN 6-7)
c)R C
The results of these calculations are plotted in Figures 6.14 - 6.20. These figures are
plots of Ci/3 versus R with contours showing the regions of DSM. The lines plotted
represent contour lines for constant values of constant DSM. A range of different colors
are used between the contour lines to make it easier to see zones between the lines of
constant values. The colors have no other significance other than to assist the reader in
viewing the plots. The single color between contours does not mean that DSM is
constant, but rather shows the region bounded by the constant value contours. Figures
6.14 - 6.20 present the regions of weak (-25% to 25%), moderate (25% to 49%, -49% to -
25%), strong (50% to 99%, -99% to -50%), and extra strong (>99%, <-99%) sensitivity
(The figures also reveal regions as R increases and decreases. There are added contour
lines to provide insight into the shape of the response surface where DSM is near zero.)
Additionally, the observed data points associated with each lake are plotted similar to
Figure 6.6. The four data points associated with the four R values used in the sensitivity
analysis are plotted and connected by a solid line to delineate points from the same lake.
These points also show what the predicted concentration for a given lake becomes as R
changes, and how the model output sensitivity changes across the R parameter space.
The predicted concentration is plotted on the y-axis to show the predicted concentration
for each lake, but the figure really is plotting the sensitivity regions of the response
surface based on the base case, Ci/3, and EQN 6-7.
69
-------
Evaluating R-MCMfor 91 VT/NH Lakes
A specific feature of EQN 6-7 must be mentioned, specifically that it is a
hyperbolic function multiplied by the derivative. As C approaches zero, the hyperbola
approaches infinity, and the function is undefined at this point. Therefore, DSM will
approach either positive or negative infinity near the x-axis. All the results in the figures
from this method of sensitivity analysis agree with the results presented in Table 6.1.
Figure 6.14 presents the sensitivity of predicted epilimnetic methylmercury
concentration. For the range of lakes and associated characteristics (parameter space)
investigated, the model prediction sensitivity to R was moderate to strong. For lakes with
predicted concentrations for Ci/3 greater than 0.05 ng/L, the model became more sensitive
to R as R increased and less sensitive to R as R decreased. When Ci/3 decreases below
approximately 0.05 ng/L, the impact of Ci/3 approaching zero becomes evident as the
model sensitivity starts increasing rapidly. This figure verifies that the epilimnion
methylmercury concentration has moderate to strong sensitivity to hypolimnion surface
area.
Figure 6.15 presents the sensitivity of predicted epilimnetic total mercury
concentration. Most of the parameter space shows that the model has a weak sensitivity
to changes in R and Ci/3. The sensitivity changes to moderate in the upper right corner of
the region, as the predicted concentration and the hypolimnion area increase. The same
asymptotic approach towards infinity arises in the lower right corner. This figure verifies
that the total epilimnetic mercury concentration is relatively insensitive to the
hypolimnion surface area.
Figure 6.16 presents the sensitivity of predicted hypolimnetic methylmercury
concentration. This figure is similar in shape to Figure 6.14. Most of the parameter
70
-------
Evaluating R-MCMfor 91 VT/NH Lakes
space shows that the sensitivity is moderate. Sensitivity increases with increasing R. The
band of moderate sensitivity spans a wider region of R than in Figure 6.14, crossing into
an area of weak sensitivity to the left, and into a strong sensitivity on the right. The
increasing hyperbolic shape as predicted concentration decreases toward zero once again
exhibits the asymptotic nature of the function. This figure verifies that the model has
predominately a moderate sensitivity to hypolimnion surface area.
Figure 6.17 presents the sensitivity of predicted hypolimnetic total mercury
concentration. This figure shows a wide region of sensitivity near zero. Additional
contours (-5 and -10) were added to specifically show the shape of the function across the
upper half of the figure. As the predicted hypolimnetic total mercury concentration
decreases and R increases, the model sensitivity increases. The contour where the
sensitivity changes from weak to moderate is curved, but occurs between the predicted
concentrations of 1 to 2 ng/L. Here the slope of the sensitivity is increasing with
decreasing C\n. This figure shows the importance of an analysis of this type. The
previously discussed average calculations showed a moderate to strong sensitivity for this
prediction, but there is a strong correlation between sensitivity and predicted
concentration value. For lakes where a high hypolimnetic total mercury concentration is
predicted, the model is relatively insensitive to the hypolimnion surface area. However,
as the predicted concentration decreases, the sensitivity changes from weak to moderate,
even passing into a strong and extra strong region as Ci/3 approaches zero.
The first and most obvious observation from Figures 6.14 - 6.20 is that the
methylmercury concentration is more sensitive to changes in R than total mercury
concentration. Both the discrete and continuous analyses showed that the methylmercury
71
-------
Evaluating R-MCMfor 91 VT/NH Lakes
concentrations in the epilimnion, hypolimnion, and sediment were all moderate to
strongly sensitive to changes in R. Fish concentrations, which are modeled as
methylmercury concentrations in fish tissue, were similarly sensitive. The total mercury
concentrations in the epilimnion, hypolimnion and sediment were less sensitive, falling
into the weak and moderately sensitive regions. There were regions within the parameter
space where the total mercury concentrations became more sensitive, but predominately,
the total mercury concentration remained relatively insensitive.
6.2.3 Re-evaluating Hypolimnion Area Sensitivity by Keeping Constant
Volumetric Flow Rate (Adjusting dwith V)
Based upon the results of the previous section's sensitivity analysis, it was unclear
as to why there was the evident inverse correlation between the hypolimnion area and the
mercury concentrations in the different media. The processes and modeling structure of
the R-MCM was explored in an attempt to understand this phenomenon. During this
review, we realized that by changing R (and thus hypolimnion area) the volumetric flow
rate through the lake was being inadvertently changed. The R-MCM calculates the
volumetric flow rate, Q, through the lake as a function of the lake volume, V, and the
hydraulic residence time, 6.
Q=^ (EQN6-8)
0
Certain parameters are defined during the initializing of the parameter set. Of particular
importance to this section, the R-MCM has the following parameters defined as part of
the input parameter vector: mean depth of each lake layer (i.e., the hypolimnion and
72
-------
Evaluating R-MCMfor 91 VT/NH Lakes
epilimnion), the lake area (epilimnion surface area), the thermocline area (hypolimnion
surface area), and the hydraulic residence time. When the R-MCM is run, the model
calculates the other required parameters. Specifically, the R-MCM first calculates the
volume of each layer in the lake as the product of the given layer's surface area and its
mean depth.
VE = AEdE (EQN 6-9)
VH = 4A = RAEdH (EQN 6-10)
Then, the total lake volume, VT, is calculated as the sum of the volumes of the two layers.
VT = AEdE+ AHdH (EQN 6-11)
AE and AH are the areas of the epilimnion and hypolimnion, respectively, and dE and du
are the mean depths of the epilimnion and hypolimnion, respectively, with R defined in
EQN 6-1. Thus,
VT = AE(dE+ RdH). (EQN 6- 1 2)
Then the volumetric flow rate, Q, is calculated via EQN 6-8. In the analyses of the
previous sections (Sections 6.2.1 - 6.2.2), it was not realized that as the hypolimnion
surface area was being updated that R-MCM was also recalculating Q. Therefore, Q was
inadvertently being decreased as V decreased. Therefore, the calculations and analyses of
Section 6.2.1 needed be to be performed again. To do this, the Q updating relationship
needed to be formulated.
Keeping Q constant based on the initial volume, VQ, for the updated volume, V;, with
corresponding do and #/, gives
6>0
(EQN 6-13)
73
-------
Evaluating R-MCMfor 91 VT/NH Lakes
and through rearrangement
ซ-^J-
and because the area of the epilimnion is not changing, AE,O = AE,I, and because the
depths of the epilimnion and hypolimnion are not changing, dฃ,o = ds,i and dn.o =
(EQN 6-15)
Using this formulation, a similar evaluation of the effect of changing R done in Section
6.2.1 and 6.2.2 was performed for the same series of R-MCM runs (R = 1/5, 1/4, 1/3, and
1/2) for the predicted mercury species concentrations in all media (MeHg and HgT in the
epilimnion, hypolimnion, and sediments). Microsoft Access and SQL programming
codes were used to calculate and update AH and 6 for each run of R value run. Particular
care was taken to keep data files and data manipulations separate and distinct to assure a
high level of data quality. Rerunning the model using the updated constant Q values
resulted in similar results as when Q was not updated (as was done in Section 6.2.2).
There were slight shifts in the predicted concentration values, and the extent of sensitivity
to changes in R was slightly decreased. However, overall, there was no appreciable
difference. It was still unclear what was impacting the predicted mercury species
concentration due to changes in hypolimnion area.
6.2.4 Investigation of Mechanism Causing Increase in Mercury
Concentration with Decreasing Hypolimnion Surface Area
Upon conclusion of the previous investigations (Sections 6.2.1 - 6.2.3), the
sensitivity of predicted mercury species concentrations to hypolimnion surface area had
74
-------
Evaluating R-MCMfor 91 VT/NH Lakes
materialized into a clearer phenomenon, however it was still unclear what the mechanism
was. Therefore, additional modeling experiments were performed to ascertain if this
phenomenon was indeed a modeling phenomenon, or if it were just an artifact of the
modeling system. A hypothetical lake was created using the default set-up capability of
the R-MCM to evaluate a simple lake system because we were interested in model
behavior independent of the VT/NH data set. The lake used was an: acidic, stratified,
medium-sized, drainage, oligotrophic lake. Using this hypothetical lake, the R-MCM
was initially run for R = 1/10, 1/2, and 9/10, keeping Q constant. During this preliminary
analysis, the same pattern of increasing predicted mercury species concentrations with
decreasing hypolimnion area was demonstrated, even for the extreme values of R = 0.1
and 0.9. These modeling results are presented in Table 6.3, labeled as the base case, run
2, and run 3, and in Figure 6.21. These results show again the previously noted inverse
correlation between predicted mercury species concentration and hypolimnion area and
R.
An additional investigation was performed, keeping hypolimnion area constant
but going to extremes with mean hypolimnion depth. In the base case, mean
hypolimnion depth was set as 5 m. For this part of the investigation, the other mean
hypolimnion depths were set at a minimal value of 0.1 m and an extreme value of 1000
m. These modeling results are also presented in Table 6.3 labeled as the base case and
run 4 and run 5, and in Figure 6.22. These model results showed that increasing
hypolimnion depth had little impact on the predicted mercury species concentrations in
the epilimnion, but produced an increase in the HgT and a decrease in Hgll concentration
in the hypolimnion. These results merely underscore the probable importance of
75
-------
Evaluating R-MCMfor 91 VT/NH Lakes
photolytic reactions and how the thickness of the hypolimnion layer impacts the
distribution and concentrations of the various mercury species, but they did not help
elucidate the processes governing the observed impact of hypolimnion surface area itself.
In reviewing Figure 6.21, it is difficult to make any general conclusions with only
3 points of data, so it was decided that additional R values would be modeled to see if
there were a limiting values that the various mercury species concentrations would
approach. Specifically, it was of interest to see how the shape of the curve reacted in the
limit that R -> 0. The model lake was set up as before, except in this case, both the
epilimnion and hypolimnion had mean depths of 5 m. A range of R values was used to
give a better idea of the shape of the response function. The R values used were: 0.0001,
0.05, 0.1, 0.25, 0.5, and 0.95. Two different cases were run for the limiting result of R =
0. The first was as if the hypolimnion were completely removed. For this case, the
model was run as a well-mixed lake with the total lake are equal to the exact dimensions
of the epilimnion surface area used in the other modeling scenarios. A second case was
run as if the total volume of the lake for the R = 0.95 case were actually a well-mixed
lake, so that the lake had nearly twice the volume as the other well-mixed case. The
modeling results for these two cases are presented in Table 6.4 (labeled as runs 1-1 and 1-
2 along with the results at R values of 0.0001 to 0.5. The data are plotted in Figure 6.23
for the range of R values, as well as for the two limiting cases of R=0. The first case is
plotted as distinct points at R= 0 as "Well-Mixed, Top Layer Only," and the second case
is plotted at R=l as "Well-Mixed, Both Layers Modeled as One." From these results, the
same pattern was demonstrated as previously seen in Sections 6.2.1 - 6.2.3.
76
-------
Evaluating R-MCMfor 91 VT/NH Lakes
The inclusion of more data points (see Figure 6.23) gives a better idea of the
shapes of the response curves and the extension of the results to the limits of R = 0 and R
= 1. The results of the two well-mixed scenarios provide additional, interesting
information. For one result, which is akin to an effective R = 0, where the hypolimnion
layer goes to zero, the predicted mercury species concentrations in the epilimnion
increased quite dramatically. As R-> 0, the predicted epilimnion mercury species
concentrations approached values of 0.14 ng/L (MeHg), 1.5 ng/L (Hgll), and 1.7 ng/L
(HgT). However, for a lake of identical dimensions, but with an effective epilimnion
only, that is, for a well-mixed lake of the same dimensions as the epilimnion in the other
scenarios, the predicted mercury species concentrations were: 0.3 ng/L (MeHg), 3.0 ng/L
(Hgll), and 3.3 ng/L (HgT), respectively. The results of the well-mixed model runs were
approximately twice the limiting value as R -> 0. It is unclear as to why this occurs, and
it is also unclear which value is the more likely value to be measured in the field.
Additionally, a simulation was made where the same dimensions as the R= 0.95 lake
were run as a well-mixed lake. This would be the case of R -$ 1, but running the model
as a well-mixed lake instead of a stratified lake. These model runs produced a similar
unexplainable gap as the R -$ 0 results just discussed (see Figure 6.23).
6.2.5 Hypothetical Lake Evaluation of the Change in Simulated Mercury
Species Concentrations as a Function of Hypolimnion Area: A
Simple Mathematical Thought Experiment
Upon conclusion of the work in the previous section, we became more confident
of the sensitivity of the model and the presence of the phenomenon of a decrease in
77
-------
Evaluating R-MCMfor 91 VT/NH Lakes
mercury species concentrations as the size of the hypolimnion area increased. Therefore,
in addition to the R-MCM modeling experiments, we believed it would be productive to
investigate the phenomenon in a purely theoretical, mathematical analysis. In this
section, we merely developed a very simple mathematical thought experiment that
allowed us to investigate only a few, simple processes so that we might understand the
physical nature of this system.
A simple construct of a stratified lake was modeled as shown here.
Vl
V2,C2
Q'
Vl
V2
--out
In this simplified two-layer lake system, there is
Qin: Volumetric Flow into the lake (into layer 1 only) [m 3/d]
Qout: Volumetric Flow out of the lake (from layer 1 only) [m 3/d]
Cin: Concentration in Inflow [ng/L]
Cout: Concentration in Outflow [ng/L]
Ci: Concentration in Layer 1 [ng/L]
C2: Concentration in Layer 2 [ng/L]
Vi: Volume of Layer 1 [m ]
V2: Volume of Layer 2 [m3]
vi: Effective Settling Velocity of particles from Layer 1 to Layer 2.
Particles settle from Layer 1 to Layer 1 sediments and into Layer 2 as
appropriate [m/d]
v2: Effective Settling Velocity of particles from Layer 2 to Layer 2
78
-------
Evaluating R-MCMfor 91 VT/NH Lakes
sediments [m/d]
Q': Exchange Flow between Layers 1 and 2. [m3/d]
The two layers were modeled as Continuous-Flow Stirred Tank Reactors
(CSTRs), so that Cout = Q and the total mass concentration of each mercury species is
assumed constant throughout each layer. Mass Balance Equations for each layer are as
follows:
Layer 1 : = Q{nCin - QoutCv - v^Q - G'fc - C2 ) (EQN 6-16)
at
Layer 2: d(V2C2) = v^c, - v2A2C2 - Q'(C2 - Q ) (EQN 6-17)
dt
In these equations, we define an "effective settling velocity." Generally, one does not
incorporate a settling term in a CSTR, however with solid particulates in the system,
settling can and will occur. To keep our model simple, we incorporate the fraction of
total concentration that might settle (i.e., that fraction of the total concentration is sorbed
to the settling matter) within the velocity term. We could have, in effect, incorporated a
term defining this fraction, such as /, as a multiplier to the concentration. This would
result in two parameters, v and/, and since we have no information on either of these, it is
just as simple, and much cleaner mathematically, to incorporate them into one lump,
"effective" parameter.
Assuming Steady State, then - ; = 0 and EQN 6-16 and 6-17 become
dt
Layer 1: QinCin -QoutCv -v^Q -Q\C, -C2) = 0 (EQN 6-18)
Layer 2: v^Q -v2A2C2 -Q'(C2 -Ct) = 0 (EQN 6-19)
After some rearrangement,
79
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Layer 1 : Q = QC*+Q*C* (EQN 6-20)
Q^+vA+Q
Layer 2: C2 = M2+6)Q (EQN 6.21)
v2A2 +<2
From these equations, we see that Ci is a function of Ci, and vice- versa. This
complicates our attempt to understand the influence of the hypolimnion area on the
concentrations in each layer. Two approaches were taken to investigate the behavior of
these coupled equations. First, a brute force method was used by assigning representative
but arbitrary values to the various parameters. An MS Excel Spreadsheet was designed
to iterate the solutions for C2 and C; as a function of R, where R = A2/Ai. Q' was
calculated by
Q' = L, (EQN 6-22)
where En is the exchange rate coefficient, as presented by Schnoor (1996) (from
Mortimer, 1941), E = 0.0142*(rf;+ rf;)1'49 and
l=-(d,+d2], (EQN 6-23)
where / is the distance between the midpoints of the lake layers [m].
Three different runs were performed with the results presented in Figure 6.24.
Run (a) represents a scenario where the particle settling rate is greater in the hypolimnion
than in the epilimnion, and Run (b) is the opposite. Run (c) uses settling velocities that
were calculated from the default run modeled previously in Section 6.2.1.
Lastly, a purely mathematical evaluation was performed using Mathematica
(Mathematica, v 5.0.1.0, Wolfram Research, Inc.) to calculate the derivatives of EQN 6-
20 and 6-21. The results were:
80
-------
Evaluating R-MCMfor 91 VT/NH Lakes
*- (EON 6-24)
(EQN6-25)
(Q (Qout +(A1-A2)v1) + A2 (Qout + g'+M )v2 )2
CinQinQ' (Qout +
dA2 (Q (Qout + (A - A2 X ) + A2 (Qout + Q'+AlVl )v2 )2
From these equations, it was clear that the shape of these derivative curves can be quite
complicated, depending on the input variables. However, one thing that can be simply
derived from these equations is that
J/~*
dA2
(EQN6-26)
Therefore, if the settling rate is greater in the hypolimnion than in the epilimnion, (i.e., V2
> vi), then C decreases as the hypolimnion area decreases. (This is true as long as AI >
A2, which is true in all real lake systems because the hypolimnion area is necessarily
smaller than the epilimnion area.) An example of EQN 6-26 behavior is presented in
Figure 6.24. In Figure 6.24 (a), vi > v2 and CT increased with R, while in (b) vi < v2 and
CT decreased with R. Using actual settling velocities from the R-MCM, the curve
becomes more gradual exhibiting the characteristic decrease in mercury concentration as
R increases seen in the R-MCM runs (6.24(c)). From this simple mathematical analysis,
it seems that the decreasing concentration is a function of the physical structure of the
lake system. The difference in lake layer particle (with sorbed Hg) settling velocities in
this simple system creates the phenomenon and sensitivity of the model output for total
mercury concentration to hypolimnion area.
In this analysis, in contrast to the R-MCM outputs, the epilimnion concentration
was found to be greater than the hypolimnion concentration; it is hypothesized that this
results is due to the oversimplification of the model. Because this simplistic
81
-------
Evaluating R-MCMfor 91 VT/NH Lakes
mathematical representation does not take into account the different mercury species, the
true variation in mercury inputs, outputs and transformation processes are not adequately
modeled. For example, the epilimnion will have a volatilization process governing
additional loss of HgO, while the hypolimnion does not have this process. Additionally,
the depth governing the photoreduction process is not adequately captured. Incorporating
processes associated with depth (as these are) would impact the ratio of epilimnion to
hypolimnion mercury concentrations.
6.2.6 Summary
The hypolimnion surface area was originally defined as a default parameter in the
R-MCM as a fraction of the epilimnion surface area, taken to be the lake surface area.
The lake surface area is a relatively straight-forward measured parameter, while
measuring the hypolimnion surface area provided a bit more of a challenge. The
hypolimnion is a physically existing layer within a lake, but its thickness is a time-
dependent variable. Specifying hypolimnion thickness is a challenge, but at any given
point in time it can be measured. Since the R-MCM is a steady-state model, both the
thickness and area of the hypolimnion must necessarily be assumed to be constant.
Specification of hypolimnion surface area is complicated because the cross-sectional area
of the hypolimnion is dependent on the hypolimnion thickness through the bathymetry of
the lake. Depending on the specific shape of the lake floor, the hypolimnion area could
vary widely for any given thickness or lake size. Given this complication, to decide upon
a representative hypolimnion surface area would be a time-intensive effort, and the
achievable accuracy of this effort is unclear. Additionally, the resulting amount of model
82
-------
Evaluating R-MCMfor 91 VT/NH Lakes
improvement that could be achieved is not clear. The model and mathematical
investigation presented in Section 6.2 have shown how the predicted mercury species
concentrations are sensitive to the hypolimnion area. Further research should be
conducted to investigate if significant improvements in mercury modeling can be made
via a more rigorous physical representation of the lake system, especially in regard to
modeling the size of the hypolimnion.
Based on our work, some conclusions regarding the sensitivity of modeled
mercury species concentrations to hypolimnion surface area can be made:
Methylmercury concentrations (epilimnion, hypolimnion, sediment, fish
tissue) were all found to be more sensitive than the total mercury
concentrations,
Methylmercury concentrations were found to have sensitivities falling in
the moderate and strong regions,
Sensitivities were generally found to be negative, suggesting an inverse
correlation between mercury species concentrations and hypolimnion
surface area, and
Generally, the predicted mercury species concentrations became more
sensitive to changes in hypolimnion surface area (K) for the larger lakes.
6.3 Comparison of Model Behavior and Observed Data
In Chapter 3, the predicted results for a series of default-parameterized lakes were
presented to illustrate the general trends and behavior that the default level model
provides. In this section, the observed data was analyzed in an effort to understand if
83
-------
Evaluating R-MCMfor 91 VT/NH Lakes
there are trends within the VT and NH data itself. Any observed trends can then be
compared to the predicted results to see if similar trends are produced by the mechanistic
model. This evaluation of observed data and comparison of trends provides insight into
the mechanisms governing mercury cycling in lake ecosystems.
6.3.1 Trends in Observed Data
The observed data from the VT and NH data set were plotted as a function of the
default-level lake characteristics, similar to the figures created by plotting the predicted
concentrations from the hypothetical default lake. The observed results were plotted
using the continuous data provided by lake-specific measurements, rather than the
categorical/discrete data entered for the default level input. The measured values for
epilimnion methylmercury and total mercury, hypolimnion methylmercury and total
mercury, sediment methylmercury and total mercury, and fish tissue mercury
concentrations are all plotted as a function of lake area, epilimnion DOC, epilimnion pH,
and lake stratification. Because precise measurements of the first three characteristics
were made for each lake, these are plotted as continuous variables, unlike lake
stratification, which is not continuous, but rather is a binary function being stratified (1)
or well-mixed (2). These data plots are presented in Figures 6.25 and 6.26. It is clear
from these figures that there is a large amount of scatter in these data.
6.3.1.1 Lake Area
For epilimnion methylmercury, there may be an inverse correlation with lake
area. There is appreciable scatter in the epilimnion methylmercury data versus lake area,
but the average value appears to be decreasing with increasing lake size. The scatter
84
-------
Evaluating R-MCMfor 91 VT/NH Lakes
amongst the data cannot be discounted, however. For the other observed mercury species
concentrations, the scatter of the data dominated over any possibly visible trends.
6.3.1.2 Epilimnion DOC
In the plots of the different mercury species concentrations in the different media
against epilimnion DOC, there did not seem to be any noticeable trend. The scatter
amongst the data overwhelmed any possible trends.
6.3.1.3 Epilimnion pH
Similarly in the plots of the different mercury species concentrations in the
different media against epilimnion pH, the scatter also tended to dominate the results.
However, for the epilimnion methylmercury there was an apparent trend of decreasing
concentration with increasing pH (i.e., decreasing acidity). This trend is similar to that
seen in the R-MCM results presented in Section 3.2.3. There also was a noticeable trend
in perch tissue mercury concentration versus the epilimnion pH. As with the other data,
there was still scatter in the fish tissue data, but the inverse correlation is noticeable. This
trend was also evident in the R-MCM output presented in Section 3.2.4. The inverse
correlation of water and fish tissue mercury concentration with pH has also been
documented in the field at other field sites. It is somewhat surprising that the epilimnion
total mercury concentration, as well as the methylmercury and total mercury
concentrations in the hypolimnion and sediment, did not demonstrate similar correlations
with epilimnion pH.
85
-------
Evaluating R-MCMfor 91 VT/NH Lakes
6.3.1.4 Lake Stratification
There is a large amount of scatter in the data when lake types were separated into
stratified and well-mixed lakes. This scatter generally dominated any trends, although
there are some minor trends among the observed results. For the epilimnion
methylmercury concentration, there is a slight increase in the average concentration in the
well-mixed lakes as compared to the stratified lakes, but there is an outlier in the
stratified lakes that is largest of any of the epilimnion methylmercury concentrations.
Similarly, for the epilimnion total mercury concentration, the average value is larger in
the well-mixed lakes. There are two stratified lakes that had appreciably higher total
mercury concentrations than the other stratified lakes, and there were two well-mixed
lakes with appreciably higher total mercury concentrations than the other well-mixed
lakes. These outlying lakes also followed the trend of having the higher concentrations in
the well-mixed lakes versus the stratified lakes.
For sediment methylmercury, there also appeared to be higher concentrations in
the well-mixed lakes than in the stratified lakes, but the data scatter is appreciable enough
to question if there really is any sort of trend. The scatter in the sediment total mercury
concentrations dominated over any possible trend with lake stratification. That was also
the situation with respect to the fish tissue mercury concentration. There is one outlying
fish mercury concentration in a stratified lake, but exclusive of this one lake, there was
little difference in the average and range of observed fish tissue mercury concentrations.
6.3.2 Percent Methylmercury for Observed and Predicted Data
An important facet of mercury modeling is the production, loading, and loss of
methylmercury within a watershed and water body. There has been much debate,
86
-------
Evaluating R-MCMfor 91 VT/NH Lakes
discussion and research on the relative importance of the methylmercury concentration in
a water body versus the total mercury concentration in the same system. Therefore, the
percent of total mercury concentration that is methylmercury as an additional model
output was evaluated. Both the predicted model results and the observed results were
compared solely on a percent methylmercury basis. The percent methylmercury
(%MeHg) was calculated according to EQN 6-8.
%MeHg = x 100 (EQN 6-8)
HgT
6.3.2. 1 Visual Analysis
The results of this comparative analysis are visually presented in Figure 6.27. In
this figure, for both the first and second column the x-axis is arbitrary, since the lakes are
plotted on the figure merely in order of lake number. Therefore, the order of the data is
irrelevant. The scatter of the data for both the predicted and observed results gives a
feeling for the range and mean of the data. The scale of the y-axis is different for each
subfigure. The observed results for the epilimnion range up to almost 80% MeHg, while
the predicted results range up to about 45% MeHg. For the percent methylmercury in the
hypolimnion both the predicted and observed data fall within a similar range of up to
about 35 - 40%. For the percent methylmercury in sediment, the predicted values have a
wider spread by about a factor of 2 than the observed results.
In the third column of Figure 6.27, the observed data are plotted versus the
predicted data. If the R-MCM were performing as a perfect model of the environment,
then these latter results would fall along the y=x line. If the predicted and observed
values displayed similar ranges, then these latter data would fall in a square splay
87
-------
Evaluating R-MCMfor 91 VT/NH Lakes
centered about the y=x line. Our data do neither. Specifically, the epilimnion data are
clustered along the x-axis, indicating that the predicted percent methylmercury was
consistently smaller than the observed. The hypolimnion data have the opposite pattern,
being clustered along the y-axis, showing how the predicted percent methylmercury was
consistently greater than the observed. For the sediment, the data are predominately
clustered near the origin, packed along the x-axis, showing how, similar to the
epilimnion, the sediment percent methylmercury is consistently smaller than the observed
by a factor of roughly 2. Also, for the sediment, however, there are two outliers, showing
that this pattern is not entirely consistent. There is one outlier where the observed
methylmercury concentration is quite high, while the corresponding predicted value is
quite low; and conversely, there is another outlier where the predicted value is high and
the observed value is quite small. The analysis of sediment percent methylmercury is
confounded by the fact that one is dealing with relatively very small concentrations, and
any errors in rounding or sampling have a dramatic impact on the calculated percent
methylmercury.
6.3.2.2 Summary Statistics
In addition to the previously discussed graphical representation of the data,
summary statistics for the predicted and observed percent methylmercury results in the
three different media (i.e., epilimnion, hypolimnion, and sediment) were calculated and
tabulated in Table 6.6. The minimum and maximum values for the three media elucidate
how well the model does at predicting the extreme values of percent methylmercury. For
both the epilimnion and sediment, the model predicted zero percent methylmercury as the
minimum, when this was not the case. This could be just an issue of significant figures
88
-------
Evaluating R-MCMfor 91 VT/NH Lakes
(i.e., the model rounds the numbers and they come out to be zero when they are just
smaller than the rounding error), or that there are difficulties in predicting methylmercury
concentrations in lakes with these more extreme characteristics. Similarly, on the other
side of the scale, there were difficulties in predicting the extreme methylmercury
percentages. For example, the percent methylmercury in the epilimnion, the R-MCM
predicted a maximum of 43% when a value of 75% was observed. The model under-
predicted this percent by almost half. For the hypolimnion, the maximum predicted
percent methylmercury (35%) was just under the maximum observed percent
methylmercury (39%). For the sediment, the R-MCM over-predicted the maximum
percent methylmercury (40%) compared to the maximum observed methylmercury
(29%) by roughly one-quarter. Therefore, relative to the maximum percent
methylmercury, the model behaved differently in the three different media. In summary,
for maximum percent methylmercury, R-MCM under-predicted for the epilimnion,
overpredicted for sediment, and was on par for the hypolimnion.
The mean percent methylmercury comparison provides insight into how well the
model is behaving overall. For the epilimnion, the model predicted a mean of 8.6%
methylmercury while the observed data had a mean of 21%. A report by Krabbenhoft et
al. (1999) suggested that the average observed percent methylmercury in water draining
mixed agricultural and forested areas was approximately 8%. For many lakes in Vermont
and New Hampshire, the land use type is mixed agricultural and forested, so the model-
predicted value of 8% would seem to be appropriate. Our mean observed value of 21%,
therefore, seems rather high for these lakes. In a report by Kelly et al. (1995), an average
percent methylmercury was reported as 15.5% for a headwater wetland, suggesting that
89
-------
Evaluating R-MCMfor 91 VT/NH Lakes
some wetlands might have percent methylmercury values on the order of 20%. The
discrepancy between the average percent epilimnion methylmercury for the R-MCM
predictions and our observations suggests that the R-MCM, as is, is unable to predict
these relatively high methylmercury percentages. This suggests that processes affecting
methylmercury concentration need to be adjusted, or additional processes and/or loadings
need to be included. Also, it seems that our study lakes in VT and NH behaved
differently than the collection of lakes studied earlier by others, where the average
methylmercury percentage was typically nearer 10%.
Our mean methylmercury percentages for the hypolimnion and sediment were
observed to be 9.2% and 2.4%, respectively, while the percentages predicted by R-MCM
were 17% and 5.7%, respectively. Compared to the epilimnion, the hypolimnion and
sediment had much lower observed mean methylmercury percentages. As mentioned
earlier, it is of interest that the R-MCM is not consistent in its predictive capability for the
percent methylmercury. Krabbenhoft et al. (1999) suggested that for a mixed agricultural
and forested land use, the methylmercury percentage should be higher for sediments than
for the water, the former with a value of approximately 10%, while a background
reference or agriculturally dominant site should have a value of approximately 5%. The
R-MCM is only predicting levels similar to the latter, and these are significantly over-
predicted compared to the observed results. No typical or average methylmercury
percentages were found reported for the hypolimnion.
6.3.3 Summary
Trends in the observed data were evaluated, and comparisons made between the
observed and predicted model results. For the observations, it was clear that a large
90
-------
Evaluating R-MCMfor 91 VT/NH Lakes
amount of scatter was present in the data that overwhelmed any trends. Most of the
trends present in the original default model results (see Section 3.2) were not as clearly
pronounced in the observed values. For example, there was a general, all media inverse
correlation demonstrated for lake area in the R-MCM default set-up lakes modeling
work. However, this trend was present in the observed data only for methylmercury
concentration in the epilimnion.
No pattern of mercury species concentration as a function of trophic status was
evident in the observed results. For epilimnion methylmercury concentration, there was a
possibly inverse correlation with pH, and an even stronger inverse correlation for fish
tissue mercury concentrations with pH for the observed data. The other observed
mercury species concentrations did not demonstrate similar trends. The R-MCM
predicted much stronger mercury species correlations with pH than the observed results
demonstrated.
The percent of total mercury that is methylmercury was calculated for all three
media (epilimnion water, hypolimnion water, and sediment). The observed results
showed a much wider range of values in the epilimnion than did the R-MCM predictions,
while the opposite was true in the sediment. The maximum methylmercury percentage
decreased from the epilimnion down to the sediment. The model was unable to predict
the high observed maximum methylmercury percentage in the epilimnion, while it over-
predicted the maximum methylmercury percentage in the hypolimnion sediment by a
factor of 2. For the mean sediment methylmercury percentage, the R-MCM predicted a
value that was half the observed value.
The results of the analysis within this section lead us to a few conclusions:
91
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Plotting observed results versus lake characteristics produced a lot of
scatter and only a few weak trends;
The range and maximum values for the epilimnion percent methylmercury
are greater for the observed than the predicted values, whereas the range
and maximum values for the sediment percent methylmercury are greater
for the predicted than for the observed values;
The R-MCM predicts a smaller mean percent methylmercury for the
epilimnion than the observed results, but predicts a larger mean percent
methylmercury for the hypolimnion and sediment;
The observed mean percent methylmercury decreases with depth
(epilimnion is largest, hypolimnion is in the middle, and sediment is the
smallest), while the modeled mean percent methylmercury is greatest in
the hypolimnion, followed by the epilimnion, with the sediment having the
smallest value;
The R-MCM predicts a mean epilimnion percent methylmercury of 8.6%,
which is very similar to a literature-reported observed average for lakes,
while our observed result, 21%, is more than twice this predicted value
and is greater than or on the order of values reported for wetlands; and
The hypolimnion and sediment mean methylmercury percentages are half
those of the predicted. However, the predicted sediment value falls within
a range of literature values for agricultural, background, or mixed
agricultural and forested watersheds.
92
-------
Evaluating R-MCMfor 91 VT/NH Lakes
The scatter present in our data and the lack of the more obvious trends present in
the R-MCM outputs underscore the complications of modeling these systems, and that
simple trends relative to a few specific characteristics are not enough to classify and
predict the various mercury species concentrations. The differences in the ranges of
percent methylmercury values point towards more specific problems. The inability to
cover the ranges necessary for the epilimnion suggests the possibility of two things.
First, there may be additional methylmercury production occurring within the water
body, resulting in the higher observed methylmercury percentages that are not modeled.
That is, if the in-situ methylation rates were higher, or demethylation rates were lower in
the model, then the predicted methylmercury percentage would be increased. The R-
MCM, as it stands, may not currently be parameterized to completely capture the
methylmercury transformation processes. Secondly, it is possible the methylmercury
loading rate into the water body may be higher than is currently modeled. It is unlikely
that this unaccounted-for mercury loading is coming directly from the atmosphere,
because it is relatively well understood that the primary atmospheric loading is in the
divalent inorganic mercury form (Hgll). Therefore, a different loading source may need
to be investigated. One possibility is that other methylmercury loading sources may be
incoming rivers and streams or direct watershed runoff. That is to say, processes
upstream and within the watershed may be increasing the methylmercury loading to the
lakes. This latter hypothesis may be particularly applicable since we observed that the
methylmercury percent was greatest in the epilimnion, and it is well understood that
methylation of mercury occurs primarily in the sediment; recall that in our study the
93
-------
Evaluating R-MCMfor 91 VT/NH Lakes
model is over-predicting the mean methylmercury percentage in the sediment, but under-
predicting it in the epilimnion.
6.4 Watershed Influences and Loading
The R-MCM focuses primarily on those processes governing mercury fate and
transport within the receiving water body, but uses a relatively simplistic approach to
model the watershed loading to the water body. This is understandable because much
attention has been focused on the mechanisms and chemistry affecting mercury behavior
within a water body, and the major source of mercury to any water body was assumed to
be via atmospheric deposition. The simplistic perception of the overall mercury transport
process can be envisioned as atmospheric transport of mercury emissions to the
atmosphere as elemental mercury and divalent mercury (and the oxidation-reduction
reactions between the two in the atmosphere), deposition as divalent mercury to the
water, and then transformation to methylmercury in major water body sediments. The
processes governing the transformation and transport of mercury once it enters a water
body are rigorously modeled in the R-MCM. However, the mechanics of loading to the
system (i.e., how the mercury actually enters the system) focuses primarily on
atmospheric inputs, with a relatively simple methodology for quantifying the watershed
mercury inputs.
The R-MCM doe account for both dry and wet deposition of atmospheric mercury
to the water body. Both depositions are modeled as fluxes to the system (mass/area/time,
9
in R-MCM, this is in ug/m /yr). Both deposition fluxes consist primarily of Hgll. The
R-MCM calculates the wet deposition input by multiplying the annual precipitation rate
by the annual average concentration of Hgll in the precipitation. The mercury flux of
94
-------
Evaluating R-MCMfor 91 VT/NH Lakes
each is multiplied by the lake area and then added together to produce the total annual
atmospheric mercury loading to the water body (mass/time).
The mercury loadings to the watershed system are calculated by using the total
annual atmospheric mercury flux calculated for the water body multiplied by the
watershed (or sub-watershed) area. The subsequent mercury loading to the water body
from the watershed system is modeled as the summation of loadings from the local
upland, local wetland, and upstream sources. The local upland term accounts for water
that runs off the land surface and enters the lake directly without passing through streams,
rivers, lakes, or wetlands. The local wetland is that portion of the catchment deemed a
wetland that is directly adjacent to the lake being modeled. The total annual atmospheric
mercury deposition flux acts upon the local wetland area and then is assumed to
immediately enter the lake. The upstream sources represent the rest of the catchment,
where atmospheric mercury deposits directly into upstream lakes, onto uplands that run
off into upstream lakes and streams, and upstream wetlands that flow into upstream lakes
and streams; all of these then flow into the lake being modeled. The total mercury
loading to the lake being modeled from all of these other sources is calculated essentially
as the sum of the products of the modeled water body's total annual atmospheric mercury
loading rate times the ratio of each sub-catchment area (upstream lake, upland, upstream
wetland) to the modeled lake surface area, a fraction that accounts for the amount of
mercury that passes through each sub-catchment unit (upstream lake, upland, upstream
wetland). The R-MCM variables describing that fraction of atmospherically loaded
mercury that passes through each sub-catchment area (i.e., terrestrial classification) are
designated by a prefix and a suffix, and are referred-to here as the "outflow parameters."
95
-------
Evaluating R-MCMfor 91 VT/NH Lakes
The prefixes, "Rl" and "R2," correspond to whether the mercury species of interest is
methylmercury (Rl) or Hgll (R2). The suffixes "Up" and "Wet" refer to whether the
mercury is passing through an upland (Up) or a wetland (Wet), respectively. The uplands
includes the area of the watershed that contributes directly through run-off to the lake, but
does not include upstream lakes and streams that flow directly into the lake. The variable
RlUp refers to the fraction of atmospheric methylmercury that passes through the upland.
R-MCM assigns default values to these outflow parameters depending on the hydrology
of the lake. The outflow parameter default values are assigned as follows.
Drainage Lake
RlUp
RlWet
R2Up
R2Wet
0.1
3
0.1
0.35
Ground Water Fed
Seepage Lake
0.05
0.3
0.05
0.1
Mounded
Seepage Lake
0
0
0
0
This table includes the three different types of lake hydrology that R-MCM incorporates.
All lake types receive input from precipitation. A "Drainage Lake" receives water has
surface water inflows and outflows. A "Ground Water Fed Seepage Lake" receives water
from ground water sources only. A "Mounded Seepage Lake" has all ground water
flowing out of the lake. In the VT/NH study, almost all lakes were drainage lakes.
An interesting facet of this approach is that the methylmercury and the divalent
mercury are modeled separately. That is, transformation between the two in the
watershed is not directly taken into account. Rather, the watershed methylation of Hgll
into MeHg is taken into account by modeling RlWet as greater than unity. Specifically,
for RlWet, the value of "3" for drainage lakes is used to account for the methylation
known to occur in wetlands (see, for example, Krabbenhoft et al. (1999) or Kelly et al.
96
-------
Evaluating R-MCMfor 91 VT/NH Lakes
(1995)). Because the methylmercury loading from the atmosphere is quite low, changes
in all Rl outflow parameters may have to be dramatic to adequately capture real-world
increases in methylmercury loading to a lake from its watershed. If the transformation of
Hgll into MeHg in the watershed is an important process for MeHg loading to the lake,
then the current R-MCM configuration may not adequately capture this process.
Although, the mercury chemistry and transformation processes that the R-MCM currently
incorporates are rigorous, unless the methylmercury loading term is accurately captured,
then the R-MCM will not be able to capture the trends in mercury species concentrations
present in the receiving water system.
In this section (Section 6.4), we investigate the effect of adjusting the values for
the various watershed element outflow parameters. Essentially, a sensitivity analysis was
performed to investigate how increases in the outflows of mercury from the various
watershed elements in the receiving lake would impact the R-MCM results. In a "real-
world sense," this could be explained as more properly accounting for possibly greater
methylation of Hgll in the VT and NH watersheds and resulting in increased loading of
MeHg and Hgll into the lakes.
6.4.1 Investigation of the Watershed Element Mercury Outflow Parameters
The mercury species loadings from the various watershed elements to the
receiving lakes are dependent on the area of the catchment or element, the atmospheric
mercury deposition to the catchment, and the fraction of the mercury deposited that
leaves the catchment or element and enters the receiving water body being modeled.
Assuming the atmospheric mercury depositions and sizes of the catchments or elements
are fixed parameters, then the only parameters that can be adjusted are the fractions of the
97
-------
Evaluating R-MCMfor 91 VT/NH Lakes
deposited mercury that leaves each catchment or element. These parameters (fractions)
are labeled as outflow parameters, as described and defined in Section 6.4. Because the
wetland fractional areas of the of VT and NH catchments are relatively small, the
analysis in this section focused on adjusting the upland parameters, RlUp and R2Up.
The concentration of Hgll in precipitation is approximately fifty times greater than the
amount of MeHg in precipitation. Furthermore, the amount of MeHg in dry deposition is
almost one hundred times less than the amount of Hgll. Hypothetically, then, if all the
Hgll that deposited onto the catchment were transformed into MeHg, then the amount of
MeHg loading to the lake could increase between fifty and one hundred times.
Therefore, as discussed previously in Section 6.4, the fraction of RlUp could be greater
than unity, indeed, as suggested here, it could be much greater than unity. Because Hgll
is the predominant deposition source, the fraction outflows of Hgll would most likely
always be less than unity. However, for sake of illustration, the test cases fractions herein
were allowed to run in excess of unity for both RlUp and R2Up.
To perform the sensitivity analysis, ranges of RlUp and R2Up values were
chosen. R-MCM currently assigns default values of 0.1 for both RlUp and R2Up.
Because we were interested in how increases in loading from the watershed would impact
the predicted lake mercury species concentration results, this value was chosen as the
base case and the minimum value. The values chosen for the sensitivity runs were RlUp:
0.1, 1, 2, and 5 and R2Up: 0.1, 1, and 2. Using the R-MCM input variables for Tier 5,
every combination of the RlUp and R2Up sensitivity run values was modeled to predict
the epilimnion MeHg and HgT, hypolimnion MeHg and HgT, sediment HgT, and Fish
98
-------
Evaluating R-MCMfor 91 VT/NH Lakes
HgT lake concentrations. The predicted results for the VT and NH Lakes for these runs
are presented in the appendix in Tables A-22 through A-27.
The predicted results for these runs are plotted against the observed values for all
types of observations in Figures 6.28 through 6.33. The results are plotted in a grid
fashion so that the RlUp values are constant on the vertical, and the R2Up values are
constant along the horizontal. The y=x lines re plotted as dashed lines to graphically
illustrate the model effectiveness for each combination of RlUp and R2Up values.
As would be expected, increasing R2Up had the greatest impact on increasing the
mean predicted value of all mercury species concentrations. This is because the
atmospheric deposition of Hgll is much greater than MeHg and therefore a small increase
in R2Up produces a large increase in total mercury load to the lake. For example,
relative to the epilimnion methylmercury concentrations, holding RlUp = 0.1 constant,
while increasing R2Up (the top row of Figure 6.28), there was a great increase in the
scatter and range of the predicted values. (This scenario could account, crudely, for
additional capture and storage of atmospheric mercury by the forest canopy and
subsequent delivery to the forest floor and lake.) By holding R2Up=0.1 constant and
increasing RlUp, the centroid of the data points moved upward and seemed to improve
the modeling. An interesting aspect of this work was that all lakes did not increase
equally. For some of the lakes, the data were seen to "jump up" quickly, while for other
lakes the data remained in the cluster. For example, in the bottom left plot of Figure 6.28
(RlUp = 5, R2Up = 0.1), there is one lake in excess of 2.5 ng/L that was not so
dramatically greater than the other lakes in the top left plot (RlUp = 0.1, R2Up = 0.1). A
possible physical explanation of this phenomenon is that the size of the watershed is
99
-------
Evaluating R-MCMfor 91 VT/NH Lakes
directly correlated to the amount of increase in loading as the RUp values are increased.
Similarly, the size of the lake in relation to that of the watershed, as well as the flushing
rate of the lake, will also have an impact on the amount of increase in predicted mercury
species concentrations. This suggests that the physical, chemical, and biological aspects
of the watershed and receiving water body all can impact the susceptibility of a lake to
increases in mercury loading, while at the same time allowing others to be better buffered
against increased loads. For the epilimnion total mercury concentrations, increases in
R2Up at constant RlUp resulted in dramatic increases in the centroid of the data, but not
an improvement in model performance.
Hypolimnion mercury species concentrations have a large bias in mercury
concentrations, and that the R-MCM is having great difficulty predicting the large
observed values (see Chapters 4 and 5). For the hypolimnion, increasing R2Up seemed
to produce more improvement in model predictability for the total mercury than for the
methylmercury concentrations.
Figure 6.32 presents the fish tissue concentrations. The fish tissue observations
and predicted values have a large amount of scatter. From these runs, it was difficult to
discern if there was any improvement by adjusting either of the watershed outflow
parameters (i.e., RlUp or R2Up). For the sediment concentrations, the large amount of
under-prediction in the sediment total mercury concentrations was corrected by
increasing R2Up at constant RlUp.
To get a better quantitative understanding of what possible improvements could
be achieved by adjusting the RlUp and R2Up values, an interpolation of the data points
was made. Using a cubic linear interpolation algorithm (Matlab v. 6, Mathworks, Inc.),
100
-------
Evaluating R-MCMfor 91 VT/NH Lakes
the error sums of squares and associated standard deviation were plotted with a resolution
of 100 points. From this interpolation, the minimum value for each was determined, and
the RlUp and R2Up values associated with these minima were extracted. Figures 6.34
and 6.35 illustrate the shapes of these surfaces. From these surfaces, the sensitivity of the
results to R2Up is obvious. Increasing R2Up for epilimnion MeHg and HgT,
hypolimnion MeHg, fish tissue HgT, and sediment HgT concentrations all resulted in a
general increase in the error sums of squares and associated standard deviations. There
were some instances where the sum of squares first decreased, but in most instances there
was a large increase as R2Up approached 2. The most noticeable difference was seen for
the hypolimnion HgT concentration where there appeared to be more improvement with
and sensitivity to R2Up rather than RlUp.
The surface minima values and associated RlUp and R2Up values are presented
in Table 6.7. To improve the results, outlying lakes were removed from the epilimnion
MeHg and HgT and the hypolimnion MeHg concentration data sets (see Table 6.7 for
details on removed data). The refined surface minima and associated RlUp and R2Up
values determined after these lakes were removed are also included in Table 6.7. A note
on the removal of these outliers is necessary at this point. It is believed that the removed
lakes may provide useful information on the reasons why the R-MCM is having difficulty
and why these lakes may be greatly over- or under-predicted. However, their great
distance from the y=x line resulted in too great an influence on the calculated standard
deviations and error sums of squares. Therefore, they were removed to allow the
investigation to be more evenly influenced by the other lakes in the study.
101
-------
Evaluating R-MCMfor 91 VT/NH Lakes
From the shape of the surfaces (Figures 6.34 and 6.35) and the estimated minima
in Table 6.7, the results suggest that an increased mercury loading from the watershed
may improve the overall modeling effort. The first thing that is seen is that for
Epilimnion MeHg and HgT, hypolimnion MeHg, and fish tissue HgT concentrations, the
R2Up value is best left at the default value of around 0.1. However, the RlUp value
appears to be about a factor of 45 too low. This suggests the possibility that watershed
loading is having an important influence on the overall R-MCM performance relative to
predicting lake mercury species concentrations for our 91 lakes. Additionally, this may
point towards the specific issue why methylmercury is not being adequately modeled by
the R-MCM at all. Lastly, the removal of outliers from our observed data sets resulted in
even larger values for the optimum RlUp for the epilimnion predicted mercury species
concentrations, although there was a slightly smaller optimum value produced for the
hypolimnion methylmercury concentration prediction. The removal of these outliers
improved the modeling and decreased the estimated standard deviation of the results.
The results of this analysis were not intended to suggest what the best default
values for RlUp and R2Up should be, but rather to investigate whether changes in
mercury outflow parameters from the various watershed elements could improve R-
MCM predictions in the receiving lake. Our results point towards the possibility that
methylmercury production (via Hgll transformation) within the watershed may be an
important process that should be better modeled. We therefore believe that the modeling
of watershed element mercury loadings to a water body (lake) needs to be improved
before more accurate predictions via R-MCM can be made for the various mercury
species concentrations in the water body.
102
-------
Evaluating R-MCMfor 91 VT/NH Lakes
7 CONCLUSIONS
In this report we used the R-MCM to predict mercury species concentrations in
the epilimnion, hypolimnion, sediment, and fish tissue for a large array of lakes in the
New England region of the United States. This is the first time that the R-MCM has been
applied to such a large data set of lakes. This is also the first time that the R-MCM has
been applied to lakes in New England. Therefore, this modeling effort, in effect,
evaluated how robust the R-MCM is when applied to lakes in an environment and
watershed ecosystems similar to those found in Vermont and New Hampshire. In
particular, this extends the boundaries of mercury cycling modeling by evaluating how
well the R-MCM is able to model mercury species concentrations for lakes in
mountainous watersheds where the lakes range from acidic to alkaline and from
oligotrophic to dystrophic. The R-MCM was originally developed for oligotrophic,
acidic, seepage lakes in the mid-western state of Wisconsin. By taking the original R-
MCM as is and applying it to a different region, our evaluation, therefore, probed into the
processes governing mercury cycling in a different watershed system and allowed for
mechanistic inference on the governing processes of mercury cycling science.
From our review of the visual inspections and statistical evaluations of all 91
modeled lakes as a whole and the lakes grouped via their characteristics, it is clear that
the modeling of mercury in watersheds is a complicated problem. There were clear
successes and failures that the R-MCM encountered while modeling these 91 lakes. One
conclusion reached immediately was that the R-MCM has difficulty translating directly
to a new environment. No calibration methods were undertaken in our evaluation
because that was the not the specific goal of this study. The goal was to evaluate the
103
-------
Evaluating R-MCMfor 91 VT/NH Lakes
level of success that could be gained by applying the original R-MCM directly to a new
environment, specifically the New England region of the United States.
First, the R-MCM was applied using the simplest and front-most user interface.
Next, the question was asked that if the model does not succeed using the simpler user
interface, then what types of parameters are needed or need to be adjusted to improve
model accuracy. Similarly, we asked how much accuracy is gained by providing certain
additional information for each lake and watershed. Indeed, can the R-MCM reach a
point with parameter refinement that would result in a strong model? Does the R-MCM
require region-specific calibration or lake-specific calibration? These are the questions
that our research set out to address. With this in mind, it is the ultimate goal to develop a
process-based, watershed-lake model that will be robust enough to be applicable to
different regions. Therefore, this study is viewed as the first step in a very challenging
journey.
In this report, we have provided a rigorous analysis of the model using both visual
inspection and a suite of statistical methods. With such a large number of lakes and such
a large number of observed variables, it becomes challenging to sort through all the data
and model results and retrieve a clear picture. We therefore have tried to separate the
pieces so that they can be synthesized into the clearest results possible.
From the first general inspection, it became clear that using the default case
resulted in a wide scatter of predicted concentrations. This scatter was reduced in the
Tier case analysis. In fact, the amount of scatter was greatly reduced simply by overriding
the default inputs and using the measured inputs as done in the Tier 1 case (i.e., for pH,
lake size, hypolimnion surface area, residence time and water column DOC). This seems
104
-------
Evaluating R-MCMfor 91 VT/NH Lakes
to suggest that the model is sensitive to these input variables, and therefore it is
worthwhile to invest time and effort in making these measurements. However, despite an
apparent decrease in the predictive scatter, a closer inspection revealed that the predicted
values were merely generally decreased across the board. Regardless of the observed
value, most predicted concentrations decreased. This suggests a decrease in the range of
R-MCM predictability when more refined values are used for the inputs.
Because of the decrease in scatter from the default case to the Tier 1 scenario, it
was originally thought that it might be worthwhile to investigate if the shift were caused
by one specific input parameter, by a combination of the input parameters, or only by all
in consort. However, because there was still a significant amount of error and bias
associated with the modeled results, these questions were not investigated. Obviously,
investigating different combinations and individual parameters for the six measured
mercury species concentrations predictions would be time intensive, and the value of that
work is not completely clear. We therefore focused on the progression of model
improvement achieved in the Tiers as outlined in Table 2.3 and described in the
associated text.
Visual inspection of the Tier results was combined with the statistical evaluation
using error sum of squares. The visual inspection indicated a decrease in prediction
scatter in going from the default case to the Tier 1 scenario. However, calculating the
error sum of squares revealed that this move did not provide much, if any, reduction in
error. These general evaluation methods suggested that as a whole, the R-MCM
performance did not improve across the Tiers. This caused us to not focus on individual
improvements from Tier to Tier. It seemed evident that the R-MCM was having
105
-------
Evaluating R-MCMfor 91 VT/NH Lakes
difficulties being applied to the VT and NH region, and, therefore, it became important to
investigate where the strengths and weaknesses of the R-MCM lay so that each could be
further investigated and understood. This led us to an evaluation of the R-MCM
performance as a function of lake type and characteristics.
Evaluating the R-MCM by lake type helped to sort out some of the model
successes and inadequacies. A visual analysis, combined with the t-test and chi-square
test showed that the R-MCM did relatively well modeling alkaline lakes. This was a
curious result because the model was designed with acidic lakes in mind. However,
results have suggested that there may be a correlation between mercury species
concentration and lake acidity. Therefore, the success of R-MCM in modeling alkaline
lakes may correspond to simpler modeling processes, where the mass balance approach
that R-MCM uses is successful. The greater difficulty in modeling acidic and
circumneutral lakes points to processes that may not be adequately incorporated into the
R-MCM.
The statistical analyses for lakes separated by characteristics provided further
insight into the model's behavior. Alkaline lakes were the only type that passed the chi-
square test, but the t-test showed that alkaline lakes had a bias towards under-prediction.
Without the additional information of the t-test, this bias would not have been detected.
Acidic and circumneutral lakes did not pass the chi-square test, but had mean error
residuals of zero.
Alkaline lakes were also found to have the smallest maximum errors. The other
statistical techniques pointed out the general tendency of the R-MCM to under-predict for
these lakes. Unfortunately, the various statistical techniques did not separate out enough
106
-------
Evaluating R-MCMfor 91 VT/NH Lakes
information to proclaim that the R-MCM could adequately model any specific type of
lake. Going into the analysis, it was hoped that at least a certain type of lake or types of
lakes could be certified as being adequately modeled by the R-MCM, but this was not
found to be the case.
Due to the tendency of the R-MCM to under-predict the lake mercury species
concentrations and the very large maximum errors noted (and the poor results of the
model efficiency test, namely that the mean observed value was found to be a better
predictor than the model for most lake variables), an investigation was made on the R-
MCM sensitivity to the various lake mercury loss processes. Specific investigations of
the photoreduction rate and particle settling velocity did not elucidate the under-
prediction problems with the R-MCM. Specifically, R-MCM performance was not found
to improve by reducing these loss rates (the predicted concentrations did not increase
enough to effectively improve the model).
Lastly, the observed data were investigated for trends as a function of lake
characteristics and compared to the trends that could be seen in the R-MCM predictions.
The predicted data generally followed the attributes that would be expected for specific
lake characteristics, while the observed data had too much scatter to discuss any possible
trends. This was discouraging because it points to the fact that mercury modeling in
these systems is even more complicated than expected. So many factors are at play in
these systems that it appears more research and modeling are required to capture all the
relevant fate processes. If this is not done, then large errors will need to be accepted with
the R-MCM - or that require an appreciable amount of calibration for each system
investigated will be required.
107
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Percent total mercury as methylmercury was isolated as an indicator of the
methylation rate of a given system. Because it is methylmercury that is being transferred
through the food chain and accumulating in fish and wildlife, methylmercury
concentrations in the various watershed and lake compartments are of particular interest.
The extent of methylation can be investigated by looking at the percent total that is
methylmercury. The present methylmercury was calculated for all observed and
predicted data in the epilimnion, hypolimnion and sediment. From these calculations, it
was found that the observed epilimnion percent methylmercury values spanned a much
greater range than did the predicted values. The observed maximum percent
methylmercury clearly decreased from epilimnion to hypolimnion to sediment, while the
predicted maximums of percent methylmercury were not much different for these
different media. The predicted mean percent methylmercury in the epilimnion was much
smaller than the observed, while the observed was smaller than the predicted in the
hypolimnion and in the sediments. It is generally believed that methylation of mercury
occurs in the sediments, so it was strange to see a gradient in percent methylmercury in
the water column that was in an opposite direction from the assumed source. This seems
to imply that the source of methylmercury in these lake systems may be coming from
more watershed related sources. This is not to discount the methylation processes
occurring in the sediments. Clearly, the percent methylmercury in the sediments is
important, but the dominant source may be watershed loading.
Further investigation into the impact of mercury watershed loading processes was
performed by evaluating the various watershed loading outflow parameters. Through
sensitivity analysis of these parameters it was found that the R-MCM error in the lake
108
-------
Evaluating R-MCMfor 91 VT/NH Lakes
modeling could be improved by greatly increasing the amount of methylmercury loading
from the watershed into the lake. Research results have suggested that litterfall and
transformation of mercury in watersheds may be more important than previously thought,
and our analysis lends support to this theory.
In conclusion, this report evaluates a currently accepted model (the R-MCM) for
predicting various mercury species concentrations in a lake water column, sediments and
fish populations within a regional set of lakes. Our analysis shows that the R-MCM
model has difficulty in prediction without some level of calibration. Therefore, the R-
MCM cannot be directly applied to a new region of lakes without calibration. An
investigation of whether certain types of lakes would be able to be adequately modeled
using the R-MCM as is was found to be fruitless. It seems for the VT and NH lakes that
the watershed mercury loading and influence may be a critical factor. Because the R-
MCM does not have a rigorous mercury watershed loading module, it is believed that this
is an important area of research that needs to be developed. It is possible that
improvements in watershed mercury modeling could interfaced with the water body
(lake) modeling capability of R-MCM to create a more robust model for successfully
modeling mercury in watersheds and water bodies without site-specific calibration.
109
-------
Evaluating R-MCMfor 91 VT/NH Lakes
8 REFERENCES
Box, G.E.P., W.G. Hunter, J.S. Hunter. 1978. Statistics for Experiments: An Introduction
to Design, Data Analysis, and Model Building. John Wiley & Sons. New York.
Kamman, N., B. Estabrook. 1998a. Quality Assurance Project Plan: Assessment of
Mercury in Hypolimnetic Lake-bed Sediments of Vermont and New Hampshire.
USEPA Region 1 - New England Regional Environmental Monitoring and
Assessment Program. July 21.
Kamman, N., B. Estabrook. 1998b. Quality Assurance Project Plan Addendum:
Assessment of Mercury in Hypolimnetic Lake-bed Sediments of Vermont and
New Hampshire. USEPA Region 1 - New England Regional Environmental
Monitority and Assessment Program.
Kamman, N., C.T. Driscoll, B. Estabrook, D.C. Evers, E.K. Miller. 2004.
Biogeochemistry of Mercury in Vermont and New Hampshire Lakes: An
Assessment of Mercury in Waters, Sediments, and Biota of Vermont and New
Hampshire Lakes. Comprehensive Final Project Report.
Kelly, C.A., J.W.M. Rudd, V.L. St. Louis, A. Heyes. 1995. Is Total Mercury
Concentration a Good Predictor of Methyl Mercury Concentration in Aquatic
Systems? Water, Air, and Soil Pollution. 80: 715-724.
Krabbenhoft, D.P., J.G. Wiener, W.G. Brumbuahg, M.L. Olson, J.F. DeWild, T.J. Sabin.
1999. A National Pilot Study of Mercury Contamination of Aquatic Ecosystems
Along Multiple Gradients. U.S. Geological Survey Toxic Substances Hydrology
Program - Proceedings of the technical meeting, Charleston, S.C., March 8-12,
1999. USGS Water Resources Investigations Report 99-4018B, Vol 2.
110
-------
Evaluating R-MCMfor 91 VT/NH Lakes
League, K. and R. E. Green. 1991. Statistical and Graphical Methods for Evaluating
Solute Transport Models: Overview and Application. Journal of Contaminant
Hydrology. 7:51-73.
Mayer, D.G. andD.G. Butler. 1993. Statistical Validation. Ecological Modelling. 68:21-
32.
Miller, E.K. 2002. Estimation and Mapping of Wet and Dry Mercury Deposition Across
the VT-NH Region. Final Report. Submitted to Neil Kamman, VTDEC.
Miller, E.K. 2003. personal communication.
Murphy, B.R. andD.W. Willis. 1996. Fisheries Techniques. Second Edition. American
Fisheries Society. Bethesda, MD.
Neter, J. and W. Wasserman. 1974. Applied Linear Statistical Models: Regression,
Analysis of Variance, and Experimental Designs. Richard D. Irwin. Homewood,
Illinois.
Tetra Tech, Inc. 1996. Regional Mercury Cycling Model: A Model for Mercury Cycling
in Lakes - R-MCM Version 1.0 Beta. Draft User's Guide and Technical
Reference. Prepared for the Electric Power Research Institute. December 1996.
Tetra Tech, Inc. 2002. Mercury in Fish in Kejimkukik Park, Nova Scotia: Application of
the Regional Mercury Cycling Model to 24 Lakes. Prepared for the Canadian
Wildlife Service. Prepared by Reed Harris, David Hutchinson, and John Radde.
USEPA, 1997. Mercury Report to Congress. EPA 425-R-97-003. Washington, D.C.
Ill
-------
Evaluating R-MCMfor 91 VT/NH Lakes
TABLES
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 2.1. Lake Categories and Characteristics
Category
Acidity
Lake
Stratification
Lake Size (Area)
Hydrology
Trophic State
Characteristic
Acidic
Circumneutral
Stratified
Small (10 ha)
Drainage Lake
Oligotrophic
Alkaline
Well-Mixed
Medium (1 km2)
Groundwater Fed
Seepage Lake
Mesotrophic
Large (25 km2)
Mounded Seepage
Eutrophic Dystrophic
Notes:
Italics: No VT NH lakes of this type in this study.
Summer anoxia: included as a category in the R-MCM default set up, but the rates and processes associated
with summer anoxia are currently unavailable in R-MCM. The model therefore defaults to no
anoxia.
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 2.2. Lake Names and Input Characteristics for Default Set-Up.
LAKE NAME
ADDER POND
ARMINGTON LAKE
BAKER (BARTON)
BAKER POND- UPPER
BEARCAMP POND
BRANCH
BRUCE
CAWLEY POND
CHASE POND
CHILDS BOG
CHITTENDEN
CLUB POND
CRANBERRY MEADOW
CURTIS
DENNIS
DUNMORE
DUTCHMAN POND
EASTMAN POND
ECHO (CHARTN)
ECHO (HUBDTN)
ELM BROOK POOL
FERN
FISH POND
FREESES POND- UPPER
GILES POND
GREAT HOSMER
GREENWOOD POND
HALL POND- UPPER
HARDWICK
HARDWOOD
HIGH (SUDBRY)
HILDRETH DAM POND
HORN POND
HORTONIA
HOWE RESERVOIR
ISLAND POND
IVANHOE- LAKE
JACKSONVILLE
JENNESS POND
KENT
LARY POND
LEFFERTS
LILY POND
LITTLE AVERILL
LONG (WESTMR)
Acidity
Acidic
Circumneutral
Alkaline
Circumneutral
Circumneutral
Acidic
Circumneutral
Circumneutral
Circumneutral
Circumneutral
Circumneutral
Acidic
Alkaline
Alkaline
Circumneutral
Circumneutral
Acidic
Circumneutral
Alkaline
Alkaline
Circumneutral
Alkaline
Alkaline
Circumneutral
Circumneutral
Alkaline
Alkaline
Acidic
Alkaline
Circumneutral
Alkaline
Circumneutral
Circumneutral
Alkaline
Acidic
Acidic
Circumneutral
Circumneutral
Circumneutral
Alkaline
Circumneutral
Circumneutral
Acidic
Alkaline
Alkaline
Stratification
Stratified
Stratified
Stratified
Well-Mixed
Stratified
Stratified
Well-Mixed
Well-Mixed
Well-Mixed
Well-Mixed
Stratified
Well-Mixed
Stratified
Stratified
Well-Mixed
Stratified
Well-Mixed
Stratified
Stratified
Stratified
Well-Mixed
Stratified
Well-Mixed
Well-Mixed
Well-Mixed
Stratified
Well-Mixed
Stratified
Well-Mixed
Well-Mixed
Stratified
Stratified
Stratified
Stratified
Well-Mixed
Stratified
Well-Mixed
Well-Mixed
Stratified
Well-Mixed
Well-Mixed
Well-Mixed
Well-Mixed
Stratified
Stratified
Lake Size
Small
Medium
Small
Medium
Medium
Small
Small
Small
Small
Small
Medium
Small
Small
Small
Medium
Medium
Small
Medium
Medium
Small
Medium
Small
Small
Small
Small
Medium
Small
Small
Medium
Small
Small
Small
Medium
Medium
Medium
Medium
Medium
Small
Medium
Small
Small
Small
Small
Medium
Small
Hydrology
Eutrophic
Oligotrophic
Eutrophic
Mesotrophic
Dystrophic
Dystrophic
Dystrophic
Mesotrophic
Dystrophic
Mesotrophic
Mesotrophic
Dystrophic
Mesotrophic
Eutrophic
Dystrophic
Oligotrophic
Oligotrophic
Eutrophic
Oligotrophic
Mesotrophic
Mesotrophic
Mesotrophic
Dystrophic
Eutrophic
Oligotrophic
Eutrophic
Eutrophic
Dystrophic
Eutrophic
Dystrophic
Eutrophic
Mesotrophic
Oligotrophic
Mesotrophic
Dystrophic
Dystrophic
Oligotrophic
Dystrophic
Oligotrophic
Mesotrophic
Eutrophic
Mesotrophic
Eutrophic
Oligotrophic
Oligotrophic
Trophic Status
Acidic
Circumneutral
Alkaline
Circumneutral
Circumneutral
Acidic
Circumneutral
Circumneutral
Circumneutral
Circumneutral
Circumneutral
Acidic
Alkaline
Alkaline
Circumneutral
Circumneutral
Acidic
Circumneutral
Alkaline
Alkaline
Circumneutral
Alkaline
Alkaline
Circumneutral
Circumneutral
Alkaline
Alkaline
Acidic
Alkaline
Circumneutral
Alkaline
Circumneutral
Circumneutral
Alkaline
Acidic
Acidic
Circumneutral
Circumneutral
Circumneutral
Alkaline
Circumneutral
Circumneutral
Acidic
Alkaline
Alkaline
-------
Evaluating R-MCMfor 91 VT/NH Lakes
LAKE NAME
LOON LAKE
LOVELL LAKE- STN 1
LYFORD
MANSFIELD
MCCONNELL
MILLSFIELD POND
MILTON
MINARDS
MITCHELL
MOOSE POND
MOUNTAIN LAKE- UPPER
NEWARK
NORTH (BRKFLD)
NOTCH
NOYES
PARAN
PARKER
PAUGUS BAY- STN 1
PAWTUCKAWAY LAKE
PEMIGEWASSET LAKE
PERCH (BENSON)
PLEASANT VALLEY
POUT POND
POWWOW POND
ROBB RESERVOIR
ROUND POND
SAB IN
SHAWS POND
SILVER LAKE
SOMERSET
SOUTH AMERICA
SPRUCE POND
STRATTON
SUNCOOK POND- UPPER
SUNRISE LAKE
SUNSET (BRKFLD)
TRIO PONDS- ONE AND
TWO
TUTTLE (HARDWK)
UNNAMED POND
WALKER POND
WILLEY POND- BIG
WILLEY POND- LITTLE
WILLOUGHBY
WILSON POND
WOLCOTT
ZEPHYR LAKE
Acidity
Acidic
Circumneutral
Alkaline
Acidic
Circumneutral
Circumneutral
Circumneutral
Circumneutral
Alkaline
Circumneutral
Circumneutral
Alkaline
Alkaline
Circumneutral
Circumneutral
Alkaline
Alkaline
Circumneutral
Circumneutral
Circumneutral
Alkaline
Alkaline
Circumneutral
Circumneutral
Acidic
Circumneutral
Alkaline
Alkaline
Acidic
Acidic
Circumneutral
Acidic
Circumneutral
Circumneutral
Circumneutral
Alkaline
Acidic
Alkaline
Acidic
Circumneutral
Acidic
Alkaline
Alkaline
Circumneutral
Circumneutral
Circumneutral
Stratification
Stratified
Stratified
Stratified
Stratified
Stratified
Well-Mixed
Well-Mixed
Stratified
Well-Mixed
Well-Mixed
Well-Mixed
Stratified
Stratified
Stratified
Well-Mixed
Well-Mixed
Stratified
Stratified
Stratified
Stratified
Stratified
Well-Mixed
Well-Mixed
Stratified
Well-Mixed
Stratified
Stratified
Well-Mixed
Stratified
Stratified
Well-Mixed
Stratified
Well-Mixed
Stratified
Well-Mixed
Stratified
Stratified
Well-Mixed
Well-Mixed
Stratified
Well-Mixed
Well-Mixed
Stratified
Well-Mixed
Stratified
Stratified
Lake Size
Small
Medium
Small
Small
Small
Medium
Small
Small
Small
Small
Small
Medium
Small
Small
Small
Small
Medium
Medium
Medium
Medium
Small
Small
Small
Medium
Medium
Small
Medium
Small
Medium
Medium
Small
Small
Small
Medium
Medium
Small
Small
Small
Small
Medium
Small
Small
Medium
Small
Small
Small
Hydrology
Oligotrophic
Mesotrophic
Mesotrophic
Dystrophic
Dystrophic
Dystrophic
Mesotrophic
Mesotrophic
Oligotrophic
Dystrophic
Dystrophic
Oligotrophic
Mesotrophic
Dystrophic
Dystrophic
Mesotrophic
Mesotrophic
Mesotrophic
Dystrophic
Dystrophic
Mesotrophic
Mesotrophic
Eutrophic
Dystrophic
Dystrophic
Oligotrophic
Mesotrophic
Oligotrophic
Oligotrophic
Mesotrophic
Dystrophic
Eutrophic
Mesotrophic
Dystrophic
Mesotrophic
Mesotrophic
Dystrophic
Eutrophic
Dystrophic
Dystrophic
Oligotrophic
Oligotrophic
Oligotrophic
Mesotrophic
Dystrophic
Mesotrophic
Trophic Status
Acidic
Circumneutral
Alkaline
Acidic
Circumneutral
Circumneutral
Circumneutral
Circumneutral
Alkaline
Circumneutral
Circumneutral
Alkaline
Alkaline
Circumneutral
Circumneutral
Alkaline
Alkaline
Circumneutral
Circumneutral
Circumneutral
Alkaline
Alkaline
Circumneutral
Circumneutral
Acidic
Circumneutral
Alkaline
Alkaline
Acidic
Acidic
Circumneutral
Acidic
Circumneutral
Circumneutral
Circumneutral
Alkaline
Acidic
Alkaline
Acidic
Circumneutral
Acidic
Alkaline
Alkaline
Circumneutral
Circumneutral
Circumneutral
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 2.3. Tiers 1 through 5 and their Associated Default Input Values and Lake
Characteristics
Tier
1 1
0>
H
CO
"O
CS
CJ
s_
0>
*4
H
^i
s_
v
*4
H
IT)
S_
V
*4
H
Parameter
pH
Lake Size
(Surface Area)
Hypolimnion Surface Area
Residence Time
Water Column DOC
Hgll concentration in
precipitation
Mean annual dry deposition
rate of Hgll
Precipitation Rate
Total Catchment Ratio
Wetland Fraction
Lake Cover Fraction
Epilimnion Depth
Hypolimnion Depth
Lake Characteristic
Acidic
Circumneutral
Alkaline
Small
Medium
Small
Medium
Small
Medium
Oligotrophic
Mesotrophic
Eutrophic
Dystrophic
n/a
n/a
n/a
n/a
n/a
n/a
Small lake
Medium Lake
Small Lake
Medium Lake
Default Value
5.3
6.5
8
10 ha
1km2
25,000 m2
50,000 m2
6 months
2 years
3mg/L
7mg/L
15 mg/L
20 mg/L
10 ug/m
3.5 ug HgII/m2/yr
0.8 m/yr
10
0.15
0.05
3m
8m
3 m
5m
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 2.4. Parameter Updates for Tier 1.
Lake Name
ADDER POND
ARMINGTON LAKE
BAKER (BARTON)
BAKER POND- UPPER
BEARCAMP POND
BRANCH
BRUCE
CAWLEY POND
CHASE POND
CHILDS BOG
CHITTENDEN
CLUB POND
CRANBERRY MEADOW
CURTIS
DENNIS
DUNMORE
DUTCHMAN POND
EASTMAN POND
ECHO (CHARTN)
ECHO (HUBDTN)
ELM BROOK POOL
FERN
FISH POND
FREESES POND- UPPER
GILES POND
GREAT HOSMER
GREENWOOD POND
HALL POND- UPPER
HARD WICK
HARDWOOD
HIGH (SUDBRY)
HILDRETH DAM POND
HORN POND
HORTONIA
HOWE RESERVOIR
ISLAND POND
IVANHOE- LAKE
JACKSONVILLE
JENNESS POND
KENT
LARY POND
LEFFERTS
LILY POND
LITTLE AVERILL
LONG (WESTMR)
LOON LAKE
LOVELL LAKE- STN 1
PH
5.75
6.26
7.32
6.7
6.61
4.6
7.05
6.53
6.17
6.09
6.98
5.84
7.47
7.42
6.82
7.17
5.86
6.93
7.44
7.97
6.37
7.52
7.4
5.9
6.77
7.52
7.62
5.8
7.65
6.13
7.46
6.61
6.33
7.8
5.83
5.6
6.04
6.11
6.11
7.64
6.4
6.37
5.87
7.32
7.54
5.7
6.9
Lake
Surface
Area
[ha]
10.52
57.55
20.64
75.07
67.58
13.76
10.93
10.00
15.78
42.65
284.10
13.15
11.33
29.14
74.87
398.63
11.29
135.57
222.59
21.85
86.52
27.92
8.09
11.09
12.95
56.66
20.03
9.19
58.68
17.81
8.09
15.18
80.13
193.85
104.33
63.94
50.18
8.09
94.09
40.07
12.54
22.26
33.79
188.99
36.42
45.28
217.72
Hypolimnion
Surface Area
[ha]
3.51
19.18
6.88
25.02
22.53
4.59
3.64
3.33
5.26
14.22
94.70
4.38
3.78
9.71
24.96
132.88
3.76
45.19
74.20
7.28
28.84
9.31
2.70
3.70
4.32
18.89
6.68
3.06
19.56
5.94
2.70
5.06
26.71
64.62
34.78
21.31
16.73
2.70
31.36
13.36
4.18
7.42
11.26
63.00
12.14
15.09
72.57
Residence
Time
[yrs]
0.53
0.59
0.31
0.13
0.15
0.50
0.14
0.01
0.01
0.71
0.72
0.05
0.07
0.47
0.04
1.15
0.71
0.30
1.05
1.96
0.00
0.97
0.05
0.01
0.01
1.79
0.71
0.67
0.01
0.45
2.00
0.06
0.12
1.67
0.16
0.00
2.00
0.01
0.63
0.10
0.04
0.02
0.77
3.70
1.59
0.50
1.67
Epilimnion
DOC
[mg/L]
7.17
2.79
10.10
4.02
4.45
4.63
5.40
3.53
4.75
2.56
3.05
5.94
2.61
3.02
5.24
2.55
3.40
3.52
2.70
2.68
5.30
3.12
8.30
5.40
2.68
6.45
3.02
4.20
3.20
4.72
2.77
4.30
3.58
3.03
4.97
5.45
2.80
5.08
2.81
3.30
6.30
2.94
6.25
9.00
1.92
3.52
3.08
Hypolimnion
DOC
[mg/L]
6.88
6.50
4.38
6.07
3.66
2.63
4.30
2.10
2.24
2.69
2.74
3.80
2.74
3.80
2.52
2.44
5.94
2.59
1.53
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Lake Name
LYFORD
MANSFIELD
MCCONNELL
MILLSFIELD POND
MILTON
MINARDS
MITCHELL
MOOSE POND
MOUNTAIN LAKE- UPPER
NEWARK
NORTH (BRKFLD)
NOTCH
NOYES
PARAN
PARKER
PAUGUS BAY- STN 1
PAWTUCKAWAY LAKE
PEMIGEW ASSET LAKE
PERCH (BENSON)
PLEASANT VALLEY
POUT POND
POWWOW POND
ROBB RESERVOIR
ROUND POND
SAB IN
SHAWS POND
SILVER LAKE
SOMERSET
SOUTH AMERICA
SPRUCE POND
STRATTON
SUNCOOK POND- UPPER
SUNRISE LAKE
SUNSET (BRKFLD)
TRIO PONDS- ONE AND
TWO
TUTTLE (HARDWK)
UNNAMED POND
WALKER POND
WILLEY POND- BIG
WILLEY POND- LITTLE
WILLOUGHBY
WILSON POND
WOLCOTT
ZEPHYR LAKE
pH
7.92
5.87
6.3
6.53
7.23
6.79
7.91
6.56
6.72
7.83
7.81
6.01
6.96
7.92
7.9
6.05
6.45
6.18
7.3
7.35
6.06
6.11
5.49
6.61
7.46
6.73
7.38
5.88
5.17
5.15
6.08
7.63
5.91
6.63
5.43
7.35
4.89
6.6
4.82
4.7
7.67
6.5
6.41
6.33
Lake
Surface
Area
[ha]
13.36
14.16
35.21
65.07
9.71
18.62
11.33
15.01
12.14
61.92
9.71
8.90
15.78
16.19
101.18
493.72
364.22
97.61
9.71
10.12
11.49
99.76
50.71
40.47
57.47
25.09
134.64
634.57
11.74
9.39
18.62
140.26
103.96
10.12
27.44
8.50
8.09
70.46
19.22
13.15
668.97
32.54
29.95
12.50
Hypolimnion
Surface Area
[ha]
4.45
4.72
11.74
21.69
3.24
6.21
3.78
5.00
4.05
20.64
3.24
2.97
5.26
5.40
33.73
164.57
121.41
32.54
3.24
3.37
3.83
33.25
16.90
13.49
19.16
8.36
44.88
211.52
3.91
3.13
6.21
46.75
34.65
3.37
9.15
2.83
2.70
23.49
6.41
4.38
222.99
10.85
9.98
4.17
Residence
Time
[yrs]
0.42
0.11
0.06
0.11
0.42
2.50
0.50
0.07
0.06
1.92
0.08
0.19
0.03
0.03
0.58
0.11
0.45
0.36
2.78
0.17
0.43
0.02
0.03
0.50
0.17
0.36
5.00
0.62
0.09
1.43
0.83
0.13
0.50
0.02
0.50
0.08
0.07
0.31
0.31
0.34
9.09
0.29
0.38
0.33
Epilimnion
DOC
[mg/L]
3.94
4.00
8.66
5.17
3.52
2.06
3.60
7.20
4.46
3.03
3.10
10.91
6.00
3.07
3.26
2.30
5.46
4.32
2.88
1.89
3.60
9.30
8.17
3.42
2.50
3.10
1.86
3.53
7.80
8.33
3.70
3.32
4.09
3.78
5.85
6.30
6.22
4.15
0.35
0.72
1.95
3.55
5.38
3.38
Hypolimnion
DOC
[mg/L]
3.22
1.91
2.57
8.36
2.61
2.10
5.55
4.82
2.78
10.11
2.15
3.28
1.55
3.22
4.10
2.66
5.27
3.85
1.84
4.61
3.66
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 2.5. Parameter Updates for Tier 3.
Lake Name
ADDER POND
ARMINGTON LAKE
BEARCAMP POND
BRANCH
BRUCE
CAWLEY POND
CHASE POND
CHILDS BOG
CLUB POND
CRANBERRY MEADOW
CURTIS
DUNMORE
DUTCHMAN POND
EASTMAN POND
ELM BROOK POOL
FERN
FISH POND
FREESES POND- UPPER
GILES POND
GREAT HOSMER
HALL POND- UPPER
HARDWOOD
HILDRETH DAM POND
HORN POND
HORTONIA
HOWE RESERVOIR
IVANHOE- LAKE
JACKSONVILLE
JENNESS POND
KENT
LARY POND
LEFFERTS
LILY POND
LITTLE AVERILL
LONG (WESTMR)
LOON LAKE
LYFORD
MANSFIELD
MCCONNELL
MILLSFIELD POND
MILTON
MINARDS
MOOSE POND
MOUNTAIN LAKE- UPPER
NEWARK
NOTCH
NOYES
Mean Hgll Cone
[ug/m3]
6.96
7.10
6.94
7.23
7.05
6.96
7.00
7.20
7.05
6.97
6.97
7.10
7.03
7.06
6.98
7.11
7.13
7.24
6.94
7.00
6.99
6.98
7.07
7.03
7.18
7.20
7.02
7.21
7.16
7.14
7.04
7.15
7.16
7.17
7.11
6.98
7.02
6.98
7.15
7.07
7.01
7.16
7.06
7.13
7.12
7.16
7.05
Mean
Precipitation
[cm/yr]
106.04
97.34
116.21
131.05
103.38
103.25
105.40
110.63
107.45
99.01
98.79
99.67
108.83
100.47
106.64
100.17
104.66
106.62
102.58
103.39
116.63
103.22
96.73
115.89
96.84
113.05
116.02
127.86
105.44
108.91
99.52
108.43
112.98
106.23
107.77
105.08
103.88
103.74
103.06
104.84
94.82
101.95
105.71
92.38
107.33
103.10
98.22
RGM1%
[ug/m2/yr]
6.88
6.33
6.53
8.30
6.05
6.49
5.97
7.82
7.84
6.76
6.25
7.07
5.54
6.07
3.43
6.77
6.40
6.49
5.92
5.41
6.14
5.59
6.30
3.64
6.54
7.58
5.83
7.53
6.36
8.00
6.99
8.64
3.77
7.37
7.34
6.07
5.15
8.74
7.70
8.31
7.49
5.68
8.20
6.49
5.95
7.76
8.15
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Lake Name
PARAN
PAUGUS BAY- STN 1
PAWTUCKAWAY LAKE
PEMIGEW ASSET LAKE
PERCH (BENSON)
PLEASANT VALLEY
POUT POND
POWWOW POND
ROBB RESERVOIR
ROUND POND
SAB IN
SHAWS POND
SILVER LAKE
SOUTH AMERICA
SPRUCE POND
STRATTON
SUNRISE LAKE
TRIO PONDS- ONE AND TWO
TUTTLE (HARDWK)
UNNAMED POND
WALKER POND
WILLEY POND- BIG
WILLEY POND- LITTLE
WILSON POND
WOLCOTT
ZEPHYR LAKE
Mean Hgll Cone
[ug/m3]
7.27
6.96
7.22
6.96
7.18
7.20
7.07
7.14
7.16
7.22
6.97
7.01
7.21
7.17
7.20
7.22
7.09
7.11
7.00
7.05
6.95
7.17
7.17
7.22
6.97
7.11
Mean
Precipitation
[cm/yr]
109.78
100.79
108.25
102.24
96.82
116.97
97.09
113.93
109.82
109.65
97.48
111.18
110.68
102.58
104.99
129.43
110.35
110.24
101.76
114.52
106.73
109.47
109.47
104.63
101.26
113.69
RGM1%
[ug/mVyr]
6.69
4.31
4.81
5.77
6.86
7.43
5.66
4.28
8.19
4.29
6.81
6.52
6.58
8.43
2.62
7.99
7.12
8.72
4.17
7.64
5.73
6.76
7.34
4.51
6.96
6.75
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 2.6. Parameter Updates for Tier 4.
Lake Name
ADDER POND
ARMINGTON LAKE
BAKER (BARTON)
BAKER POND- UPPER
BEARCAMP POND
BRANCH
BRUCE
CAWLEY POND
CHASE POND
CHILDS BOG
CHITTENDEN
CLUB POND
CRANBERRY MEADOW
CURTIS
DENNIS
DUNMORE
DUTCHMAN POND
EASTMAN POND
ECHO (HUBDTN)
ECHO (CHARTN)
ELM BROOK POOL
FERN
FISH POND
FREESES POND- UPPER
GILES POND
GREENWOOD POND
GREAT HOSMER
HALL POND- UPPER
HARD WICK
HARDWOOD
HIGH (SUDBRY)
HILDRETH DAM POND
HORN POND
HORTONIA
HOWE RESERVOIR
ISLAND POND
IVANHOE- LAKE
JACKSONVILLE
JENNESS POND
KENT
LARY POND
LITTLE AVERILL
LEFFERTS
LILY POND
LONG (WESTMR)
LOON LAKE
LOVELL LAKE- STN 1
Total
Catchment
Ratio
9.84
8.62
27.22
32.12
45.17
8.71
12.67
304.45
229.87
7.50
13.50
54.00
67.04
11.74
27.85
12.27
3.10
13.65
9.07
26.61
1,277.22
6.32
45.39
154.57
330.49
5.57
5.14
8.02
519.69
5.45
7.65
102.17
57.11
8.30
24.57
137.44
2.20
166.45
6.90
23.00
72.02
5.11
68.35
2.61
7.13
19.02
4.60
Wetland Fraction
0.01
0.02
0.02
0.02
0.03
0.09
0.05
0.02
0.03
0.03
0.02
0.09
0.05
0.02
0.11
0.04
0.04
0.05
0.02
0.01
0.09
0.01
0.02
0.09
0.02
0.13
0.01
0.02
0.05
0.01
0.00
0.02
0.02
0.06
0.05
0.03
0.03
0.07
0.07
0.00
0.07
0.04
0.04
0.09
0.01
0.02
0.02
Lake Cover Fraction
0.11
0.10
0.06
0.03
0.02
0.20
0.14
0.04
0.08
0.09
0.12
0.03
0.10
0.14
0.02
0.13
0.24
0.07
0.13
0.27
0.15
0.16
0.03
0.02
0.02
0.11
0.22
0.08
0.04
0.29
0.14
0.01
0.15
0.20
0.06
0.05
0.14
0.04
0.16
0.09
0.02
0.18
0.06
0.16
0.11
0.05
0.17
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Lake Name
LYFORD
MANSFIELD
MCCONNELL
MILLSFIELD POND
MILTON
MINARDS
MITCHELL
MOOSE POND
MOUNTAIN LAKE- UPPER
NEWARK
NORTH (BRKFLD)
NOTCH
NOYES
PARAN
PARKER
PAUGUS BAY- STN 1
PAWTUCKAWAY LAKE
PEMIGEW ASSET LAKE
PERCH (BENSON)
PLEASANT VALLEY
POUT POND
POWWOW POND
ROBB RESERVOIR
ROUND POND
SHAWS POND
SILVER LAKE
SOMERSET
SOUTH AMERICA
SPRUCE POND
STRATTON
SUNCOOK POND- UPPER
SUNRISE LAKE
SUNSET (BRKFLD)
TRIO PONDS- ONE AND TWO
TUTTLE (HARDWK)
UNNAMED POND
WALKER POND
WILLEY POND- BIG
WILLEY POND- LITTLE
WILLOUGHBY
WILSON POND
WOLCOTT
SAB IN
ZEPHYR LAKE
Total
Catchment
Ratio
8.06
43.49
40.62
39.60
10.13
3.39
165.14
19.71
70.83
2.62
33.54
21.73
60.72
231.80
20.67
181.82
13.73
12.80
3.58
24.96
4.63
78.16
46.68
3.35
16.55
3.24
11.24
16.55
2.83
4.78
91.51
7.22
105.56
11.33
15.29
12.61
32.82
19.21
8.65
6.41
11.71
11.43
62.48
13.90
Wetland Fraction
0.02
0.00
0.12
0.01
0.00
0.01
0.01
0.00
0.02
0.00
0.04
0.09
0.05
0.07
0.01
0.04
0.08
0.03
0.00
0.03
0.01
0.08
0.04
0.08
0.06
0.03
0.04
0.04
0.29
0.19
0.06
0.07
0.04
0.00
0.08
0.02
0.04
0.10
0.05
0.00
0.05
0.04
0.04
0.05
Lake Cover Fraction
0.15
0.13
0.07
0.03
0.15
0.22
0.05
0.04
0.01
0.31
0.08
0.04
0.17
0.04
0.08
0.17
0.08
0.08
0.29
0.09
0.21
0.07
0.02
0.20
0.08
0.23
0.14
0.06
0.26
0.02
0.04
0.09
0.04
0.06
0.00
0.06
0.04
0.07
0.05
0.19
0.05
0.10
0.13
0.06
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 2.7. Parameter Updates for Tier 5.
Lake Name
ADDER POND
ARMINGTON LAKE
BAKER (BARTON)
BAKER POND- UPPER
BEARCAMP POND
BRANCH
BRUCE
CAWLEY POND
CHASE POND
CHILDS BOG
CHITTENDEN
CLUB POND
CRANBERRY MEADOW
CURTIS
DENNIS
DUNMORE
DUTCHMAN POND
EASTMAN POND
ECHO (CHARTN)
ECHO (HUBDTN)
ELM BROOK POOL
FERN
FISH POND
FREESES POND- UPPER
GILES POND
GREAT HOSMER
GREENWOOD POND
HALL POND- UPPER
HARD WICK
HARDWOOD
HIGH (SUDBRY)
HILDRETH DAM POND
HORN POND
HORTONIA
HOWE RESERVOIR
ISLAND POND
IVANHOE- LAKE
JACKSONVILLE
JENNESS POND
KENT
LARY POND
LEFFERTS
LILY POND
LITTLE AVERILL
LONG (WESTMR)
LOON LAKE
LOVELL LAKE- STN 1
Epilimnion
Depth
[m]
6
9
10
6
11
9
3
4
3.9
4.5
9
3
8
10
1
25
3
11
32.79
11
6
13
3
3
6
14
4.5
14
3.7
4
13.9
9
9.5
18
5.5
5.375
7
2
8.83
5
3.5
2
2.5
32
21
9
12
Hypolimnion
Depth
[m]
3
4.4
4.5
0
4.5
5
0
0
0
0
2.5
0
3.25
4
0
17
0
4.5
23.79
5
0
4.5
0
0
0
6.5
0
8.5
0
0
6.9
4
2.75
10
0
1.875
0
0
3.33
0
0
0
0
24
15
3
2.5
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Lake Name
LYFORD
MANSFIELD
MCCONNELL
MILLSFIELD POND
MILTON
MINARDS
MITCHELL
MOOSE POND
MOUNTAIN LAKE- UPPER
NEWARK
NORTH (BRKFLD)
NOTCH
NOYES
PARAN
PARKER
PAUGUS BAY- STN 1
PAWTUCKAWAY LAKE
PEMIGEW ASSET LAKE
PERCH (BENSON)
PLEASANT VALLEY
POUT POND
POWWOW POND
ROBB RESERVOIR
ROUND POND
SAB IN
SHAWS POND
SILVER LAKE
SOMERSET
SOUTH AMERICA
SPRUCE POND
STRATTON
SUNCOOK POND- UPPER
SUNRISE LAKE
SUNSET (BRKFLD)
TRIO PONDS- ONE AND TWO
TUTTLE (HARDWK)
UNNAMED POND
WALKER POND
WILLEY POND- BIG
WILLEY POND- LITTLE
WILLOUGHBY
WILSON POND
WOLCOTT
ZEPHYR LAKE
Epilimnion
Depth
[m]
5.8
6.5
6
4.5
4.32
14.04
4
4.5
6
9
5
8.6
3.9
5
13
23.5
14.75
9.2
11
8
2.5
3.25
2.4
7
17
5
25
25
1.9
5
4.9
13.25
4
9.5
9.75
2
2
13
7
4.5
80
4.5
7
5.85
Hypolimnion
Depth
[m]
1.55
4
2.5
0
0
6.74
0
0
0
2.5
1.5
4.6
0
0
5.5
13.5
9
3.95
3.5
0
0
0.75
0
2.5
10.5
0
14.875
15.25
0
1.5
0
6.75
0
4
4.75
0
0
6.5
0
0
66.75
0
2
1.1
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 3.1. Combinations of Characteristics for Default Runs of R-MCM.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
ACIDITY
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
STRATIFICATION
STRATIFIED
STRATIFIED
STRATIFIED
WELL-MIXED
WELL-MIXED
WELL-MIXED
STRATIFIED
STRATIFIED
STRATIFIED
WELL-MIXED
WELL-MIXED
WELL-MIXED
STRATIFIED
STRATIFIED
STRATIFIED
WELL-MIXED
WELL-MIXED
WELL-MIXED
STRATIFIED
STRATIFIED
STRATIFIED
WELL-MIXED
WELL-MIXED
WELL-MIXED
STRATIFIED
STRATIFIED
STRATIFIED
WELL-MIXED
WELL-MIXED
WELL-MIXED
STRATIFIED
STRATIFIED
STRATIFIED
WELL-MIXED
WELL-MIXED
WELL-MIXED
STRATIFIED
STRATIFIED
STRATIFIED
WELL-MIXED
WELL-MIXED
WELL-MIXED
STRATIFIED
STRATIFIED
STRATIFIED
SIZE
SMALL
SMALL
SMALL
SMALL
SMALL
SMALL
MEDIUM
MEDIUM
MEDIUM
MEDIUM
MEDIUM
MEDUIM
LARGE
LARGE
LARGE
LARGE
LARGE
LARGE
SMALL
SMALL
SMALL
SMALL
SMALL
SMALL
MEDIUM
MEDIUM
MEDIUM
MEDIUM
MEDIUM
MEDUIM
LARGE
LARGE
LARGE
LARGE
LARGE
LARGE
SMALL
SMALL
SMALL
SMALL
SMALL
SMALL
MEDIUM
MEDIUM
MEDIUM
TROPHIC STATE
OLIGOTROPHIC
OLIGOTROPHIC
OLIGOTROPHIC
OLIGOTROPHIC
OLIGOTROPHIC
OLIGOTROPHIC
OLIGOTROPHIC
OLIGOTROPHIC
OLIGOTROPHIC
OLIGOTROPHIC
OLIGOTROPHIC
OLIGOTROPHIC
OLIGOTROPHIC
OLIGOTROPHIC
OLIGOTROPHIC
OLIGOTROPHIC
OLIGOTROPHIC
OLIGOTROPHIC
MESOTROPHIC
MESOTROPHIC
MESOTROPHIC
MESOTROPHIC
MESOTROPHIC
MESOTROPHIC
MESOTROPHIC
MESOTROPHIC
MESOTROPHIC
MESOTROPHIC
MESOTROPHIC
MESOTROPHIC
MESOTROPHIC
MESOTROPHIC
MESOTROPHIC
MESOTROPHIC
MESOTROPHIC
MESOTROPHIC
EUTROPHIC
EUTROPHIC
EUTROPHIC
EUTROPHIC
EUTROPHIC
EUTROPHIC
EUTROPHIC
EUTROPHIC
EUTROPHIC
-------
Evaluating R-MCMfor 91 VT/NH Lakes
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
ACIDITY
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
ACIDIC
CIRCUMNEUTRAL
ALKALINE
STRATIFICATION
WELL-MIXED
WELL-MIXED
WELL-MIXED
STRATIFIED
STRATIFIED
STRATIFIED
WELL-MIXED
WELL-MIXED
WELL-MIXED
STRATIFIED
STRATIFIED
STRATIFIED
WELL-MIXED
WELL-MIXED
WELL-MIXED
STRATIFIED
STRATIFIED
STRATIFIED
WELL-MIXED
WELL-MIXED
WELL-MIXED
STRATIFIED
STRATIFIED
STRATIFIED
WELL-MIXED
WELL-MIXED
WELL-MIXED
SIZE
MEDIUM
MEDIUM
MEDUIM
LARGE
LARGE
LARGE
LARGE
LARGE
LARGE
SMALL
SMALL
SMALL
SMALL
SMALL
SMALL
MEDIUM
MEDIUM
MEDIUM
MEDIUM
MEDIUM
MEDUIM
LARGE
LARGE
LARGE
LARGE
LARGE
LARGE
TROPHIC STATE
EUTROPHIC
EUTROPHIC
EUTROPHIC
EUTROPHIC
EUTROPHIC
EUTROPHIC
EUTROPHIC
EUTROPHIC
EUTROPHIC
DYSTROPHIC
DYSTROPHIC
DYSTROPHIC
DYSTROPHIC
DYSTROPHIC
DYSTROPHIC
DYSTROPHIC
DYSTROPHIC
DYSTROPHIC
DYSTROPHIC
DYSTROPHIC
DYSTROPHIC
DYSTROPHIC
DYSTROPHIC
DYSTROPHIC
DYSTROPHIC
DYSTROPHIC
DYSTROPHIC
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 3.2. Predicted Results for Combination of Default Lake Characteristics.
MeHg in Total Hg in MeHg in Total Hg in MeHg in MeHg in
Epilimnion Epilimnion Hypolimn Hypolimnion MeHg in HgT in Prey Fish Pred Fish
(ng/L, (ng/L, ion (ng/L, (ng/L, Sediments Sediments Muscle Muscle
unfiltered) unfiltered) unfiltered) unfiltered) (ug/g dry) (ug/g dry) (ug/g) (ug/g)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
0.14
0.14
0.05
0.38
0.3
0.09
0.08
0.07
0.03
0.15
0.11
0.04
0.06
0.05
0.03
0.11
0.08
0.03
0.29
0.29
0.13
0.68
0.57
0.21
0.15
0.14
0.06
0.29
0.24
0.08
0.11
0.1
0.04
0.21
0.17
0.06
0.24
0.25
0.18
0.76
0.46
0.38
1.91
1.77
0.92
2.91
2.52
1.02
1.32
1.11
0.46
1.79
1.39
0.48
1.14
0.94
0.39
1.56
1.19
0.41
2.03
1.98
1.13
3
2.78
1.3
1.39
1.28
0.56
1.86
1.64
0.6
1.2
1.09
0.47
1.62
1.41
0.51
1.19
1.21
1.06
1.98
1.69
1.41
0.905
0.8
0.272
0.691
0.572
0.197
0.537
0.439
0.17
2.041
1.946
0.772
1.083
0.963
0.318
0.809
0.71
0.242
0.1557
1.698
1.181
3.723
3.912
2.239
2.292
2.46
1.122
1.95
2.303
0.99
5.121
5.112
3.324
3.04
3.007
1.636
2.487
2.598
1.422
5.46
5.768
5.269
0.016
0.007
0.002
0.01
0.005
0.001
0.011
0.004
0.001
0.005
0.002
0.001
0.008
0.003
0.001
0.004
0.002
0
0.015
0.006
0.002
0.008
0.004
0.001
0.008
0.003
0.001
0.004
0.002
0.001
0.006
0.002
0.001
0.003
0.002
0
0.006
0.003
0.002
0.003
0.002
0.001
0.222
0.166
0.012
0.301
0.237
0.039
0.154
0.098
0.004
0.197
0.136
0.016
0.133
0.077
0.003
0.176
0.118
0.013
0.103
0.085
0.014
0.118
0.104
0.029
0.071
0.054
0.005
0.082
0.067
0.013
0.061
0.045
0.003
0.075
0.061
0.011
0.025
0.021
0.011
0.027
0.025
0.015
0.43
0.3
0.1
1.19
0.67
0.18
0.24
0.15
0.05
0.51
0.26
0.07
0.17
0.1
0.04
0.35
0.18
0.05
0.36
0.26
0.11
0.9
0.54
0.18
0.19
0.13
0.04
0.41
0.23
0.07
0.13
0.09
0.03
0.28
0.16
0.05
0.11
0.09
0.06
0.37
0.23
0.14
0.81
0.58
0.2
2.25
1.29
0.35
0.43
0.27
0.09
0.9
0.46
0.13
0.28
0.18
0.07
0.6
0.31
0.1
0.65
0.49
0.2
1.65
1
0.34
0.34
0.23
0.08
0.72
0.41
0.17
0.22
0.15
0.05
0.48
0.27
0.08
0.2
0.17
0.12
0.67
0.58
0.26
-------
Evaluating R-MCMfor 91 VT/NH Lakes
MeHg in Total Hg in MeHg in Total Hg in MeHg in MeHg in
Epilimnion Epilimnion Hypolimn Hypolimnion MeHg in HgT in Prey Fish Pred Fish
(ng/L, (ng/L, ion (ng/L, (ng/L, Sediments Sediments Muscle Muscle
unfiltered) unfiltered) unfiltered) unfiltered) (ug/g dry) (ug/g dry) (ug/g) (ug/g)
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
0.15
0.15
0.09
0.4
0.32
0.16
0.12
0.11
0.06
0.31
0.25
0.11
0.5
0.49
0.24
0.86
0.79
0.35
0.19
0.19
0.1
0.28
0.26
0.1
0.13
0.13
0.07
0.19
0.18
0.09
0.94
0.94
0.64
1.4
1.31
0.77
0.85
0.85
0.55
1.25
1.18
0.65
4.17
3.98
1.86
5.41
5.08
2.14
2.1
1.97
0.83
2.52
2.33
0.88
1.69
1.58
0.68
2.08
1.91
0.74
0.994
1.044
0.558
0.782
0.82
0.413
1.853
1.591
0.715
0.727
0.609
0.257
0.532
0.301
0.188
3.869
3.807
3.08
3.071
3.215
2.79
5.711
5.315
2.901
2.714
2.538
1.328
2.157
1.879
1.125
0.004
0.002
0.001
0.002
0.001
0.001
0.003
0.001
0.001
0.002
0.001
0
0.009
0.004
0.002
0.006
0.003
0.001
0.004
0.001
0.001
0.002
0.001
0
0.003
0.001
0
0.001
0.001
0
0.02
0.017
0.006
0.023
0.021
0.009
0.018
0.015
0.004
0.022
0.02
0.008
0.086
0.073
0.014
0.118
0.106
0.035
0.043
0.035
0.004
0.059
0.052
0.015
0.035
0.027
0.003
0.05
0.044
0.012
0.08
0.06
0.03
0.21
0.13
0.06
0.05
0.04
0.02
0.15
0.09
0.04
0.38
0.26
0.12
0.72
0.44
0.18
0.15
0.1
0.05
0.24
0.14
0.06
0.1
0.07
0.03
0.16
0.1
0.04
0.13
0.11
0.06
0.37
0.23
0.11
0.09
0.07
0.04
0.26
0.16
0.07
0.83
0.59
0.28
1.55
0.97
0.41
0.27
0.18
0.08
0.43
0.26
0.12
0.17
0.25
0.05
0.27
0.33
0.07
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 3.3. Summary of Results for the Combination of Default Lakes run by R-MCM.
Includes minimum, maximum, mean, median, and the range of predicted values.
Minimum
Maximum
Mean
Median
Range
Epilimnion
MeHg
ng/L,
unfiltered
0.03
0.86
0.21
0.15
0.83
HgT
ng/L,
unfiltered
0.39
5.41
1.54
1.31
5.02
Hypolimnion
MeHg
ng/L,
unfiltered
0.156
2.04
0.77
0.713
1.89
HgT
ng/L,
unfiltered
0.99
5.77
3.08
2.86
4.78
Sediments
MeHg
ug/g dry
0
0.016
0.00317
0.002
0.016
HgT
ug/g dry
0.003
0.301
0.059
0.035
0.298
Fish Muscle
Prey
ug/g
0.02
1.19
0.201
0.135
1.17
Predator
ug/g
0.04
2.25
0.384
0.26
2.21
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 3.4. Examples of Literature Published Ranges for Mercury Concentrations in Different Media in the Environment.
Water (ng/L)
MeHg
0.068-0.61
0.01-1.481,
mean 0.15
HgT
1 -5
0.27-1106.7,
mean 16.6
Sediment (ug/g)
MeHg
0.01-10.851,
mean 1.87
HgT
1.85-451.7,
mean 21.1
Fish (ug/g)
0.05-1.2,
mean 0.5
0.018-5.84,
mean .0478
Reference
1
2
3
1. Driscoll, C.T., C. Yan, C. L. Schofield, R. Munson, J. Holsapple. 1994. The Mercury Cycle and Fish in the Adirondack Lakes.
Environmental Science & Technology. 28:3, 136A-143A.
2. Brumbaugh, W.G., D.P. Krabbenhoft, D.R. Helsel, J.G. Wiener, K.R. Echols. 2001. A National Pilot Study of Mercury
Contamination of Aquatic Ecosystems Along Multiple Gradients: Bioaccumulation in Fish, USGS/BRD/BSR-2001-0009.
3. Krabbenhoft, J.G.Wiener, W.G. Brumbuahg, M.L. Olson, J.F. DeWild, T.J. Sabin. 1999. A National Pilot Study of Mercury
Contamination of Aquatic Ecosystems Along Multiple Gradients. U.S. Geological Survey Toxic Substances Hydrology Program
- Proceedings of the technical meeting, Charleston, S.C., March 8-12, 1999. USGS Water Resources Investigations Report 99-
4018B,Vol2.
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.1. Model Performance Statistics for Default Scenario: %2' Reference %2 at 90% Confidence, and n, number of observations.
Values that pass the %2 goodness of fit test (%2 Value
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.2.
EPI_MeHg
EPI_HgT
HYP_MeHg
HYP_HgT
Fish_Hg
Sed_Hg
Model Performance Statistics for Tier 1 Scenario: % ' Reference % at 90% Confidence, and n, number of observations.
Values that pass the %2 goodness of fit test (%2 Value
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.3.
EPI_MeHg
EPI_HgT
HYP_MeHg
HYP_HgT
Fish_Hg
Sed_Hg
Model Performance Statistics for Tier 2 Scenario: % ' Reference % at 90% Confidence, and n, number of observations.
Values that pass the %2 goodness of fit test (%2 Value
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.4.
EPI_MeHg
EPI_HgT
HYP_MeHg
HYP_HgT
Fish_Hg
Sed_Hg
Model Performance Statistics for Tier 3 Scenario: % ' Reference % at 90% Confidence, and n, number of observations.
Values that pass the %2 goodness of fit test (%2 Value
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.5.
EPI_MeHg
EPI_HgT
HYP_MeHg
HYP_HgT
Fish_Hg
Sed_Hg
Model Performance Statistics for Tier 4 Scenario: % ' Reference % at 90% Confidence, and n, number of observations.
Values that pass the %2 goodness of fit test (%2 Value
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.6.
EPI_MeHg
EPI_HgT
HYP_MeHg
HYP_HgT
Fish_Hg
Sed_Hg
Model Performance Statistics for Tier 5 Scenario: % ' Reference % at 90% Confidence, and n, number of observations.
Values that pass the %2 goodness of fit test (%2 Value
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.7. Model Performance Statistics for Default Scenario: T-test Value, Reference Value at 90% Confidence, number of observations, mean residual,
and Standard Deviation.Values that pass the T-Test at 90% Confidence (T-Test Value< Reference T-Value) are in bold italics.
T-Test Value
Acidic Neutral
Alkaline
T-Test Reference (a=90%) d_bar: mean residual Standard Deviation n
Acidic Neutral Alkaline Acidic Neutral Alkaline Acidic Neutral Alkaline Acidic Neutral Alkaline
0.48 1.25 -4.71
1.92 2.43 -1.53
0.06 0.39 1.00
-2.82 -4.22 -4.27
0.31 3.14 0.55
-4.05 -9.50 -13.93
1.75 1.68 1.70
1.75 1.68 1.70
1.86 1.75 1.75
1.86 1.75 1.75
2.02 1.71 1.75
1.75 1.68 1.70
0.04 0.06 -0.08
1.01 0.67 -0.28
0.03 0.10 0.07
10.14 -7.29 -3.70
0.03 0.21 0.02
-0.17 -0.15 -0.17
0.35 0.35 0.09
2.18 1.84 0.98
1.66 1.11 0.29
10.77 7.12 3.57
0.25 0.34 0.16
0.17 0.11 0.07
17 45 29
17 45 29
9 17 17
9 17 17
6 25 16
17 45 29
EPIJMeHg
EPI_HgT
HYPJMeHg
HYP_HgT
Fish_Hg
Sed_Hg
EPIJMeHg
EPI_HgT
HYPJMeHg
HYP_HgT
Fish_Hg
Sed_Hg
EPIJMeHg
EPI_HgT
Fish_Hg
Sed_Hg
T-Test Value
Small Medium
3.23
3.29
0.68
-3.64
4.14
-9.54
-2.36
-1.23
-0.20
-4.77
-0.84
-13.22
T-Test Value
Mixed Stratified
2.49
2.25
4.91
-7.57
-1.91
0.87
0.12
-12.09
T-Test Reference (a=90%)
Small Medium
d_bar: mean
residual
Small Medium
Standard Deviation
Small Medium
Small Medium
Oligo
T-Test Value
Meso Eutro
T-Test Reference (a=90%)
Dystro Oligo
Meso
Eutro
Dystro
Oligo
d_bar: mean residual
Meso
Eutro
Dystro Oligo
Standard Deviation
Meso
Eutro Dystro Oligo Meso Eutro Dystro
-2.53 0.76 -0.12 1.00
0.13 2.51 -2.74 4.34
1.15 3.97 -1.00 -1.00
-3.05 -4.02 -1.00 -1.00
1.75 2.63 1.18 -0.29
-3.68 -11.21 -7.96 -7.52
1.73 1.70 1.77 1.70
1.73 1.70 1.77 1.70
1.86 1.76 1.94 1.80
1.86 1.76 1.94 1.80
1.81 1.72 2.92 1.80
1.73 1.70 1.77 1.70
-0.10 0.03 0.00 0.08
0.04 0.43 -1.31 1.52
0.12 0.57 -0.25 -0.38
-11.38 -4.23 -3.92 -7.07
0.16 0.18 0.21 -0.02
-0.09 -0.17 -0.19 -0.19
0.17 0.18 0.15 0.45
1.24 0.93 1.80 1.89
0.30 0.55 0.65 1.31
11.19 4.07 10.38 24.50
0.30 0.31 0.30 0.22
0.11 0.08 0.09 0.13
19 29 14 29
19 29 14 29
9 15 7 12
9 15 7 12
11 21 3 12
19 29 14 29
EPIJMeHg
EPI_HgT
HYPJMeHg
HYP_HgT
Fish_Hg
Sed_Hg
Notes:
"Neutral" --> Circumneutral
"Mixed" -> Well-Mixed
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.8. Model Performance Statistics for Tier 1 Scenario: T-test Value, Reference Value at 90% Confidence, number of observations, mean residual,
and Standard Deviation. Values that pass the T-Test at 90% Confidence (T-Test Value< Reference T-Value) are in bold italics.
EPI_MeHg
EPI_HgT
HYPJMeHg
HYP_HgT
Fish_Hg
Sed_Hg
T-Test Value T-Test Reference (cc=90%) d_bar: mean residual Standard Deviation n
Acidic Neutral Alkaline Acidic Neutral Alkaline Acidic Neutral Alkaline Acidic Neutral Alkaline Acidic Neutral Alkaline
0.48 1.25 -4.71
1.92 2.43 -1.53
0.06 0.39 1.00
-2.82 -4.22 -4.27
0.31 3.14 0.55
-4.05 -9.50 -13.93
1.75 1.68 1.70
1.75 1.68 1.70
1.86 1.75 1.75
1.86 1.75 1.75
2.02 1.71 1.75
1.75 1.68 1.70
0.04 0.06 -0.08
1.01 0.67 -0.28
0.03 0.10 0.07
10.14 -7.29 -3.70
0.03 0.21 0.02
-0.17 -0.15 -0.17
0.35 0.35 0.09
2.18 1.84 0.98
1.66 1.11 0.29
10.77 7.12 3.57
0.25 0.34 0.16
0.17 0.11 0.07
17 45 29
17 45 29
9 17 17
9 17 17
6 25 16
17 45 29
EPIJMeHg
EPI_HgT
HYPJMeHg
HYP_HgT
Fish_Hg
Sed_Hg
EPIJMeHg
EPI_HgT
Fish_Hg
Sed_Hg
T-Test Value
Small Medium
T-Test Reference (a=90%)
Small Medium
1.67
1.67
1.72
1.72
1.73
1.67
1.69
1.69
1.72
1.72
1.71
1.69
T-Test Reference (a=90%)
Mixed Stratified
1.68
1.68
1.78
1.68
1.68
1.68
1.69
1.68
d_bar: mean
residual
Small Medium
Standard Deviation
Small Medium
n
Small Medium
Oligo
T-Test Value
Meso Eutro
Dystro
T-Test Reference (a=90%)
Oligo
Meso
Eutro Dystro
Oligo
d_bar: mean residual
Meso
Eutro
Dystro Oligo
Standard Deviation
Meso
Eutro Dystro Oligo Meso Eutro Dystro
-2.53 0.76 -0.12 1.00
0.13 2.51 -2.74 4.34
1.15 3.97 -1.00 -1.00
-3.05 -4.02 -1.00 -1.00
1.75 2.63 1.18 -0.29
-3.68 -11.21 -7.96 -7.52
1.73 1.70 1.77 1.70
1.73 1.70 1.77 1.70
1.86 1.76 1.94 1.80
1.86 1.76 1.94 1.80
1.81 1.72 2.92 1.80
1.73 1.70 1.77 1.70
-0.10 0.03 0.00 0.08
0.04 0.43 -1.31 1.52
0.12 0.57 -0.25 -0.38
-11.38 -4.23 -3.92 -7.07
0.16 0.18 0.21 -0.02
-0.09 -0.17 -0.19 -0.19
0.17 0.18 0.15 0.45
1.24 0.93 1.80 1.89
0.30 0.55 0.65 1.31
11.19 4.07 10.38 24.50
0.30 0.31 0.30 0.22
0.11 0.08 0.09 0.13
19 29 14 29
19 29 14 29
9 15 7 12
9 15 7 12
11 21 3 12
19 29 14 29
EPIJMeHg
EPI_HgT
HYPJMeHg
HYP_HgT
Fish_Hg
Sed_Hg
Notes:
"Neutral" --> Circumneutral
"Mixed" --> Well-Mixed
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.9. Model Performance Statistics for Tier 2 Scenario: T-test Value, Reference Value at 90% Confidence, number of observations, mean residual,
and Standard Deviation. Values that pass the T-Test at 90% Confidence (T-Test Value< Reference T-Value) are in bold italics.
Acidic
T-Test Value
Neutral Alkaline
T-Test Reference (oc=90%) d_bar: mean residual Standard Deviation n
Acidic Neutral Alkaline Acidic Neutral Alkaline Acidic Neutral Alkaline Acidic Neutral Alkaline
-2.85 -4.65 -7.32
-2.88 -5.02 -3.09
-0.32 -1.17 1.00
-1.96 -4.77 -3.55
-3.02 0.20 -0.11
-5.74 -12.18 -13.62
1.75 1.68 1.70
1.75 1.68 1.70
1.86 1.75 1.75
1.86 1.75 1.75
2.02 1.71 1.75
1.75 1.68 1.70
-0.22 -0.23 -0.14
-0.96 -1.27 -0.65
-0.19 -0.34 0.01
-8.89 -8.44 -3.31
-0.25 0.01 0.00
-0.21 -0.19 -0.17
0.31 0.33 0.10
1.37 1.70 1.13
1.76 1.19 0.03
13.60 7.29 3.84
0.20 0.34 0.12
0.15 0.11 0.07
17 45 29
17 45 29
9 17 17
9 17 17
6 25 16
17 45 29
EPI_MeHg
EPI_HgT
HYPJMeHg
HYP_HgT
Fish_Hg
Sed_Hg
EPIJMeHg
EPI_HgT
HYPJMeHg
HYP_HgT
Fish_Hg
Sed_Hg
EPI_MeHg
EPI_HgT
Fish_Hg
Sed_Hg
T-Test Value
Small Medium
-6.22
-4.52
-0.36
-3.81
1.41
-12.37
-3.80
-4.95
-1.41
-3.44
-3.85
-13.66
T-Test Value
Mixed Stratified
-4.59
-4.55
0.86
-10.34
-5.09
-4.81
-2.85
-13.40
T-Test Reference (o=90%)
Small Medium
1.67
1.67
1.72
1.72
1.73
1.67
1.69
1.69
1.72
1.72
1.71
1.69
T-Test Reference (o=90%)
Mixed Stratified
1.68
1.68
1.78
1.68
1.68
1.68
1.69
1.68
d_bar: mean
residual
Small Medium
Standard Deviation
Small Medium
Small Medium
Oligo
T-Test Value
Meso Eutro
Dystro
T-Test Reference (o=90%)
Oligo Meso Eutro Dystro
Oligo
d_bar: mean residual
Meso Eutro
Dystro Oligo
Standard Deviation
Meso Eutro
Dystro Oligo Meso Eutro Dystro
-3.49 -3.55 -3.30 -1.00
-1.27 -3.07 -3.59 -5.51
0.93 2.20 -1.00 -1.00
-2.79 -3.61 -1.00 -1.00
-1.15 1.03 1.38 -2.61
-5.16 -13.39 -8.71 -10.26
1.73 1.70 1.77 1.70
1.73 1.70 1.77 1.70
1.86 1.76 1.94 1.80
1.86 1.76 1.94 1.80
1.81 1.72 2.92 1.80
1.73 1.70 1.77 1.70
-0.12 -0.12 -0.19 -0.34
-0.31 -0.49 -2.04 -1.50
0.06 0.37 -0.78 -0.66
-11.06 -4.39 -4.73 -6.78
-0.05 0.07 0.10 -0.21
-0.12 -0.19 -0.19 -0.24
0.15 0.18 0.21 1.84
1.08 0.86 2.13 1.46
0.18 0.66 2.07 2.30
11.90 4.72 12.50 23.47
0.14 0.31 0.13 0.27
0.10 0.08 0.08 0.12
19 29 14 29
19 29 14 29
9 15 7 12
9 15 7 12
11 21 3 12
19 29 14 29
EPIJMeHg
EPI_HgT
HYPJMeHg
HYP_HgT
Fish_Hg
Sed_Hg
Notes:
"Neutral" --> Circumneutral
"Mixed" -> Well-Mixed
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.10. Model Performance Statistics for Tier 3 Scenario: T-test Value, Reference Value at 90% Confidence, number of observations, mean residual
,and Standard Deviation. Values that pass the T-Test at 90% Confidence (T-Test Value< Reference T-Value) are in bold italics.
EPI_MeHg
EPI_HgT
HYPJMeHg
HYP_HgT
Fish_Hg
Sed_Hg
T-Test Value T-Test Reference (cc=90%) d_bar: mean residual Standard Deviation n
Acidic Neutral Alkaline Acidic Neutral Alkaline Acidic Neutral Alkaline Acidic Neutral Alkaline Acidic Neutral Alkaline
-2.85 -4.65 -7.32
-2.88 -5.02 -3.09
-0.32 -1.17 1.00
-1.96 -4.77 -3.55
-3.02 0.20 -0.11
-5.74 -12.18 -13.62
1.75 1.68 1.70
1.75 1.68 1.70
1.86 1.75 1.75
1.86 1.75 1.75
2.02 1.71 1.75
1.75 1.68 1.70
-0.22 -0.23 -0.14
-0.96 -1.27 -0.65
-0.19 -0.34 0.01
-8.89 -8.44 -3.31
-0.25 0.01 0.00
-0.21 -0.19 -0.17
0.31 0.33 0.10
1.37 1.70 1.13
1.76 1.19 0.03
13.60 7.29 3.84
0.20 0.34 0.12
0.15 0.11 0.07
17 45 29
17 45 29
9 17 17
9 17 17
6 25 16
17 45 29
T-Test Value
Small Medium
T-Test Reference (o=90%)
Small Medium
d_bar: mean residual
Small Medium
Standard Deviation
Small Medium
n
Small Medium
EPIJMeHg
EPI_HgT
HYPJMeHg
HYP_HgT
Fish_Hg
Sed_Hg
EPI_MeHg
EPI_HgT
Fish_Hg
Sed_Hg
-6.22
-4.52
-0.36
-3.81
1.41
-12.37
-3.80
-4.95
-1.41
-3.44
-3.85
-13.66
T-Test Value
Mixed Stratified
-4.59
-4.55
0.86
-10.34
-5.09
-4.81
-2.85
-13.40
-0.18
-0.96
-0.11
-7.31
0.11
-0.20
-0.23
-1.11
-0.23
-5.74
-0.13
-0.18
d_bar: mean residual
Mixed Stratified
-0.22
-1.26
0.11
-0.17
-0.18
-0.79
-0.08
-0.20
Oligo
T-Test Value
Meso Eutro
Dystro
Oligo
T-Test Reference (o=90%)
Meso
Eutro Dystro
Oligo
d_bar: mean residual
Meso
Eutro
Dystro Oligo
Standard Deviation
Meso
Eutro Dystro Oligo Meso Eutro Dystro
-3.49 -3.55 -3.30 -1.00
-1.27 -3.07 -3.59 -5.51
0.93 2.20 -1.00 -1.00
-2.79 -3.61 -1.00 -1.00
-1.15 1.03 1.38 -2.61
-5.16 -13.39 -8.71 -10.26
1.73 1.70 1.77 1.70
1.73 1.70 1.77 1.70
1.86 1.76 1.94 1.80
1.86 1.76 1.94 1.80
1.81 1.72 2.92 1.80
1.73 1.70 1.77 1.70
-0.12 -0.12 -0.19 -0.34
-0.31 -0.49 -2.04 -1.50
0.06 0.37 -0.78 -0.66
-11.06 -4.39 -4.73 -6.78
-0.05 0.07 0.10 -0.21
-0.12 -0.19 -0.19 -0.24
0.15 0.18 0.21 1.84
1.08 0.86 2.13 1.46
0.18 0.66 2.07 2.30
11.90 4.72 12.50 23.47
0.14 0.31 0.13 0.27
0.10 0.08 0.08 0.12
19 29 14 29
19 29 14 29
9 15 7 12
9 15 7 12
11 21 3 12
19 29 14 29
EPI_MeHg
EPI_HgT
HYPJMeHg
HYP_HgT
Fish_Hg
Sed_Hg
Notes:
"Neutral" --> Circumneutral
"Mixed" -> Well-Mixed
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.11. Model Performance Statistics for Tier 4 Scenario: T-test Value, Reference Value at 90% Confidence, number of observations, mean residual,
and Standard Deviation. Values that pass the T-Test at 90% Confidence (T-Test Value< Reference T-Value) are in bold italics.
T-Test Value T-Test Reference (cc=90%) d_bar: mean residual Standard Deviation n
Acidic Neutral Alkaline Acidic Neutral Alkaline Acidic Neutral Alkaline Acidic Neutral Alkaline Acidic Neutral Alkaline
-2.73 -4.01 -4.89
-1.35 -3.17 -0.23
-0.46 -0.78 1.00
-2.00 -4.22 -2.87
-2.14 1.16 1.34
-5.04 -10.03 -8.84
1.75 1.68 1.70
1.75 1.68 1.70
1.86 1.75 1.75
1.86 1.75 1.75
2.02 1.71 1.75
1.75 1.68 1.70
-0.21 -0.21 -0.11
-0.55 -0.78 -0.05
-0.26 -0.22 0.02
-9.03 -7.74 -2.87
-0.22 0.09 0.03
-0.19 -0.16 -0.14
0.31 0.34 0.12
1.68 1.64 1.17
1.71 1.18 0.07
13.51 7.57 4.12
0.25 0.40 0.10
0.15 0.11 0.09
17 45 29
17 45 29
9 17 17
9 17 17
6 25 16
17 45 29
EPIJMeHg
EPI_HgT
HYPJMeHg
HYP_HgT
Fish_Hg
Sed_Hg
EPIJMeHg
EPI_HgT
HYPJMeHg
HYP_HgT
Fish_Hg
Sed Hg
EPIJMeHg
EPI_HgT
Fish_Hg
Sed Hg
T-Test Value
Small Medium
-4.77 -3.61
-1.64 -3.30
-0.55 -0.66
-3.86 -2.84
2.26 -2.79
-9.98 -10.57
T-Test Value
Mixed Stratified
-3.64 -4.60
-2.31 -2.13
1.49 -1.30
-8.35 -10.93
T-Test Reference (a=90%)
Small Medium
1.67 1.69
1.67 1.69
1.72 1.72
1.72 1.72
1.73 1.71
1.67 1.69
T-Test Reference (a=90%)
Mixed Stratified
T-Test Value
Oligo Meso Eutro
1.68 1.68
1.68 1.68
1.78 1.69
1.68 1.68
d_bar: mean residual
Small Medium
-0.15 -0.22
-0.36 -0.75
-0.16 -0.12
-7.28 -4.94
0.20 -0.09
-0.17 -0.15
d_bar: mean residual
Mixed Stratified
-0.18 -0.17
-0.67 -0.36
0.21 -0.04
-0.14 -0.18
T-Test Reference (a=90%)
Dystro Oligo Meso Eutro Dystro
Oligo
Standard Deviation
Small Medium
0.23 0.36
1.64 1.32
1.32 0.83
8.64 8.16
0.40 0.17
0.13 0.08
Standard Deviation
Mixed Stratified
0.32 0.26
1.85 1.21
0.52 0.16
0.10 0.12
d_bar: mean residual
Meso Eutro
Small Medium
Standard Deviation
Dystro Oligo
Meso
Eutro Dystro Oligo Meso Eutro Dystro
-2.50 -2.18 -3.49 -1.00
0.94 0.59 -3.43 -3.01
0.42 3.16 -1.00 -1.00
-2.63 -2.97 -1.00 -1.00
-0.35 1.88 1.64 -1.80
-3.33 -9.05 -8.04 -9.41
1.73 1.70 1.77 1.70
1.73 1.70 1.77 1.70
1.86 1.76 1.94 1.80
1.86 1.76 1.94 1.80
1.81 1.72 2.92 1.80
1.73 1.70 1.77 1.70
-0.10 -0.08 -0.19 -0.32
0.22 0.08 -1.91 -0.88
0.03 0.45 -0.79 -0.62
-10.88 -3.67 -4.65 -6.35
-0.01 0.15 0.09 -0.15
-0.09 -0.15 -0.19 -0.21
0.17 0.19 0.20 1.70
1.02 0.74 2.08 1.58
0.24 0.56 2.08 2.16
12.43 4.78 12.30 21.99
0.14 0.38 0.09 0.29
0.11 0.09 0.09 0.12
19 29 14 29
19 29 14 29
9 15 7 12
9 15 7 12
11 21 3 12
19 29 14 29
EPIJMeHg
EPI_HgT
HYPJMeHg
HYP_HgT
Fish_Hg
Sed_Hg
Notes:
"Neutral" --> Circumneutral
"Mixed" -> Well-Mixed
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.12. Model Performance Statistics for Tier 5 Scenario: T-test Value, Reference Value at 90% Confidence, number of observations, mean residual,
and Standard Deviation. Values that pass the T-Test at 90% Confidence (T-Test Value< Reference T-Value) are are in bold italics.
EPI_MeHg
EPI_HgT
HYPJMeHg
HYP_HgT
Fish_Hg
Sed_Hg
T-Test Value T-Test Reference (cc=90%) d_bar: mean residual Standard Deviation n
Acidic Neutral Alkaline Acidic Neutral Alkaline Acidic Neutral Alkaline Acidic Neutral Alkaline Acidic Neutral Alkaline
-3.78 -4.72 -8.58
-2.60 -3.97 -3.03
-1.49 -1.88 -1.00
-2.77 -4.81 -5.09
-3.45 0.57 -2.30
-5.62 -10.55 -11.42
1.75 1.68 1.70
1.75 1.68 1.70
1.86 1.75 1.75
1.86 1.75 1.75
2.02 1.71 1.75
1.75 1.68 1.70
-0.26 -0.23 -0.15
-0.90 -0.95 -0.49
-0.80 -0.49 -0.24
-11.27 -8.59 -4.27
-0.26 0.04 -0.05
-0.21 -0.18 -0.17
0.28 0.32 0.09
1.44 1.60 0.86
1.61 1.07 0.99
12.23 7.36 3.45
0.18 0.34 0.09
0.16 0.11 0.08
17 45 29
17 45 29
9 17 17
9 17 17
6 25 16
17 45 29
T-Test Value
Small Medium
T-Test Reference (o=90%)
Small Medium
d_bar: mean residual
Small Medium
Standard Deviation
Small Medium
EPIJMeHg
EPI_HgT
HYPJMeHg
HYP_HgT
Fish_Hg
Sed_Hg
EPI_MeHg
EPI_HgT
Fish_Hg
Sed_Hg
-0.20
-0.75
-0.60
-8.80
0.08
-0.19
-0.22
-0.86
-0.32
-6.15
-0.11
-0.16
d_bar: mean residual
Mixed Stratified
-0.20
-0.75
0.19
-0.14
-0.21
-0.81
-0.11
-0.21
Oligo
T-Test Value
Meso Eutro
T-Test Reference (o=90%)
Dystro Oligo
Meso
Eutro
Dystro
Oligo
d_bar: mean residual
Meso
Eutro
Dystro Oligo
Standard Deviation
Meso
Eutro
Dystro Oligo Meso Eutro Dystro
-4.29 -4.06 -4.00 -1.00
-1.31 -1.92 -3.71 -4.21
-2.18 0.84 -1.00 -1.00
-3.04 -4.46 -1.00 -1.00
-2.55 1.14 1.27 -2.98
-4.94 -10.86 -8.03 -8.75
1.73 1.70 1.77 1.70
1.73 1.70 1.77 1.70
1.86 1.76 1.94 1.80
1.86 1.76 1.94 1.80
1.81 1.72 2.92 1.80
1.73 1.70 1.77 1.70
-0.14 -0.12 -0.20 -0.34
-0.23 -0.28 -1.97 -1.11
-0.12 0.08 -0.98 -1.08
-12.16 -4.90 -5.68 -8.10
-0.09 0.08 0.06 -0.19
-0.12 -0.18 -0.19 -0.22
0.14 0.16 0.19 1.85
0.77 0.79 1.99 1.41
0.16 0.39 2.59 3.75
12.00 4.26 15.04 28.05
0.12 0.33 0.08 0.22
0.11 0.09 0.09 0.13
19 29 14 29
19 29 14 29
9 15 7 12
9 15 7 12
11 21 3 12
19 29 14 29
EPI_MeHg
EPI_HgT
HYPJMeHg
HYP_HgT
Fish_Hg
Sed_Hg
Notes:
"Neutral" --> Circumneutral
"Mixed" -> Well-Mixed
-------
w
|
2
w
2
cr
1
0
o
&
Alkaline
"3
S i
1
g
o
VH
U
o
"o
fi
ฃ "3
g pq
pq
k^
3 2
2 H
Ui ^-H
pq ^
OJ
o3
a1
CO
flj i H
S S
O
ฃ
^ 00 ft g CN 0
BS^SL^
^ 2^ฐ 2 g
^ oo r^ ^/ r^ ,x
^1 ^ ^ ^1 ^j VO
oo ^ ^ ON ^ oo
W
oo
S- .2
2 "^
^ -fc:
W oo
8
C3
a"
oo -a
aj i5
o 'aj
g ON in in
H ON (N r~
g ^ ON [<
^ OO VO OO
2 In oq p
jn S ฐ p
M^
^ O
s 13
cr &
oo g
J 8
^ 2
Q
O
O
13
D.
S
0
o
ซ ^ s 2 - ic
^2^^oo
^
SSSฐฐ"
iX ; iX ON OO O
$ S 3 - ฃ o
i
ซ CN
S o
0 ซ
P 00
S ฐ
cn
d
cn
in
d
o
13
D.
g
G
o
p.
G 2
S "3
e- w
0
a
' N d)
^ 'Z3
& g
o oo
fcl
W
S
3 ,_
-ง 'S
Ctf k_H
s S
'a)
^
* (N cn cn
vq ON in in
^ cn d d
ON oo vo in
ON o oo ^t-
d mi d d
!-H
w
II
S p.
o
S 0
2 ^
^
O
S
D-
0
O
^ t~- ro z? O
\O ^O \O - V~)
- ^ r4 o d
O oo o cn vo
cn o ON r^ in
d mi r>i ON d
(O oo c^i ,-J cn
in o r~ ฐ; oo
d r4 ^H 2 d
^-* O ON 2S vo
^(- ^^ ^t- H oo
d cn d ^ d
cn
in
d
in
cn
d
in
d
(N
d
I
งB?
w w
I H ^ -,
I I OH OH
E E fe fe
pq pq K W
(X .2
M CL,
pq pq
S
^ ^ "S
W HH 00
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.14. Model Performance Statistics for Default Scenario:
Coefficient of Determination, Modeling Efficiency, and Coefficient of Residual Mass.
Coefficient of Determination (CD)
Acidic Circumneutral Alkaline
Modeling Efficiency (EF)
Acidic Circumneutral Alkaline
Coefficient of Residual Mass (CRM)
Acidic Circumneutral Alkaline
1.23
0.54
14.17
0.94
0.44
0.47
1.26
0.91
2.61
0.87
0.12
0.42
0.68
4.51
1.24
0.72
0.25
0.14
0.19
-0.84
0.93
-0.07
-1.27
-1.12
0.20
-0.10
0.62
-0.15
-7.28
-1.35
-0.47
0.78
0.19
-0.40
-3.01
-5.95
-0.12
-0.54
-0.03
0.70
-0.08
0.61
-0.20
-0.35
-0.11
0.68
-0.88
0.64
0.39
0.22
-0.15
0.58
-0.16
0.88
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Coefficient of Determination (CD)
Small Medium
0.37
0.66
5.03
1.47
0.10
0.44
4.73
2.65
4.87
0.93
1.01
0.15
Modeling Efficiency (EF)
Small Medium
-1.68
-0.52
0.80
0.32
-8.58
-1.26
0.79
0.62
0.79
-0.08
0.01
-5.65
Coefficient of Residual Mass (CRM)
Small Medium
-0.33
-0.48
-0.18
0.60
-1.56
0.68
0.48
0.15
0.05
0.74
0.08
0.74
Coefficient of Determination (CD)
Well-Mixed Stratified
EPI_Me
EPI_T
Fish
Sed
0.70
0.82
0.06
0.40
2.50
0.94
0.90
0.31
Modeling Efficiency (EF)
Well-
Mixed Stratified
Coefficient of Residual Mass (CRM)
-0.42
-0.23
-14.55
-1.50
0.60
-0.06
-0.11
-2.19
Well-Mixed
-0.35
-0.39
-1.80
0.60
Stratified
0.28
-0.10
-0.01
0.77
Oligo
Coefficient of Determination (CD)
Meso Eutro Dystro
Oligo
Modeling Efficiency (EF)
Meso Eutro Dystro
Coefficient of Residual Mass (CRM)
Oligo Meso Eutro Dystro
0.84
1.60
0.27
0.98
0.09
0.33
0.29
0.55
0.11
0.97
0.08
0.24
0.96 1.83 -0.18 -2.50 -0.04 0.45 0.45 -0.11 0.01 -0.21
2.07 0.38 0.38 -0.83 0.52 -1.66 -0.03 -0.38 0.51 -0.72
10.91 3.61 -2.75 -7.89 0.91 0.72 -0.35 -1.55 0.17 0.26
0.93 0.85 -0.02 -0.03 -0.08 -0.17 0.84 0.58 0.45 0.66
0.00 0.36 -9.63 -12.07 -362.50 -1.79 -0.74 -0.99 -3.45 0.05
0.21 0.47 -1.99 -3.13 -3.85 -1.14 0.45 0.76 0.91 0.70
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.15. Model Performance Statistics for Tier 1 Scenario:
Maximum Error and Root Mean Square Error.
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Acidic
Maximum Error (ME)
Circumneutral
Alkaline
1.27
5.04
3.24
27.52
0.46
0.53
1.81
6.84
2.87
27.03
1.17
0.59
0.43
4.00
1.10
11.95
0.27
0.30
Root Mean Square Error (RMSE)
Acidic Circumneutral Alkaline
105.39
85.37
128.86
111.00
80.49
91.46
124.25
108.60
128.03
104.37
146.86
93.10
83.72
102.45
55.24
81.25
90.81
92.49
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Maximum Error (ME)
Small Medium
0.79
6.84
3.24
27.52
1.17
0.59
1.81
5.81
2.52
24.85
0.46
0.36
Root Mean Square Error (RMSE)
Small Medium
93.86
104.73
129.18
104.85
184.64
94.04
151.51
104.07
128.09
113.88
86.85
90.40
Maximum Error (ME)
Well-Mixed Stratified
EPI_Me
EPI_T
Fish
Sed
1.27
6.84
1.17
0.47
1.81
5.81
0.39
0.59
Root Mean Square Error (RMSE)
Well-
Mixed Stratified
108.98
109.33
194.71
91.07
122.24
93.83
79.08
93.28
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Oliso
Maximum Error (ME)
Meso Eutro Dystro
0.44
3.13
0.35
27.52
0.26
0.25
0.54
2.89
1.25
16.06
1.17
0.47
0.79
6.84
3.24
11.56
0.27
0.35
1.81
5.81
3.19
22.13
0.63
0.59
Oliso
Root Mean Square Error (RMSE)
Meso Eutro Dystro
89.09
85.73
47.74
118.48
67.60
73.00
95.25
85.48
169.36
90.02
184.57
89.42
85.51
110.98
155.04
156.07
259.68
100.46
458.03
97.23
138.50
1,198.50
15.99
12.49
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.16. Model Performance Statistics for Tier 1 Scenario:
Coefficient of Determination, Modeling Efficiency, and Coefficient of Residual Mass.
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Coefficient of Determination (CD)
Acidic Circumneutral Alkaline
Modeling Efficiency (EF) Coefficient of Residual Mass (CRM)
Acidic Circumneutral Alkaline Acidic Circumneutral Alkaline
1.01
1.26
6.43
0.85
0.12
0.34
1.13
1.11
1.93
0.60
0.13
0.27
0.38
1.92
1.65
0.74
0.44
0.15
0.01
0.21
0.84
-0.18
-7.10
-1.94
0.11
0.10
0.48
-0.65
-6.73
-2.74
-1.67
0.48
0.39
-0.35
-1.29
-5.72
0.60
0.47
0.31
0.66
0.64
0.74
0.71
0.63
0.47
0.82
-0.10
0.81
0.68
0.49
0.13
0.59
0.03
0.86
Coefficient of Determination (CD)
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Small
Medium
0.51
1.67
4.83
1.04
0.14
0.34
2.03
0.94
2.78
0.76
0.58
0.12
Modeling Efficiency (EF)
Small Medium
-0.97
0.40
0.79
0.04
-6.25
-1.98
0.51
-0.06
0.64
-0.31
-0.71
-7.27
Coefficient of Residual Mass (CRM)
Small Medium
0.59
0.51
0.27
0.72
-0.60
0.80
0.84
0.66
0.46
0.70
0.52
0.83
EPI_Me
EPI_T
Fish
Sed
Coefficient of Determination (CD)
Well-
Mixed Stratified
0.74
1.17
0.07
0.26
1.58
1.48
1.17
0.27
Modeling Efficiency (EF)
Well-
Mixed Stratified
Coefficient of Residual Mass (CRM)
-0.35
0.15
-12.78
-2.88
0.37
0.32
0.14
-2.64
Well-
Mixed
0.62
0.61
-0.52
0.77
Stratified
0.73
0.51
0.35
0.84
Coefficient of Determination (CD)
Oligo Meso Eutro Dystro
Oligo
Modeling Efficiency (EF)
Meso Eutro Dystro
Coefficient of Residual Mass (CRM)
Oligo Meso Eutro Dystro
0.76
2.14
1.78
0.97
1.06
0.29
0.26
0.82
0.22
0.76
0.07
0.20
0.76 1.00 -0.32 -2.83 -0.32 0.00 0.56 0.51 0.57 0.85
0.95 0.62 0.53 -0.23 -0.06 -0.62 0.19 0.39 0.77 0.69
2.20 1.65 0.44 -3.56 0.55 0.39 0.02 -0.72 0.59 0.59
0.55 0.59 -0.03 -0.32 -0.80 -0.70 0.85 0.65 0.59 0.68
0.01 0.14 0.06 -12.71 -90.27 -6.00 0.24 -0.44 -1.84 0.56
0.20 0.30 -2.47 -3.98 -4.00 -2.36 0.54 0.83 0.93 0.88
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.17. Model Performance Statistics for Tier 2 Scenario:
Maximum Error and Root Mean Square Error.
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Acidic
Maximum Error (ME)
Circumneutral
Alkaline
1.27
5.13
3.34
27.14
0.46
0.53
1.81
6.84
2.78
26.84
1.08
0.59
0.43
4.00
1.19
11.69
0.25
0.30
Root Mean Square Error (RMSE)
Acidic Circumneutral Alkaline
105.27
86.64
128.98
107.27
79.95
92.11
123.79
109.04
130.25
102.18
138.98
94.04
83.58
103.10
7.22
78.55
87.77
93.43
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Maximum Error (ME)
Small Medium
0.79
6.84
3.34
27.14
1.08
0.59
1.81
5.81
2.47
24.23
0.46
0.36
Root Mean Square Error (RMSE)
Small Medium
93.42
105.37
131.88
101.33
173.26
94.76
151.28
104.70
127.46
111.53
86.19
91.60
Maximum Error (ME)
Well-Mixed Stratified
EPI_Me
EPI_T
Fish
Sed
1.27
6.84
1.08
0.48
1.81
5.81
0.40
0.59
Root Mean Square Error (RMSE)
Well-
Mixed Stratified
108.50
109.74
183.09
91.77
122.09
94.83
78.68
94.23
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Oliso
Maximum Error (ME)
Meso Eutro Dystro
0.44
3.19
0.33
27.14
0.26
0.26
0.54
2.90
1.48
15.21
1.08
0.48
0.79
6.84
3.34
10.70
0.25
0.35
1.81
5.81
2.78
20.14
0.58
0.59
Oliso
Root Mean Square Error (RMSE)
Meso Eutro Dystro
88.14
85.56
54.82
116.68
66.54
75.25
93.57
85.53
202.70
86.21
172.98
90.45
85.89
111.80
145.09
143.30
239.24
101.00
458.03
97.97
108.53
1,108.00
15.68
12.55
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.18. Model Performance Statistics for Tier 2 Scenario:
Coefficient of Determination, Modeling Efficiency, and Coefficient of Residual Mass.
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Coefficient of Determination (CD)
Acidic Circumneutral Alkaline
Modeling Efficiency (EF) Coefficient of Residual Mass (CRM)
Acidic Circumneutral Alkaline Acidic Circumneutral Alkaline
1.02
1.20
5.12
0.96
0.12
0.33
1.14
1.07
1.90
0.65
0.15
0.26
0.38
1.80
1.18
0.85
0.48
0.15
0.02
0.17
0.80
-0.04
-7.06
-1.99
0.12
0.06
0.47
-0.54
-5.87
-2.83
-1.65
0.44
0.15
-0.17
-1.10
-5.88
0.61
0.51
0.15
0.61
0.64
0.76
0.71
0.66
0.37
0.78
-0.06
0.83
0.68
0.52
-0.02
0.52
0.03
0.87
Coefficient of Determination (CD)
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Small
Medium
0.52
1.60
3.76
1.20
0.16
0.33
2.06
0.91
2.87
0.81
0.60
0.12
Modeling Efficiency (EF)
Small Medium
-0.93
0.37
0.73
0.17
-5.39
-2.04
0.51
-0.10
0.65
-0.23
-0.66
-7.51
Coefficient of Residual Mass (CRM)
Small Medium
Coefficient of Determination (CD)
Well-Mixed Stratified
Modeling Efficiency (EF)
Well-Mixed Stratified
EPI_Me
EPI_T
Fish
Sed
0.75 1.60
1.14 1.39
0.08 1.20
0.25 0.27
-0.33 0.37
0.12 0.28
-11.14 0.17
-2.93 -2.74
0.64 0.73
0.64 0.54
-0.44 0.36
0.78 0.85
Coefficient of Determination (CD)
Oligo Meso Eutro Dystro
Oligo
Modeling Efficiency (EF)
Meso Eutro Dystro
Coefficient of Residual Mass (CRM)
Oligo Meso Eutro Dystro
0.79
2.21
1.04
1.04
1.15
0.27
0.27
0.80
0.14
0.85
0.08
0.20
0.77 1.00 -0.26 -2.65 -0.29 0.00 0.56 0.52 0.58 0.85
0.92 0.60 0.55 -0.25 -0.09 -0.68 0.25 0.43 0.79 0.71
2.52 1.62 0.04 -5.93 0.60 0.38 -0.17 -1.03 0.55 0.47
0.64 0.65 0.04 -0.17 -0.56 -0.53 0.82 0.60 0.54 0.63
0.01 0.15 0.13 -11.05 -77.72 -5.68 0.23 -0.39 -1.67 0.58
0.20 0.29 -2.69 -4.07 -4.08 -2.40 0.58 0.84 0.93 0.89
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.19. Model Performance Statistics for Tier 3 Scenario:
Maximum Error and Root Mean Square Error.
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Maximum Error (ME)
Acidic Circumneutral Alkaline
1.27
4.99
3.34
27.14
0.45
0.49
1.81
6.82
2.78
26.84
1.74
0.58
0.43
4.00
1.19
11.69
0.28
0.30
Root Mean Square Error (RMSE)
Acidic Circumneutral Alkaline
104.68
85.66
128.97
107.27
78.21
87.24
127.56
108.22
130.18
102.29
189.05
90.94
80.94
104.10
7.22
78.62
97.16
90.91
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Maximum Error (ME)
Small Medium
0.82
6.82
3.34
27.14
1.74
0.58
1.81
5.81
2.47
24.23
0.45
0.36
Root Mean Square Error (RMSE)
Small Medium
97.54
105.16
131.87
101.37
243.13
90.85
150.70
103.46
127.33
111.62
87.31
89.23
Maximum Error (ME)
Well-Mixed Stratified
EPI_Me
EPI_T
Fish
Sed
1.27
6.82
1.74
0.47
1.81
5.81
0.38
0.58
Root Mean Square Error (RMSE)
Well-
Mixed Stratified
113.65
110.33
253.28
89.40
120.01
92.39
80.07
90.16
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Oliso
Maximum Error (ME)
Meso Eutro Dystro
0.43
3.19
0.33
27.14
0.25
0.24
0.82
2.89
1.48
15.21
1.74
0.47
0.79
6.82
3.34
10.70
0.25
0.35
1.81
5.81
2.78
20.14
0.76
0.58
Oliso
Root Mean Square Error (RMSE)
Meso Eutro Dystro
90.32
91.57
54.82
116.68
66.51
72.49
110.72
93.89
202.36
86.51
252.08
86.34
84.01
109.94
145.09
143.30
243.63
100.46
436.75
94.71
108.52
1,111.70
16.46
12.03
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.20. Model Performance Statistics for Tier 3 Scenario:
Coefficient of Determination, Modeling Efficiency, and Coefficient of Residual Mass.
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Coefficient of Determination (CD)
Acidic Circumneutral Alkaline
Modeling Efficiency (EF) Coefficient of Residual Mass (CRM)
Acidic Circumneutral Alkaline Acidic Circumneutral Alkaline
1.09
1.32
5.12
0.96
0.14
0.36
1.05
1.18
1.86
0.64
0.08
0.28
0.42
2.10
1.18
0.86
0.34
0.15
0.08
0.24
0.80
-0.04
-6.26
-1.76
0.05
0.16
0.46
-0.55
-12.12
-2.53
-1.39
0.52
0.15
-0.17
-1.90
-5.49
0.52
0.34
0.14
0.61
0.57
0.67
0.63
0.55
0.37
0.79
-0.35
0.77
0.62
0.41
-0.02
0.52
-0.10
0.84
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Coefficient of Determination (CD)
Small Medium
0.47
1.75
3.76
1.20
0.08
0.35
2.15
1.02
2.81
0.81
0.60
0.13
Modeling Efficiency (EF)
Small Medium
-1.12
0.43
0.73
0.17
-11.54
-1.83
0.53
0.01
0.64
-0.24
-0.68
-6.99
Coefficient of Residual Mass (CRM)
Small Medium
0.49
0.41
0.10
0.66
-0.94
0.75
0.81
0.60
0.38
0.67
0.43
0.80
EPI_Me
EPI_T
Fish
Sed
Coefficient of Determination (CD)
Well-
Mixed Stratified
0.68
1.23
0.04
0.26
1.74
1.57
1.07
0.29
Modeling Efficiency (EF)
Well-
Mixed Stratified
-0.48
0.19
-22.86
-2.82
0.42
0.36
0.07
-2.44
Coefficient of Residual Mass (CRM)
Well-Mixed
Stratified
0.54
0.53
-0.85
0.71
0.67
0.42
0.23
0.81
Coefficient of Determination (CD)
Oligo Meso Eutro Dystro
Oligo
Modeling Efficiency (EF)
Meso Eutro Dystro
Coefficient of Residual Mass (CRM)
Oligo Meso Eutro Dystro
0.73
1.68
1.04
1.04
0.98
0.28
0.18
0.63
0.14
0.84
0.04
0.21
0.78
0.98
2.52
0.64
0.01
0.20
1.05
0.68
1.62
0.65
0.13
0.32
-0.36
0.40
0.04
0.04
-0.02
-2.61
-4.49
-0.59
-5.96
-0.18
-24.40
-3.67
-0.28
-0.02
0.60
-0.56
-79.33
-4.00
0.05
-0.47
0.38
-0.54
-6.64
-2.16
0.46
0.05
-0.17
0.82
0.10
0.45
0.37
0.25
-1.01
0.60
-0.78
0.79
0.53
0.76
0.55
0.54
-1.84
0.93
0.81
0.60
0.47
0.63
0.49
0.85
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.21. Model Performance Statistics for Tier 4 Scenario:
Maximum Error and Root Mean Square Error.
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Acidic
Maximum Error (ME)
Circumneutral
Alkaline
1.26
5.44
3.60
26.59
0.44
0.49
1.81
6.48
2.62
27.57
1.45
0.56
0.43
3.68
0.84
12.22
0.20
0.30
Root Mean Square Error (RMSE)
Acidic Circumneutral Alkaline
102.80
91.15
125.96
107.29
80.64
84.72
122.59
93.35
125.84
99.02
164.55
83.82
76.65
92.62
15.28
77.59
76.82
85.42
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Maximum Error (ME)
Small Medium
0.76
6.48
3.60
27.57
1.45
0.56
1.81
5.68
2.41
25.07
0.44
0.36
Root Mean Square Error (RMSE)
Small Medium
90.24
95.02
122.24
100.18
213.75
85.47
150.20
92.68
135.76
109.44
78.46
83.37
Maximum Error (ME)
Well-Mixed Stratified
EPI_Me
EPI_T
Fish
Sed
1.26
6.48
1.45
0.47
1.81
5.68
0.38
0.56
Root Mean Square Error (RMSE)
Well-
Mixed Stratified
106.97
99.16
220.79
80.44
119.16
83.67
74.85
86.68
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Oliso
Maximum Error (ME)
Meso Eutro Dystro
0.43
2.09
0.49
27.57
0.30
0.28
0.60
1.56
1.81
15.33
1.45
0.47
0.76
6.48
3.60
10.75
0.20
0.36
1.81
5.68
2.62
19.36
0.67
0.56
Oliso
Root Mean Square Error (RMSE)
Meso Eutro Dystro
89.47
79.71
70.65
118.47
62.43
66.50
92.38
64.22
193.33
80.33
220.47
79.77
84.03
107.08
146.13
140.94
198.19
100.34
423.34
84.13
102.02
1,037.90
14.85
11.23
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.22. Model Performance Statistics for Tier 4 Scenario:
Coefficient of Determination, Modeling Efficiency, and Coefficient of Residual Mass.
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Coefficient of Determination (CD)
Acidic Circumneutral Alkaline
Modeling Efficiency (EF) Coefficient of Residual Mass (CRM)
Acidic Circumneutral Alkaline Acidic Circumneutral Alkaline
1.19
1.57
5.35
0.93
0.14
0.39
1.26
2.23
2.82
0.77
0.10
0.35
0.45
0.86
2.00
1.09
0.49
0.18
0.16
0.36
0.81
-0.07
-6.28
-1.54
0.21
0.55
0.65
-0.30
-8.76
-1.84
-1.22
-0.16
0.50
0.08
-1.04
-4.43
0.58
0.29
0.20
0.62
0.56
0.66
0.63
0.40
0.24
0.72
-0.38
0.70
0.52
0.04
-0.04
0.45
-0.25
0.73
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Coefficient of Determination (CD)
Small Medium
0.62
2.23
5.43
1.24
0.10
0.43
2.24
1.51
2.45
0.99
0.83
0.15
Modeling Efficiency (EF)
Small Medium
-0.60
0.55
0.82
0.19
-8.56
-1.32
0.55
0.34
0.59
-0.01
-0.21
-5.82
Coefficient of Residual Mass (CRM)
Small Medium
0.49
0.20
0.15
0.65
-0.98
0.68
0.80
0.46
0.19
0.58
0.38
0.73
Coefficient of Determination (CD)
Well-Mixed Stratified
EPI_Me
EPI_T
Fish
Sed
0.87
1.65
0.06
0.34
1.87
2.19
1.49
0.33
Modeling Efficiency (EF)
Well-Mixed Stratified
-0.14
0.39
-16.81
-1.93
0.46
0.54
0.33
-2.01
Coefficient of Residual Mass (CRM)
Well-Mixed Stratified
0.53
0.34
-0.87
0.64
0.66
0.25
0.17
0.74
Coefficient of Determination (CD)
Oligo Meso Eutro Dystro
Oligo
Modeling Efficiency (EF)
Meso Eutro Dystro
Coefficient of Residual Mass (CRM)
Oligo Meso Eutro Dystro
0.73
0.62
0.95
1.07
1.13
0.38
0.28
1.07
0.17
1.19
0.05
0.27
0.92 1.14 -0.36 -2.54 -0.08 0.12 0.45 0.35 0.58 0.79
1.06 1.17 -0.61 0.07 0.05 0.15 -0.17 -0.07 0.74 0.42
2.58 1.94 -0.06 -4.80 0.61 0.49 -0.10 -1.25 0.55 0.44
0.66 0.73 0.06 0.16 -0.51 -0.37 0.81 0.50 0.53 0.59
0.02 0.17 0.11 -18.03 -53.59 -5.05 0.07 -0.85 -1.50 0.42
0.20 0.37 -1.60 -2.64 -3.88 -1.68 0.41 0.69 0.92 0.78
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.23. Model Performance Statistics for Tier 5 Scenario:
Maximum Error and Root Mean Square Error.
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Acidic
Maximum Error (ME)
Circumneutral
Alkaline
1.24
5.40
3.64
27.96
0.38
0.55
1.80
6.48
2.98
28.31
1.07
0.60
0.39
3.30
1.29
12.39
0.19
0.31
Root Mean Square Error (RMSE)
Acidic Circumneutral Alkaline
106.30
87.95
132.71
110.81
78.59
93.00
120.44
95.83
124.13
103.61
139.59
88.88
84.84
78.39
207.59
85.51
73.31
94.26
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Maximum Error (ME)
Small Medium
0.76
6.48
3.64
28.31
1.07
0.60
1.80
5.38
2.66
26.04
0.69
0.36
Root Mean Square Error (RMSE)
Small Medium
91.42
94.25
129.87
108.23
160.55
93.63
149.74
90.21
125.52
108.94
96.23
87.08
Maximum Error (ME)
Well-Mixed Stratified
EPI_Me
EPI_T
Fish
Sed
1.24
6.48
1.07
0.49
1.80
5.38
0.51
0.60
Root Mean Square Error (RMSE)
Well-Mixed Stratified
101.51
92.75
175.80
82.76
126.58
90.23
83.94
95.99
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Oliso
Maximum Error (ME)
Meso Eutro Dystro
0.47
2.11
0.43
28.31
0.38
0.31
0.52
2.06
0.84
16.02
1.07
0.49
0.76
6.48
3.64
11.50
0.15
0.36
1.80
5.38
3.59
22.68
0.51
0.60
Oliso
Root Mean Square Error (RMSE)
Meso Eutro Dystro
90.96
61.52
57.69
123.03
70.25
77.26
90.69
71.91
106.37
87.19
185.71
87.24
83.40
106.13
181.54
172.33
141.78
102.29
461.27
83.82
176.86
1,324.20
13.36
12.08
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.24. Model Performance Statistics for Tier 5 Scenario:
Coefficient of Determination, Modeling Efficiency, and Coefficient of Residual Mass
Coefficient of Determination (CD)
Modeling Efficiency (EF)
Alkaline
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
EPI Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
EPI Me
EPI_T
Fish
Sed
Acidic
0.89
1.49
4.05
0.72
0.13
0.35
Coefficient
Small
0.51
1.94
3.86
0.86
0.19
0.36
Coefficient
Well-
Mixed
0.92
2.09
0.09
0.34
Circumneutral
1.24
1.87
2.41
0.64
0.16
0.31
Alkaline Acidic
0.32
1.07
2.20
0.59
0.89
0.15
-0.13
0.33
0.75
-0.39
-6.89
-1.90
Circumneutral
0.19
0.47
0.59
-0.56
-5.34
-2.17
of Determination (CD) Modeling Efficiency (EF)
Medium
2.09
1.38
2.75
0.87
0.40
0.13
of Determination (CD)
Stratified
1.31
1.48
1.08
0.26
Small
-0.97
0.48
0.74
-0.16
-4.23
-1.79
Medium
0.52
0.28
0.64
-0.15
-1.48
-6.41
Modeling Efficiency (EF)
Well-
Mixed
-0.08
0.52
-9.90
-1.93
Stratified
0.24
0.33
0.08
-2.79
Coefficient of Determination (CD) J
EPI_Me
EPI_T
HYP_Me
HYP_T
Fish
Sed
Oligo
0.66
0.84
0.90
0.86
0.89
0.28
Meso
0.28
1.26
0.68
0.75
0.08
0.22
Eutro Dystro
0.80 1.03
1.00 1.11
1.72 1.42
0.46 0.61
0.03 0.22
0.20 0.34
Oligo
-0.51
-0.18
-0.11
-0.16
-0.12
-2.62
Coefficient of Residual Mass (CRM)
Acidic Circumneutral Alkaline
Coefficient of Residual Mass (CRM)
Small Medium
0.65
0.43
0.57
0.79
-0.38
0.78
0.81
0.53
0.53
0.72
0.45
0.78
Coefficient of Residual Mass (CRM)
Well-
Mixed Stratified
0.60
0.38
-0.76
0.68
0.83
0.56
0.51
0.86
Modeling Efficiency (EF)
Meso Eutro Dystro
Coefficient of Residual Mass (CRM)
Oligo Meso Eutro Dystro
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 5.25. Formulas for Statistical Evaluation Parameters.
t =
d =
n
d2
n-l n-l
P(t
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 6.1. Hypolimnion Surface Area Sensitivity Analysis Results.
EPI_MeHg
EPI_HgT
HYP_MeHg
HYP_HgT
Sed_MeHg
Sed_HgT
Fish Tissue
Average Change
from 1/5 to l/2a
Percent Absolute
35.83% 0.026
13.25% 0.105
36.23% 0.209
19.73% 0.581
25.55% 0.001
5.12% 0.003
39.09% 0.060
Max Change from
1/5 to 1/2 a
Percent Absolute
66.67% 0.120
32.35% 0.330
65.51% 0.891
58.28% 2.918
100.00% 0.011
50.00% 0.020
60.00% 0.190
Average A(C,/J) b
A(C,l/2) A(C,l/4) A(C,l/5) Average
-42.90% -82.30% -75.72% -66.98%
-14.36% -17.52% -18.38% -16.75%
-42.99% -59.71% -63.43% -55.38%
-22.88% -27.62% -30.26% -26.92%
-26.15% -60.34% -67.80% -51.43%
-5.07% -6.68% -10.42% -7.39%
-50.43% -60.94% -62.13% -57.84%
Notes: a Change from 1/5 to 1/2 calculated as: Absolute Change = C1/5 C1/2, Percent Change =100
where: C# is the predicted concentration for R, where R = 1/5 or 1/2.
' C -C
Vx CT> Vx 1
r -C
*-s-\ / < ^1
'1/5
C
1/5
C
1/3
1/3
where:
Sensitivity:
C
R
CIK
CSR
1/3
R8
1
1
I) =
=
=
=
=
=
=
Extra Strong 1
(>100%) |
the sensitivity of model output C to parameter R [percent]
model output [mercury concentration]
parameter being varied [hypolimnion surface area/epilimnion surface
calculated model output value for base case (i.e., R = 1/3)
calculated model output value for change in parameter, R
model parameter value for base case (i.e., R = 1/3)
model parameter value in sensitivity simulation
Strong Moderate Weak
(50% - 99%) (25%-49%) (<25%)
area]
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 6.2. Polynomial Linear Regression Results. Coefficients, Standard Errors, Adjusted R2, and F Significance.
EPI_MeHg
EPI_HgT
HYP_MeHg
HYP_HgT
Sed_MeHg
Sed_HgT
Fish Tissue
Po
0.0030
-0.0061
0.042
0.75
0.0013
0.0022
0.0015
Std
Error
0.0058
0.018
0.039
0.30
0.0013
0.0027
0.0070
Pi
-0.043
-0.081
-0.50
-3.3
-0.0043
-0.011
-0.056
Std
Error
0.034
0.10
0.23
1.8
0.0079
0.016
0.040
P2
0.11
0.26
0.97
2.2
0.0071
0.0098
0.15
Std
Error
0.046
0.14
0.32
2.4
0.011
0.023
0.055
P3
1.9
1.2
1.8
1.1
1.1
1.06
1.76
Std
Error
0.072
0.020
0.052
0.050
0.14
0.018
0.037
P4
1.2
-0.0016
-0.0032
-0.0055
25.0
-0.086
0.21
Std
Error
0.50
0.010
0.041
0.0030
6.0
0.068
0.11
Ps
-2.9
-0.68
-2.36
-0.051
-1.8
-0.13
-2.3
Std
Error
0.13
0.037
0.098
0.11
0.31
0.041
0.068
Adj
R2
0.97
0.997
0.98
0.97
0.82
0.996
0.994
F
Significance
1.2E-144
1.3E-238
2.1E-148
1.1E-129
8.2E-58
1.6E-229
7.6E-143
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 6.3. Hypolimnion Area Sensitivity Evaluation. Hypothetical Default Lake Conditions and Results:
Changing R and DH, Constant Q.
Hypolimnion Area
Sensitivity
Hydraulic Res Time
pH_EPI/HYP
DOC_EPI/HYP
Lake Area
Hypo Area
EPI_Depth
HYP_Depth
Lake Volume
Vol Flow Rate
Epilimnion
Hypolimnion
Sediment
(Hypolimnion)
Units
days
mg/L
m2
m2
m
m
m3
m3/day
MeHg
Hgll
HgT
MeHg
Hgll
HgT
MeHg
Hgll
HgT
Acidity
Acidic
Base Case
365
5.3
3
1000000
250000
8
5
9250000
25,342
1
0.08
1.217
1.318
0.69
1.192
2.291
0.011
0.144
0.154
Stratification
Stratified
Run 2
493
5.3
3
1000000
900000
8
5
12500000
25,342
2
0.028
0.834
0.878
0.207
0.816
1.215
0.003
0.098
0.101
Size
Medium
Run 3
335
5.3
3
1000000
100000
8
5
8500000
25,342
3
0.109
1.362
1.493
0.949
1.334
2.801
0.015
0.161
0.175
Hydrology
Drainage
Run 4
317
5.3
3
1000000
250000
8
0.1
8025000
25,342
4
0.081
1.218
1.319
0.709
1.205
2.229
0.011
0.145
0.156
Trophic
Status
Oligotrophic
Run5
10181
5.3
3
1000000
250000
8
1000
258000000
25,342
5
0.081
1.184
1.308
0.777
0.372
6.049
0.011
0.045
0.056
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 6.4. Hypolimnion Area Sensitivity Evaluation. Hypothetical Default Lake Conditions and Results: Changing R, Constant Q.
Hypolimnion Area
Sensitivity
Hydraulic Res Time
pH_EPI/HYP
DOC_EPI/HYP
Lake Area
Hypo Area
EPI_Depth
HYP_Depth
Lake Volume
Vol Flow Rate
Raw Data
Epilimnion
Hypolimnion
Sediment (Hypolimnion)
Units
days
mg/L
m2
m2
m
m
m3
m3/day
R
MeHg
Hgll
HgT
MeHg
Hgll
HgT
MeHg
Hgll
HgT
Acidity
Acidic
Run 1-1
385
5.3
3
1,000,000
950,000
5
5
9,750,000
25,342
1-1
0.95
0.025
0.818
0.859
0.16
0.802
1.17
0.002
0.097
0.099
Stratification
Stratified
Run 1-2
385
5.3
3
1,000,000
0
9.75
0
9,750,000
25,342
1-2
0
0.135
1.471
1.631
N/A
N/A
N/A
N/A
N/A
N/A
Size
Medium
Run 1-3
217
5.3
3
1,000,000
100,000
5
5
5,500,000
25,342
1-3
0.1
0.109
1.375
1.506
0.849
1.347
2.862
0.014
0.162
0.176
Hydrology
Drainage
Run 1-4
247
5.3
3
1,000,000
250,000
5
5
6,250,000
25,342
1-4
0.25
0.079
1.228
1.326
0.608
1.202
2.321
0.01
0.145
0.155
Trophic
Status
Oligotrophic
Run 1-5
296
5.3
3
1,000,000
500,000
5
5
7,500,000
25,342
1-5
0.5
0.049
1.042
1.108
0.368
1.02
1.739
0.006
0.123
0.129
Run 1-6
207
5.3
3
1,000,000
50,000
5
5
5,250,000
25,342
1-6
0.05
0.123
1.432
1.577
0.955
1.403
3.093
0.016
0.169
0.184
Run 1-7
197
5.3
3
1,000,000
0
5
0
5,000,000
25,342
1-7
0
0.25
2.989
3.273
N/A
N/A
N/A
N/A
N/A
N/A
Run 1-8
197
5.3
3
1,000,000
100
5
5
5,000,500
25,342
1-8
0.0001
0.138
1.494
1.656
1.079
1.464
3.357
0.018
0.176
0.194
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 6.5. Results for Mathematical Analysis of Simple, Stratified Lake System.
Run (a) R Hypolimnion Area [m2] Epilimnion CT [ng/L] Hypolimnion CT [ng/L]
1 1,000,000 0.167 0.330
0.9 900,000 0.167 0.330
0.8 800,000 0.167 0.330
0.7 700,000 0.167 0.329
0.6 600,000 0.167 0.328
0.5 500,000 0.167 0.327
0.4 400,000 0.167 0.325
0.3 300,000 0.167 0.323
0.2 200,000 0.167 0.317
0.1 100,000 0.166 0.304
0.05 50,000 0.166 0.283
0.01 10,000 0.164 0.219
0 0 0.163 0.167
Run (b) R Hypolimnion Area [m2] Epilimnion CT [ng/L] Hypolimnion CT [ng/L]
1 1,000,000 0.315 0.162
0.9 900,000 0.315 0.162
0.8 800,000 0.315 0.162
0.7 700,000 0.315 0.162
0.6 600,000 0.315 0.163
0.5 500,000 0.315 0.163
0.4 400,000 0.315 0.164
0.3 300,000 0.315 0.166
0.2 200,000 0.315 0.168
0.1 100,000 0.315 0.176
0.05 50,000 0.316 0.189
0.01 10,000 0.318 0.244
0 0 0.322 0.320
Run(c) R H;
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.05
0.01
0
mnion Area [m ]
1,000,000
900,000
800,000
700,000
600,000
500,000
400,000
300,000
200,000
100,000
50,000
10,000
0
Epilimnion CT [ng/L]
4.131
4.166
4.207
4.258
4.321
4.402
4.511
4.664
4.895
5.284
5.597
5.956
6.069
Hypolimnion CT [ng/L]
2.412
2.476
2.554
2.648
2.767
2.920
3.125
3.413
3.846
4.575
5.163
5.839
6.051
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 6.6. Comparison of Summary Statistics for Percent Methylmercury: Predicted and
Observed Results.
Minimum
Maximum
Mean
Median
Std Dev.
Range
Percent MeHg in
Epilimnion
Predicted
0
43
8.6
6.3
7.5
43
Observed
2.8
75
21
18
14
72
Percent MeHg in
Hypolimnion
Predicted
0.15
35
17
16
8.8
35
Observed
0.72
39
9.2
6.2
8.0
38
Percent MeHg in
Sediments
Predicted
0
40
5.7
4
5.8
40
Observed
0.22
29
2.4
1.5
3.6
29
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table 6.7. Minima for Error Sum of Squares and Associated Standard Deviations and
Their Associated RlUp and R2Up Values.
EPI MeHg
EPI HgT
HYP MeHg
HYP HgT
Fish Tissue
Sediment HgT
RlUp
1.7
4.2
1.8
1
0.1
0.1
R2Up
0.1
0.1
0.1
1.19
0.1
0.72
Minimum
Sum of Squares
9.03
211
48.1
3200
3.59
3.21
Minimum
Standard Deviation
0.315
1.54
1.06
8.65
0.276
0.194
After Removal of Outliers for EPI MeHg, EPI HgT, and HYP MeHg:
EPI MeHg
EPI HgT
HYP MeHg
RlUp
4.7
4.6
1.4
R2Up
0.1
0.1
0.1
Minimum
Sum of Squares
2.96
206
9.67
Minimum
Standard Deviation
0.184
1.52
0.499
Investigating EPI MeHg and HgT.
For MeHg, Removed 3 outlier lakes:
Lake No. 54, Mitchell Lake, was removed because the predicted values greatly
exceeded the other lakes.
Lakes No. 69 and 70, Powwow and Robb Reservoir, were removed because they
had observed values greatly exceeding the other lakes.
For HgT, Removed 1 outlier lake:
Lake No. 54, Mitchell Lake, was removed because the predicted values greatly
exceeded the other lakes.
HYP_MeHg:
Removed 4 outlier lakes because all had observed values greatly exceeding the other
lakes: Lakes No. 6 (Branch), 59 (Notch), 64 (Pawtuckaway Lake), and 77 (Spruce
Pond).
-------
Evaluating R-MCMfor 91 VT/NH Lakes
APPENDIX
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-l. Observed Results from the VT/NH REMAP Study.
Lake Name
ADDER POND
ARMINGTON LAKE
BAKER (BARTON)
BAKER POND- UPPER
BEARCAMP POND
BRANCH
BRUCE
CAWLEY POND
CHASE POND
CHILDS BOG
CHITTENDEN
CLUB POND
CRANBERRY
MEADOW
CURTIS
DENNIS
DUNMORE
DUTCHMAN POND
EASTMAN POND
ECHO (CHARTN)
ECHO (HUBDTN)
ELM BROOK POOL
FERN
FISH POND
FREESES POND- UPPER
GILES POND
GREAT HOSMER
GREENWOOD POND
HALL POND- UPPER
HARDWICK
HARDWOOD
HIGH (SUDBRY)
HILDRETH DAM POND
HORN POND
HORTONIA
HOWE RESERVOIR
ISLAND POND
IVANHOE- LAKE
JACKSONVILLE
JENNESS POND
Epilimnion
MeHg
[ng/L]
0.289
0.218
0.096
0.290
0.214
0.490
0.472
0.541
0.317
0.390
0.113
0.184
0.215
0.175
0.101
0.098
0.299
0.194
0.072
0.188
0.100
0.231
0.363
0.787
0.038
0.123
0.243
0.299
0.113
0.249
0.175
0.349
0.141
0.166
0.241
0.295
0.209
0.588
0.475
HgT
[ng/L]
2.143
0.908
0.940
1.007
1.124
1.776
1.520
0.717
3.358
0.692
1.142
0.957
1.118
0.648
1.398
0.507
1.790
1.639
0.594
0.352
0.587
1.035
2.680
6.910
0.511
0.679
0.600
1.890
4.040
1.435
1.720
1.960
1.506
0.656
1.168
1.578
1.790
2.410
3.101
Hypolimnion
MeHg
[ng/L]
1.004
1.649
1.753
4.447
0.938
0.670
0.388
0.142
0.263
0.374
0.394
0.273
1.415
1.005
0.163
0.127
0.379
HgT
[ng/L]
15.191
8.830
8.503
25.843
13.430
2.569
4.685
2.094
2.727
5.625
9.763
7.880
8.522
8.775
5.580
1.266
2.421
Fish
HgT
[ug/g]
0.231
0.209
0.321
0.389
0.210
0.074
0.046
0.111
0.276
0.117
0.261
0.163
0.267
0.049
0.085
0.417
0.215
0.132
0.386
0.382
0.255
0.215
Sediment
HgT
[ug/g]
0.166
0.296
0.090
0.255
0.290
0.480
0.170
0.210
0.457
0.263
0.170
0.180
0.175
0.164
0.190
0.270
0.110
0.230
0.174
0.282
0.110
0.283
0.210
0.090
0.160
0.200
0.278
0.390
0.072
0.200
0.315
0.195
0.166
0.268
0.220
0.363
0.140
0.187
0.195
-------
Evaluating R-MCMfor 91 VT/NH Lakes
KENT
LARY POND
LEFFERTS
LILY POND
LITTLE AVERILL
LONG (WESTMR)
LOON LAKE
LOVELL LAKE- STN 1
LYFORD
MANSFIELD
MCCONNELL
MILLSFffiLD POND
MILTON
MINARDS
MITCHELL
MOOSE POND
MOUNTAIN LAKE-
UPPER
NEWARK
NORTH (BRKFLD)
NOTCH
NOYES
PARAN
PARKER
PAUGUS BAY- STN 1
PAWTUCKAWAY
LAKE
PEMIGEWASSET LAKE
PERCH (BENSON)
PLEASANT VALLEY
POUT POND
POWWOW POND
ROBB RESERVOIR
ROUND POND
SABIN
SHAWS POND
SILVER LAKE
SOMERSET
SOUTH AMERICA
SPRUCE POND
STRATTON
Epilimnion
MeHg
0.177
0.298
0.260
0.510
0.201
0.067
0.520
0.318
0.305
0.243
0.180
0.336
0.142
0.230
0.264
0.665
0.095
0.165
0.136
0.329
0.269
0.231
0.217
0.141
0.187
0.456
0.162
0.262
0.665
1.813
1.272
0.287
0.223
0.177
0.077
0.116
0.695
0.327
0.073
HgT
1.370
4.010
1.868
6.380
0.501
0.817
0.861
0.503
0.946
0.950
1.962
2.152
1.214
2.422
4.115
3.840
0.396
1.066
0.846
5.296
2.581
1.242
1.741
3.120
2.257
2.057
0.586
0.845
2.140
5.872
2.900
0.690
1.216
0.577
2.224
1.885
3.140
1.456
1.090
Hypolimnion
MeHg
0.353
0.288
0.552
0.277
0.205
0.448
0.584
3.186
0.156
0.051
2.908
1.213
0.573
0.822
0.433
0.326
0.195
0.165
4.454
HgT
6.678
2.015
29.786
4.711
7.992
3.784
13.346
13.660
4.911
5.210
20.741
8.703
9.468
4.540
29.604
6.540
27.134
6.402
11.518
Fish
HgT
0.115
0.189
0.117
0.156
0.087
0.076
0.225
0.233
0.204
0.331
0.190
0.308
0.468
0.233
0.128
0.462
0.463
Sediment
HgT
0.180
0.220
0.220
0.280
0.240
0.140
0.140
0.190
0.190
0.290
0.280
0.170
0.244
0.237
0.185
0.190
0.170
0.256
0.142
0.622
0.270
0.133
0.110
0.210
0.290
0.163
0.300
0.310
0.190
0.175
0.188
0.370
0.135
0.170
0.237
0.256
0.125
0.370
0.310
-------
Evaluating R-MCMfor 91 VT/NH Lakes
SUNCOOK POND-
UPPER
SUNRISE LAKE
SUNSET (BRKFLD)
TRIO PONDS- ONE
AND TWO
TUTTLE (HARDWK)
UNNAMED POND
WALKER POND
WILLEY POND- BIG
WILLEY POND- LITTLE
WILLOUGHBY
WILSON POND
WOLCOTT
ZEPHYR LAKE
Epilimnion
MeHg
0.107
0.266
0.232
0.300
0.502
0.408
0.155
0.169
0.386
0.183
0.218
0.322
0.213
HgT
1.380
0.788
0.451
2.267
2.930
1.027
0.932
0.867
0.712
1.116
0.835
1.132
0.899
Hypolimnion
MeHg
0.733
0.344
0.178
0.892
0.265
0.303
0.221
HgT
12.303
10.277
4.750
14.282
5.267
5.659
19.048
Fish
HgT
0.331
0.156
0.156
0.172
0.068
0.100
0.696
0.112
Sediment
HgT
0.257
0.229
0.190
0.608
0.180
0.128
0.225
0.380
0.230
0.128
0.550
0.215
0.140
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-2. Predicted Results for the Default Run
Lake Name
ADDER POND
ARMINGTON LAKE
BAKER (BARTON)
BAKER POND- UPPER
BEARCAMP POND
BRANCH
BRUCE
CAWLEY POND
CHASE POND
CHILDS BOG
CHITTENDEN
CLUB POND
CRANBERRY
MEADOW
CURTIS
DENNIS
DUNMORE
DUTCHMAN POND
EASTMAN POND
ECHO (CHARTN)
ECHO (HUBDTN)
ELM BROOK POOL
FERN
FISH POND
FREESES POND- UPPER
GILES POND
GREAT HOSMER
GREENWOOD POND
HALL POND- UPPER
HARDWICK
HARDWOOD
HIGH (SUDBRY)
HILDRETH DAM POND
HORN POND
HORTONIA
HOWE RESERVOIR
ISLAND POND
IVANHOE- LAKE
JACKSONVILLE
JENNESS POND
Epilimnion
MeHg
[ng/L]
0.240
0.070
0.180
0.240
0.190
0.490
0.780
0.570
0.780
0.570
0.140
0.840
0.130
0.180
0.260
0.070
0.370
0.150
0.030
0.130
0.240
0.060
0.350
0.600
0.290
0.090
0.230
0.490
0.160
0.780
0.180
0.290
0.070
0.060
0.280
0.190
0.110
0.780
0.070
HgT
[ng/L]
1.190
1.110
1.060
1.640
1.970
4.160
5.070
2.770
5.070
2.770
1.280
5.390
1.130
1.060
2.330
1.110
2.910
0.940
0.460
1.130
1.640
0.670
2.130
1.830
2.510
0.640
1.240
4.160
0.770
5.070
1.060
1.980
1.110
0.560
2.520
2.100
1.390
5.070
1.110
Hypolimnion
MeHg
[ng/L]
1.556
1.180
0.609
1.817
0.963
1.180
1.044
0.197
0.769
0.152
0.558
1.818
1.180
1.938
0.572
0.318
0.727
HgT
[ng/L]
5.458
5.267
2.528
5.666
3.007
5.267
3.807
1.122
3.319
1.201
3.080
5.667
5.267
5.100
2.459
1.636
2.714
Fish
HgT
[ug/g]
0.190
0.260
0.110
0.390
0.170
0.650
0.760
0.920
0.760
0.920
0.220
1.200
0.180
0.110
0.250
0.260
1.970
0.100
0.090
0.180
0.390
0.140
0.320
0.400
1.130
0.050
0.640
0.650
0.100
0.760
0.110
0.450
0.260
0.080
0.410
0.260
0.440
0.760
0.260
Sediment
HgT
[ug/g]
0.020
0.110
0.010
0.070
0.040
0.090
0.110
0.100
0.110
0.100
0.050
0.120
0.030
0.010
0.050
0.110
0.300
0.020
0.020
0.030
0.070
0.020
0.040
0.030
0.240
0.010
0.010
0.090
0.010
0.110
0.010
0.080
0.110
0.010
0.060
0.050
0.140
0.110
0.110
-------
Evaluating R-MCMfor 91 VT/NH Lakes
KENT
LARY POND
LEFFERTS
LILY POND
LITTLE AVERILL
LONG (WESTMR)
LOON LAKE
LOVELL LAKE- STN 1
LYFORD
MANSFIELD
MCCONNELL
MILLSFffiLD POND
MILTON
MINARDS
MITCHELL
MOOSE POND
MOUNTAIN LAKE-
UPPER
NEWARK
NORTH (BRKFLD)
NOTCH
NOYES
PARAN
PARKER
PAUGUS BAY- STN 1
PAWTUCKAWAY
LAKE
PEMIGEWASSET LAKE
PERCH (BENSON)
PLEASANT VALLEY
POUT POND
POWWOW POND
ROBB RESERVOIR
ROUND POND
SABIN
SHAWS POND
SILVER LAKE
SOMERSET
SOUTH AMERICA
SPRUCE POND
STRATTON
Epilimnion
MeHg
0.210
0.600
0.570
0.760
0.030
0.050
0.140
0.140
0.130
0.490
0.480
0.260
0.570
0.290
0.090
0.780
0.780
0.030
0.130
0.480
0.780
0.210
0.060
0.140
0.190
0.190
0.130
0.210
0.600
0.170
0.280
0.130
0.060
0.090
0.080
0.150
0.780
0.240
0.570
HgT
1.300
1.830
2.770
1.980
0.460
0.920
1.910
1.280
1.130
4.160
3.970
2.330
2.770
1.980
1.020
5.070
5.070
0.460
1.130
3.970
5.070
1.300
0.560
1.280
1.970
1.970
1.130
1.300
1.830
1.950
2.520
1.770
0.560
1.020
1.320
1.390
5.070
1.190
2.770
Hypolimnion
MeHg
0.197
0.270
0.897
0.963
0.769
1.938
0.197
1.567
0.318
0.963
0.609
0.609
0.769
0.200
0.794
0.318
0.690
1.083
1.556
HgT
1.122
2.235
3.706
3.007
3.319
5.099
1.122
5.286
1.636
3.007
2.538
2.538
3.319
1.368
3.900
1.636
2.291
3.040
5.458
Fish
HgT
0.310
0.400
0.920
0.630
0.090
0.180
0.720
0.220
0.180
0.650
0.460
0.250
0.920
0.450
0.300
0.760
0.760
0.090
0.180
0.460
0.760
0.310
0.080
0.220
0.170
0.170
0.180
0.310
0.400
0.140
0.410
0.510
0.080
0.300
0.410
0.330
0.760
0.190
0.920
Sediment
HgT
0.030
0.030
0.100
0.030
0.020
0.040
0.210
0.050
0.030
0.090
0.090
0.050
0.100
0.080
0.040
0.110
0.110
0.020
0.030
0.090
0.110
0.030
0.010
0.050
0.040
0.040
0.030
0.030
0.030
0.040
0.060
0.170
0.010
0.040
0.150
0.060
0.110
0.020
0.100
-------
Evaluating R-MCMfor 91 VT/NH Lakes
SUNCOOK POND-
UPPER
SUNRISE LAKE
SUNSET (BRKFLD)
TRIO PONDS- ONE
AND TWO
TUTTLE (HARDWK)
UNNAMED POND
WALKER POND
WILLEY POND- BIG
WILLEY POND- LITTLE
WILLOUGHBY
WILSON POND
WOLCOTT
ZEPHYR LAKE
Epilimnion
MeHg
0.190
0.240
0.130
0.490
0.220
0.840
0.190
0.370
0.090
0.030
0.570
0.480
0.290
HgT
1.970
1.640
1.130
4.160
1.230
5.390
1.970
2.910
1.020
0.460
2.770
3.970
1.980
Hypolimnion
MeHg
0.609
0.769
1.817
0.609
0.197
1.567
1.938
HgT
2.538
3.319
5.666
2.538
1.133
5.385
5.099
Fish
HgT
0.170
0.390
0.180
0.650
0.110
1.200
0.170
1.970
0.300
0.090
0.920
0.460
0.450
Sediment
HgT
0.040
0.070
0.030
0.090
0.010
0.120
0.040
0.300
0.040
0.020
0.100
0.090
0.080
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-3. Predicted Results for Tier 1.
Lake Name
ADDER POND
ARMINGTON LAKE
BAKER (BARTON)
BAKER POND- UPPER
BEARCAMP POND
BRANCH
BRUCE
CAWLEY POND
CHASE POND
CHILDS BOG
CHITTENDEN
CLUB POND
CRANBERRY
MEADOW
CURTIS
DENNIS
DUNMORE
DUTCHMAN POND
EASTMAN POND
ECHO (CHARTN)
ECHO (HUBDTN)
ELM BROOK POOL
FERN
FISH POND
FREESES POND- UPPER
GILES POND
GREAT HOSMER
GREENWOOD POND
HALL POND- UPPER
HARDWICK
HARDWOOD
HIGH (SUDBRY)
HILDRETH DAM POND
HORN POND
HORTONIA
HOWE RESERVOIR
ISLAND POND
IVANHOE- LAKE
JACKSONVILLE
JENNESS POND
Epilimnion
MeHg
[ng/L]
0.120
0.040
0.090
0.020
0.010
0.160
0.080
0.000
0.000
0.560
0.070
0.020
0.020
0.080
0.000
0.050
0.280
0.040
0.040
0.100
0.000
0.030
0.020
0.000
0.000
0.080
0.210
0.260
0.000
0.420
0.100
0.020
0.010
0.060
0.030
0.000
0.210
0.000
0.050
HgT
[ng/L]
0.670
0.830
0.660
0.340
0.330
1.710
0.910
0.100
0.120
1.870
0.680
0.400
0.330
0.530
0.130
0.710
2.580
0.350
0.590
0.920
0.030
0.520
0.390
0.070
0.080
0.600
0.830
1.920
0.040
2.470
0.600
0.290
0.270
0.540
0.460
0.030
2.190
0.090
0.880
Hypolimnion
MeHg
[ng/L]
0.773
0.554
0.126
1.254
0.669
0.425
0.293
0.260
0.739
0.089
0.326
1.870
0.541
0.179
0.080
0.402
0.002
HgT
[ng/L]
3.635
3.115
1.035
3.717
2.687
3.819
2.197
1.806
3.631
0.835
3.362
4.391
4.595
1.076
0.861
2.159
18.728
Fish
HgT
[ug/g]
0.110
0.180
0.070
0.070
0.040
0.700
0.190
0.020
0.020
1.560
0.170
0.070
0.060
0.090
0.010
0.160
1.200
0.050
0.120
0.210
0.000
0.120
0.040
0.010
0.010
0.070
0.350
0.680
0.000
1.050
0.110
0.050
0.040
0.110
0.090
0.000
0.920
0.010
0.200
Sediment
HgT
[ug/g]
0.020
0.090
0.010
0.020
0.020
0.090
0.040
0.010
0.010
0.120
0.040
0.020
0.020
0.020
0.010
0.060
0.240
0.010
0.040
0.020
0.000
0.020
0.010
0.000
0.010
0.010
0.020
0.100
0.000
0.120
0.020
0.020
0.020
0.020
0.020
0.000
0.240
0.000
0.100
-------
Evaluating R-MCMfor 91 VT/NH Lakes
KENT
LARY POND
LEFFERTS
LILY POND
LITTLE AVERILL
LONG (WESTMR)
LOON LAKE
LOVELL LAKE- STN 1
LYFORD
MANSFIELD
MCCONNELL
MILLSFffiLD POND
MILTON
MINARDS
MITCHELL
MOOSE POND
MOUNTAIN LAKE-
UPPER
NEWARK
NORTH (BRKFLD)
NOTCH
NOYES
PARAN
PARKER
PAUGUS BAY- STN 1
PAWTUCKAWAY
LAKE
PEMIGEWASSET LAKE
PERCH (BENSON)
PLEASANT VALLEY
POUT POND
POWWOW POND
ROBB RESERVOIR
ROUND POND
SABIN
SHAWS POND
SILVER LAKE
SOMERSET
SOUTH AMERICA
SPRUCE POND
STRATTON
Epilimnion
MeHg
0.050
0.030
0.010
0.560
0.070
0.070
0.080
0.130
0.070
0.030
0.010
0.010
0.270
0.190
0.080
0.030
0.030
0.040
0.020
0.060
0.010
0.010
0.030
0.010
0.060
0.050
0.180
0.120
0.410
0.000
0.000
0.070
0.020
0.140
0.050
0.080
0.040
0.160
0.590
HgT
0.540
0.250
0.200
1.340
1.350
1.020
1.400
0.920
0.790
0.570
0.330
0.310
1.320
1.090
0.990
0.510
0.430
0.530
0.350
0.960
0.250
0.210
0.430
0.230
0.830
0.690
1.180
0.730
1.000
0.060
0.110
1.310
0.280
1.530
0.660
0.790
0.670
0.830
2.200
Hypolimnion
MeHg
0.270
0.404
0.464
1.259
0.514
1.697
0.237
0.316
0.241
0.164
0.386
0.364
1.478
0.426
0.145
0.411
0.725
1.217
HgT
2.326
3.297
2.271
3.958
2.770
5.182
1.395
1.350
1.607
1.098
1.710
1.541
5.209
0.024
2.573
1.457
2.289
2.193
4.326
Fish
HgT
0.140
0.040
0.050
0.540
0.120
0.280
0.330
0.280
0.140
0.120
0.030
0.040
0.600
0.500
0.270
0.070
0.080
0.110
0.050
0.100
0.030
0.030
0.070
0.050
0.120
0.120
0.390
0.370
0.430
0.000
0.010
0.270
0.040
0.570
0.200
0.210
0.170
0.160
1.410
Sediment
HgT
0.020
0.010
0.010
0.030
0.060
0.070
0.140
0.060
0.020
0.030
0.010
0.020
0.060
0.080
0.040
0.020
0.020
0.020
0.010
0.030
0.010
0.010
0.010
0.020
0.040
0.040
0.060
0.040
0.030
0.000
0.000
0.120
0.010
0.140
0.050
0.060
0.030
0.020
0.120
-------
Evaluating R-MCMfor 91 VT/NH Lakes
SUNCOOK POND-
UPPER
SUNRISE LAKE
SUNSET (BRKFLD)
TRIO PONDS- ONE
AND TWO
TUTTLE (HARDWK)
UNNAMED POND
WALKER POND
WILLEY POND- BIG
WILLEY POND- LITTLE
WILLOUGHBY
WILSON POND
WOLCOTT
ZEPHYR LAKE
Epilimnion
MeHg
0.010
0.150
0.000
0.190
0.070
0.030
0.040
0.370
0.100
0.050
0.260
0.140
0.110
HgT
0.250
1.040
0.120
1.830
0.430
0.510
0.590
1.190
0.880
0.570
1.310
1.440
0.900
Hypolimnion
MeHg
0.080
0.058
1.289
0.284
0.355
0.951
1.057
HgT
1.067
1.375
3.443
1.458
1.805
2.869
2.988
Fish
HgT
0.030
0.410
0.020
0.510
0.050
0.170
0.090
3.990
0.580
0.170
0.640
0.310
0.270
Sediment
HgT
0.010
0.060
0.010
0.080
0.010
0.020
0.030
0.320
0.040
0.030
0.080
0.070
0.060
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-4. Predicted Results for Tier 2.
Lake Name
ADDER POND
ARMINGTON LAKE
BAKER (BARTON)
BAKER POND- UPPER
BEARCAMP POND
BRANCH
BRUCE
CAWLEY POND
CHASE POND
CHILDS BOG
CHITTENDEN
CLUB POND
CRANBERRY
MEADOW
CURTIS
DENNIS
DUNMORE
DUTCHMAN POND
EASTMAN POND
ECHO (CHARTN)
ECHO (HUBDTN)
ELM BROOK POOL
FERN
FISH POND
FREESES POND- UPPER
GILES POND
GREAT HOSMER
GREENWOOD POND
HALL POND- UPPER
HARDWICK
HARDWOOD
HIGH (SUDBRY)
HILDRETH DAM POND
HORN POND
HORTONIA
HOWE RESERVOIR
ISLAND POND
IVANHOE- LAKE
JACKSONVILLE
JENNESS POND
Epilimnion
MeHg
[ng/L]
0.120
0.050
0.090
0.020
0.020
0.150
0.080
0.000
0.000
0.530
0.070
0.020
0.020
0.080
0.000
0.050
0.260
0.040
0.040
0.100
0.000
0.030
0.020
0.000
0.000
0.080
0.200
0.250
0.000
0.410
0.100
0.020
0.010
0.060
0.030
0.000
0.200
0.000
0.050
HgT
[ng/L]
0.630
0.780
0.620
0.320
0.310
1.590
0.850
0.090
0.110
1.740
0.630
0.380
0.310
0.500
0.120
0.670
2.400
0.330
0.560
0.870
0.020
0.480
0.370
0.070
0.080
0.560
0.770
1.790
0.040
2.300
0.570
0.270
0.260
0.510
0.430
0.030
2.040
0.080
0.820
Hypolimnion
MeHg
[ng/L]
0.959
0.531
0.164
1.925
0.837
0.487
0.338
0.301
0.924
0.106
0.365
2.354
0.706
0.205
0.093
0.469
0.002
HgT
[ng/L]
4.495
2.958
1.220
5.699
3.370
4.366
2.497
2.044
4.386
1.003
3.729
5.486
6.017
1.173
0.885
2.521
18.727
Fish
HgT
[ug/g]
0.110
0.180
0.070
0.070
0.040
0.680
0.190
0.020
0.020
1.470
0.160
0.070
0.060
0.080
0.010
0.170
1.140
0.050
0.130
0.210
0.000
0.110
0.040
0.010
0.010
0.070
0.330
0.660
0.000
1.000
0.100
0.050
0.040
0.120
0.090
0.000
0.880
0.010
0.190
Sediment
HgT
[ug/g]
0.020
0.080
0.010
0.020
0.020
0.080
0.040
0.010
0.010
0.120
0.040
0.020
0.020
0.010
0.010
0.050
0.230
0.010
0.030
0.020
0.000
0.020
0.010
0.000
0.010
0.010
0.020
0.100
0.000
0.110
0.020
0.020
0.020
0.010
0.020
0.000
0.220
0.000
0.090
-------
Evaluating R-MCMfor 91 VT/NH Lakes
KENT
LARY POND
LEFFERTS
LILY POND
LITTLE AVERILL
LONG (WESTMR)
LOON LAKE
LOVELL LAKE- STN 1
LYFORD
MANSFIELD
MCCONNELL
MILLSFffiLD POND
MILTON
MINARDS
MITCHELL
MOOSE POND
MOUNTAIN LAKE-
UPPER
NEWARK
NORTH (BRKFLD)
NOTCH
NOYES
PARAN
PARKER
PAUGUS BAY- STN 1
PAWTUCKAWAY
LAKE
PEMIGEWASSET LAKE
PERCH (BENSON)
PLEASANT VALLEY
POUT POND
POWWOW POND
ROBB RESERVOIR
ROUND POND
SABIN
SHAWS POND
SILVER LAKE
SOMERSET
SOUTH AMERICA
SPRUCE POND
STRATTON
Epilimnion
MeHg
0.050
0.030
0.010
0.530
0.070
0.070
0.080
0.130
0.070
0.040
0.010
0.020
0.260
0.190
0.080
0.030
0.030
0.040
0.020
0.060
0.010
0.010
0.040
0.010
0.060
0.050
0.180
0.120
0.390
0.000
0.000
0.080
0.020
0.140
0.060
0.080
0.040
0.160
0.560
HgT
0.510
0.230
0.180
1.250
1.260
0.950
1.310
0.860
0.740
0.530
0.310
0.290
1.230
1.020
0.930
0.480
0.400
0.500
0.330
0.900
0.230
0.190
0.400
0.220
0.780
0.650
1.100
0.690
0.930
0.060
0.100
1.220
0.260
1.430
0.620
0.740
0.620
0.770
2.050
Hypolimnion
MeHg
0.347
0.522
0.547
1.461
0.578
1.932
0.292
0.408
0.254
0.254
0.442
0.417
1.766
0.000
0.472
0.171
0.526
0.703
1.118
HgT
3.016
4.282
2.651
4.573
3.030
5.879
1.658
1.727
1.550
1.550
1.889
1.719
6.235
0.024
2.768
1.662
2.901
2.092
3.935
Fish
HgT
0.140
0.040
0.050
0.510
0.120
0.280
0.320
0.270
0.140
0.120
0.030
0.040
0.570
0.480
0.270
0.070
0.080
0.120
0.050
0.100
0.030
0.030
0.070
0.050
0.120
0.120
0.380
0.360
0.410
0.000
0.010
0.260
0.050
0.550
0.210
0.210
0.170
0.150
1.330
Sediment
HgT
0.020
0.010
0.010
0.030
0.050
0.060
0.130
0.050
0.020
0.030
0.010
0.010
0.060
0.080
0.030
0.020
0.020
0.020
0.010
0.030
0.010
0.010
0.010
0.020
0.040
0.040
0.060
0.040
0.020
0.000
0.000
0.110
0.010
0.130
0.040
0.050
0.020
0.020
0.110
-------
Evaluating R-MCMfor 91 VT/NH Lakes
SUNCOOK POND-
UPPER
SUNRISE LAKE
SUNSET (BRKFLD)
TRIO PONDS- ONE
AND TWO
TUTTLE (HARDWK)
UNNAMED POND
WALKER POND
WILLEY POND- BIG
WILLEY POND- LITTLE
WILLOUGHBY
WILSON POND
WOLCOTT
ZEPHYR LAKE
Epilimnion
MeHg
0.010
0.150
0.000
0.190
0.070
0.030
0.040
0.350
0.100
0.050
0.250
0.140
0.110
HgT
0.230
0.970
0.120
1.700
0.400
0.480
0.550
1.110
0.830
0.540
1.220
1.350
0.840
Hypolimnion
MeHg
0.095
0.066
1.823
0.335
0.384
1.156
1.368
HgT
1.191
1.403
4.892
1.654
1.751
3.748
3.839
Fish
HgT
0.030
0.390
0.020
0.490
0.050
0.170
0.090
3.780
0.600
0.180
0.610
0.300
0.260
Sediment
HgT
0.010
0.060
0.010
0.080
0.010
0.020
0.030
0.300
0.040
0.030
0.070
0.060
0.050
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-5. Predicted Results for Tier 3.
Lake Name
ADDER POND
ARMINGTON LAKE
BAKER (BARTON)
BAKER POND- UPPER
BEARCAMP POND
BRANCH
BRUCE
CAWLEY POND
CHASE POND
CHILDS BOG
CHITTENDEN
CLUB POND
CRANBERRY
MEADOW
CURTIS
DENNIS
DUNMORE
DUTCHMAN POND
EASTMAN POND
ECHO (CHARTN)
ECHO (HUBDTN)
ELM BROOK POOL
FERN
FISH POND
FREESES POND- UPPER
GILES POND
GREAT HOSMER
GREENWOOD POND
HALL POND- UPPER
HARDWICK
HARDWOOD
HIGH (SUDBRY)
HILDRETH DAM POND
HORN POND
HORTONIA
HOWE RESERVOIR
ISLAND POND
IVANHOE- LAKE
JACKSONVILLE
JENNESS POND
Epilimnion
MeHg
[ng/L]
0.150
0.050
0.090
0.030
0.020
0.240
0.100
0.000
0.000
0.770
0.090
0.030
0.020
0.100
0.000
0.060
0.320
0.050
0.040
0.120
0.000
0.040
0.020
0.000
0.000
0.090
0.200
0.320
0.000
0.480
0.140
0.020
0.010
0.070
0.030
0.000
0.260
0.000
0.060
HgT
[ng/L]
0.830
0.960
0.620
0.350
0.420
2.630
1.060
0.110
0.140
2.560
0.850
0.540
0.390
0.600
0.120
0.870
2.960
0.400
0.670
1.170
0.030
0.620
0.470
0.090
0.090
0.660
0.770
2.390
0.040
2.750
0.790
0.330
0.280
0.630
0.630
0.030
2.670
0.120
1.060
Hypolimnion
MeHg
[ng/L]
0.959
0.531
0.164
1.926
0.837
0.487
0.338
0.301
0.924
0.106
0.365
2.354
0.706
0.205
0.093
0.469
0.002
HgT
[ng/L]
4.495
2.958
1.220
5.699
3.370
4.366
2.497
2.044
4.586
1.003
3.729
5.486
6.017
1.173
0.885
2.521
18.727
Fish
HgT
[ug/g]
0.140
0.210
0.070
0.080
0.050
1.080
0.220
0.020
0.020
2.130
0.210
0.100
0.070
0.100
0.010
0.200
1.390
0.050
0.140
0.270
0.000
0.140
0.050
0.010
0.020
0.080
0.330
0.860
0.000
1.180
0.140
0.050
0.040
0.130
0.120
0.000
1.130
0.020
0.240
Sediment
HgT
[ug/g]
0.020
0.100
0.010
0.020
0.020
0.140
0.040
0.010
0.010
0.170
0.050
0.020
0.020
0.020
0.010
0.070
0.280
0.010
0.040
0.030
0.000
0.020
0.010
0.000
0.010
0.010
0.020
0.130
0.000
0.130
0.020
0.020
0.030
0.020
0.030
0.000
0.290
0.010
0.120
-------
Evaluating R-MCMfor 91 VT/NH Lakes
KENT
LARY POND
LEFFERTS
LILY POND
LITTLE AVERILL
LONG (WESTMR)
LOON LAKE
LOVELL LAKE- STN 1
LYFORD
MANSFIELD
MCCONNELL
MILLSFffiLD POND
MILTON
MINARDS
MITCHELL
MOOSE POND
MOUNTAIN LAKE-
UPPER
NEWARK
NORTH (BRKFLD)
NOTCH
NOYES
PARAN
PARKER
PAUGUS BAY- STN 1
PAWTUCKAWAY
LAKE
PEMIGEWASSET LAKE
PERCH (BENSON)
PLEASANT VALLEY
POUT POND
POWWOW POND
ROBB RESERVOIR
ROUND POND
SABIN
SHAWS POND
SILVER LAKE
SOMERSET
SOUTH AMERICA
SPRUCE POND
STRATTON
Epilimnion
MeHg
0.070
0.040
0.020
0.590
0.090
0.090
0.090
0.150
0.080
0.050
0.020
0.020
0.330
0.220
0.080
0.040
0.030
0.050
0.020
0.070
0.010
0.010
0.040
0.010
0.070
0.050
0.210
0.170
0.450
0.000
0.000
0.080
0.020
0.180
0.070
0.080
0.060
0.150
0.890
HgT
0.740
0.300
0.270
1.390
1.750
1.330
1.640
1.060
0.850
0.780
0.430
0.420
1.620
1.240
0.930
0.690
0.490
0.630
0.410
1.260
0.320
0.260
0.400
0.230
0.920
0.770
1.410
1.010
1.090
0.060
0.150
1.390
0.330
1.910
0.830
0.740
0.910
0.740
3.310
Hypolimnion
MeHg
0.347
0.522
0.547
1.461
0.578
1.932
0.292
0.408
0.254
0.169
0.442
0.417
1.766
0.472
0.171
0.526
0.703
1.118
HgT
3.016
4.282
2.651
4.573
3.030
5.879
1.658
1.727
1.550
1.104
1.889
1.719
6.235
0.024
2.768
1.662
2.901
2.092
3.935
Fish
HgT
0.200
0.050
0.080
0.560
0.160
0.360
0.390
0.320
0.160
0.160
0.040
0.050
0.730
0.570
0.270
0.090
0.090
0.130
0.060
0.130
0.040
0.040
0.070
0.060
0.140
0.140
0.470
0.510
0.480
0.000
0.020
0.300
0.050
0.710
0.260
0.210
0.220
0.150
2.130
Sediment
HgT
0.030
0.010
0.020
0.030
0.080
0.090
0.160
0.060
0.020
0.050
0.020
0.020
0.080
0.090
0.030
0.030
0.020
0.020
0.010
0.040
0.010
0.010
0.010
0.020
0.040
0.040
0.070
0.060
0.030
0.000
0.010
0.130
0.020
0.170
0.060
0.050
0.040
0.020
0.180
-------
Evaluating R-MCMfor 91 VT/NH Lakes
SUNCOOK POND-
UPPER
SUNRISE LAKE
SUNSET (BRKFLD)
TRIO PONDS- ONE
AND TWO
TUTTLE (HARDWK)
UNNAMED POND
WALKER POND
WILLEY POND- BIG
WILLEY POND- LITTLE
WILLOUGHBY
WILSON POND
WOLCOTT
ZEPHYR LAKE
Epilimnion
MeHg
0.010
0.200
0.010
0.270
0.070
0.040
0.040
0.480
0.110
0.050
0.280
0.180
0.150
HgT
0.290
1.350
0.130
2.620
0.420
0.690
0.670
1.570
0.880
0.540
1.380
1.750
1.160
Hypolimnion
MeHg
0.095
0.066
1.823
0.335
0.384
1.156
1.368
HgT
1.191
1.403
4.892
1.654
1.751
3.478
3.839
Fish
HgT
0.030
0.530
0.020
0.720
0.050
0.230
0.110
5.260
0.630
0.180
0.680
0.380
0.360
Sediment
HgT
0.010
0.080
0.010
0.120
0.010
0.030
0.040
0.430
0.040
0.030
0.080
0.080
0.080
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-6. Predicted Results for Tier 4.
Lake Name
ADDER POND
ARMINGTON LAKE
BAKER (BARTON)
BAKER POND- UPPER
BEARCAMP POND
BRANCH
BRUCE
CAWLEY POND
CHASE POND
CHILDS BOG
CHITTENDEN
CLUB POND
CRANBERRY
MEADOW
CURTIS
DENNIS
DUNMORE
DUTCHMAN POND
EASTMAN POND
ECHO (CHARTN)
ECHO (HUBDTN)
ELM BROOK POOL
FERN
FISH POND
FREESES POND- UPPER
GILES POND
GREAT HOSMER
GREENWOOD POND
HALL POND- UPPER
HARDWICK
HARDWOOD
HIGH (SUDBRY)
HILDRETH DAM POND
HORN POND
HORTONIA
HOWE RESERVOIR
ISLAND POND
IVANHOE- LAKE
JACKSONVILLE
JENNESS POND
Epilimnion
MeHg
[ng/L]
0.140
0.040
0.130
0.040
0.030
0.260
0.110
0.040
0.030
0.660
0.100
0.070
0.080
0.110
0.010
0.060
0.220
0.050
0.090
0.110
0.010
0.050
0.040
0.030
0.020
0.070
0.170
0.260
0.010
0.440
0.120
0.050
0.030
0.060
0.050
0.000
0.160
0.020
0.050
HgT
[ng/L]
0.840
0.890
0.990
0.550
0.830
3.040
1.340
1.100
1.320
2.250
1.040
1.370
1.530
0.720
0.190
1.040
2.180
0.450
1.840
1.190
0.480
0.800
0.950
0.430
0.810
0.570
0.650
2.080
0.360
2.650
0.740
1.120
1.090
0.690
0.980
0.060
1.690
0.750
1.010
Hypolimnion
MeHg
[ng/L]
0.937
0.811
0.293
2.131
0.950
0.560
0.368
0.628
0.847
0.145
0.298
1.945
0.637
0.649
0.294
0.447
0.011
HgT
[ng/L]
4.439
4.566
2.057
6.487
4.009
5.103
2.750
5.361
4.513
1.689
3.154
4.675
5.511
2.985
2.119
2.672
18.776
Fish
HgT
[ug/g]
0.140
0.170
0.110
0.110
0.090
1.200
0.260
0.220
0.180
1.840
0.240
0.250
0.270
0.110
0.020
0.200
0.990
0.060
0.300
0.250
0.060
0.180
0.090
0.070
0.150
0.060
0.280
0.710
0.040
1.090
0.120
0.170
0.140
0.130
0.170
0.000
0.690
0.100
0.210
Sediment
HgT
[ug/g]
0.020
0.100
0.020
0.030
0.040
0.160
0.060
0.080
0.070
0.150
0.070
0.060
0.080
0.020
0.010
0.080
0.210
0.020
0.120
0.030
0.030
0.030
0.030
0.010
0.090
0.010
0.020
0.110
0.010
0.130
0.020
0.070
0.100
0.020
0.050
0.000
0.180
0.040
0.110
-------
Evaluating R-MCMfor 91 VT/NH Lakes
KENT
LARY POND
LEFFERTS
LILY POND
LITTLE AVERILL
LONG (WESTMR)
LOON LAKE
LOVELL LAKE- STN 1
LYFORD
MANSFIELD
MCCONNELL
MILLSFffiLD POND
MILTON
MINARDS
MITCHELL
MOOSE POND
MOUNTAIN LAKE-
UPPER
NEWARK
NORTH (BRKFLD)
NOTCH
NOYES
PARAN
PARKER
PAUGUS BAY- STN 1
PAWTUCKAWAY
LAKE
PEMIGEWASSET LAKE
PERCH (BENSON)
PLEASANT VALLEY
POUT POND
POWWOW POND
ROBB RESERVOIR
ROUND POND
SABIN
SHAWS POND
SILVER LAKE
SOMERSET
SOUTH AMERICA
SPRUCE POND
STRATTON
Epilimnion
MeHg
0.100
0.110
0.050
0.390
0.060
0.070
0.100
0.110
0.070
0.110
0.040
0.020
0.340
0.150
0.420
0.040
0.060
0.020
0.040
0.100
0.040
0.110
0.040
0.140
0.070
0.050
0.160
0.260
0.360
0.000
0.010
0.060
0.060
0.220
0.040
0.090
0.060
0.110
0.670
HgT
1.150
0.880
0.910
0.940
1.480
1.170
2.030
0.840
0.840
2.280
1.050
0.740
1.770
0.910
5.930
0.840
1.270
0.440
0.830
1.790
1.400
2.380
0.560
2.700
1.090
0.840
1.110
1.680
0.900
0.190
0.290
1.030
1.310
2.450
0.610
0.860
1.110
0.570
2.500
Hypolimnion
MeHg
0.263
0.403
0.630
1.090
0.504
1.365
0.157
0.567
0.281
1.861
0.498
0.411
1.324
0.002
0.328
0.596
0.321
0.776
0.854
HgT
2.520
3.674
3.195
3.551
2.887
4.232
1.125
2.386
2.038
7.698
2.204
1.817
4.820
0.156
2.030
5.640
2.063
2.365
3.019
Fish
HgT
0.280
0.160
0.250
0.380
0.120
0.280
0.450
0.240
0.140
0.420
0.090
0.080
0.780
0.400
1.440
0.100
0.210
0.070
0.110
0.180
0.150
0.390
0.080
0.620
0.160
0.140
0.350
0.810
0.390
0.000
0.030
0.210
0.190
0.880
0.160
0.230
0.260
0.110
1.600
Sediment
HgT
0.050
0.030
0.070
0.020
0.060
0.080
0.200
0.050
0.020
0.140
0.040
0.040
0.090
0.070
0.210
0.030
0.070
0.020
0.030
0.060
0.060
0.070
0.010
0.260
0.050
0.050
0.060
0.100
0.020
0.000
0.010
0.090
0.070
0.230
0.040
0.060
0.040
0.010
0.140
-------
Evaluating R-MCMfor 91 VT/NH Lakes
SUNCOOK POND-
UPPER
SUNRISE LAKE
SUNSET (BRKFLD)
TRIO PONDS- ONE
AND TWO
TUTTLE (HARDWK)
UNNAMED POND
WALKER POND
WILLEY POND- BIG
WILLEY POND- LITTLE
WILLOUGHBY
WILSON POND
WOLCOTT
ZEPHYR LAKE
Epilimnion
MeHg
0.040
0.170
0.020
0.240
0.070
0.040
0.070
0.410
0.110
0.030
0.270
0.180
0.160
HgT
1.110
1.190
0.540
2.560
0.440
0.710
1.170
1.370
0.880
0.490
1.370
1.910
1.280
Hypolimnion
MeHg
0.335
0.267
1.679
0.538
0.260
1.189
1.189
HgT
3.452
2.325
4.677
2.718
1.498
3.715
3.715
Fish
HgT
0.120
0.460
0.080
0.660
0.050
0.210
0.180
4.510
0.630
0.120
0.660
0.390
0.380
Sediment
HgT
0.040
0.070
0.040
0.120
0.010
0.030
0.060
0.370
0.040
0.020
0.080
0.090
0.090
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-7. Predicted Results for Tier 5.
Lake Name
ADDER POND
ARMINGTON LAKE
BAKER (BARTON)
BAKER POND- UPPER
BEARCAMP POND
BRANCH
BRUCE
CAWLEY POND
CHASE POND
CHILDS BOG
CHITTENDEN
CLUB POND
CRANBERRY
MEADOW
CURTIS
DENNIS
DUNMORE
DUTCHMAN POND
EASTMAN POND
ECHO (CHARTN)
ECHO (HUBDTN)
ELM BROOK POOL
FERN
FISH POND
FREESES POND- UPPER
GILES POND
GREAT HOSMER
GREENWOOD POND
HALL POND- UPPER
HARDWICK
HARDWOOD
HIGH (SUDBRY)
HILDRETH DAM POND
HORN POND
HORTONIA
HOWE RESERVOIR
ISLAND POND
IVANHOE- LAKE
JACKSONVILLE
JENNESS POND
Epilimnion
MeHg
[ng/L]
0.11
0.04
0.05
0.06
0.02
0.09
0.11
0.02
0.02
0.51
0.1
0.07
0.03
0.06
0.1
0.02
0.22
0.04
0.03
0.04
0.02
0.01
0.04
0.03
0.01
0.06
0.14
0.05
0.03
0.31
0.07
0.01
0.03
0.03
0.08
0
0.17
0.04
0.05
HgT
[ng/L]
0.72
0.84
0.56
0.72
0.66
1.55
1.34
0.84
1.02
1.9
1.01
1.37
0.75
0.48
1.28
0.47
2.18
0.39
0.69
0.51
0.63
0.35
0.95
0.43
0.42
0.46
0.57
0.79
0.74
2.13
0.51
0.48
1.02
0.41
1.39
0.08
1.81
1.11
0.99
Hypolimnion
MeHg
[ng/L]
0.754
0.363
0.22
0.861
0.931
0.31
0.301
0.182
0.333
0.024
0.222
0.484
0.368
0.232
0.28
0.229
0.018
HgT
[ng/L]
3.696
2.566
1.759
3.161
3.364
3.284
2.369
2.243
1.872
0.523
2.591
1.719
3.896
2.095
1.866
1.588
11.718
Fish
HgT
[ug/g]
0.11
0.16
0.05
0.16
0.07
0.48
0.26
0.16
0.13
1.46
0.23
0.25
0.11
0.07
0.24
0.08
0.99
0.05
0.09
0.1
0.09
0.03
0.09
0.07
0.07
0.05
0.23
0.17
0.09
0.79
0.07
0.06
0.13
0.07
0.27
0.01
0.74
0.16
0.21
Sediment
HgT
[ug/g]
0.02
0.09
0.01
0.04
0.04
0.08
0.06
0.06
0.06
0.13
0.06
0.06
0.04
0.01
0.06
0.03
0.21
0.01
0.03
0.01
0.04
0.01
0.03
0.01
0.04
0.01
0.01
0.04
0.02
0.1
0.01
0.03
0.1
0.01
0.07
0
0.2
0.06
0.11
-------
Evaluating R-MCMfor 91 VT/NH Lakes
KENT
LARY POND
LEFFERTS
LILY POND
LITTLE AVERILL
LONG (WESTMR)
LOON LAKE
LOVELL LAKE- STN 1
LYFORD
MANSFIELD
MCCONNELL
MILLSFffiLD POND
MILTON
MINARDS
MITCHELL
MOOSE POND
MOUNTAIN LAKE-
UPPER
NEWARK
NORTH (BRKFLD)
NOTCH
NOYES
PARAN
PARKER
PAUGUS BAY- STN 1
PAWTUCKAWAY
LAKE
PEMIGEWASSET LAKE
PERCH (BENSON)
PLEASANT VALLEY
POUT POND
POWWOW POND
ROBB RESERVOIR
ROUND POND
SABIN
SHAWS POND
SILVER LAKE
SOMERSET
SOUTH AMERICA
SPRUCE POND
STRATTON
Epilimnion
MeHg
0.05
0.09
0.1
0.42
0.03
0.02
0.05
0.09
0.04
0.04
0.02
0.06
0.23
0.08
0.31
0.02
0.02
0.02
0.03
0.03
0.03
0.06
0.03
0.03
0.03
0.04
0.08
0.07
0.39
0.01
0.03
0.03
0.02
0.12
0.02
0.03
0.12
0.11
0.49
HgT
0.74
0.77
1.32
0.98
0.54
0.33
1.17
0.69
0.55
1.24
0.63
1.28
1.37
0.55
4.67
0.57
0.66
0.41
0.62
0.77
1.09
1.47
0.41
1.06
0.65
0.79
0.6
0.75
0.94
0.49
0.93
0.67
0.71
1.67
0.31
0.41
1.69
0.55
2.03
Hypolimnion
MeHg
0.114
0.148
0.481
0.87
0.319
0.716
0.155
0.205
0.183
0.586
0.25
0.374
0.647
0.007
0.293
0.239
0.17
0.275
0.812
HgT
1.193
1.153
1.822
2.639
1.655
2.719
0.957
1.248
1.468
4.463
1.533
1.672
2.484
0.396
1.295
3.901
1.091
1.314
2.893
Fish
HgT
0.15
0.14
0.41
0.4
0.05
0.07
0.24
0.19
0.08
0.18
0.05
0.17
0.55
0.21
1.07
0.06
0.09
0.07
0.08
0.07
0.11
0.22
0.05
0.2
0.08
0.12
0.17
0.25
0.41
0.01
0.12
0.13
0.08
0.53
0.08
0.08
0.46
0.11
1.21
Sediment
HgT
0.03
0.02
0.1
0.02
0.02
0.02
0.11
0.04
0.01
0.08
0.02
0.06
0.07
0.04
0.16
0.02
0.03
0.01
0.02
0.02
0.05
0.05
0.01
0.1
0.03
0.04
0.03
0.04
0.02
0.01
0.04
0.06
0.03
0.15
0.02
0.03
0.07
0.01
0.12
-------
Evaluating R-MCMfor 91 VT/NH Lakes
SUNCOOK POND-
UPPER
SUNRISE LAKE
SUNSET (BRKFLD)
TRIO PONDS- ONE
AND TWO
TUTTLE (HARDWK)
UNNAMED POND
WALKER POND
WILLEY POND- BIG
WILLEY POND- LITTLE
WILLOUGHBY
WILSON POND
WOLCOTT
ZEPHYR LAKE
Epilimnion
MeHg
0.03
0.35
0.01
0.07
0.11
0.07
0.04
0.17
0.08
0.01
0.18
0.08
0.11
HgT
0.74
1.86
0.22
1.19
0.57
1.03
0.79
0.85
0.65
0.16
1.04
1.15
1.01
Hypolimnion
MeHg
0.195
0.094
0.582
0.319
0.105
0.592
1.065
HgT
2.657
3.492
2.085
2.041
0.459
2.121
3.029
Fish
HgT
0.07
0.85
0.03
0.23
0.08
0.35
0.11
2.12
0.46
0.05
0.45
0.19
0.27
Sediment
HgT
0.02
0.11
0.01
0.06
0.01
0.05
0.04
0.26
0.03
0.003
0.06
0.05
0.07
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-8. Predicted Epilimnetic Methylmercury Concentrations for the
Hypolimnion Surface Area Sensitivity Analysis. Data are presented in columns of
observed, and then for the fraction used to estimate the hypolimnion surface area (e.g.,
1/2 data corresponds to a hypolimnion area =1/2 lake surface area) in the model.
Lake Name
ADDER POND
ARMINGTON LAKE
BAKER (BARTON)
BEARCAMP POND
BRANCH
CHITTENDEN
CRANBERRY MEADOW
CURTIS
DUNMORE
EASTMAN POND
ECHO (CHARTN)
ECHO (HUBDTN)
FERN
GREAT HOSMER
HALL POND- UPPER
HIGH (SUDBRY)
HILDRETH DAM POND
HORN POND
HORTONIA
ISLAND POND
JENNESS POND
LITTLE AVERILL
LONG (WESTMR)
LOON LAKE
LOVELL LAKE- STN 1
LYFORD
MANSFIELD
MCCONNELL
MINARDS
NEWARK
NORTH (BRKFLD)
NOTCH
PARKER
PAUGUS BAY- STN 1
PAWTUCKAWAY LAKE
PEMIGEWASSET LAKE
PERCH (BENSON)
POWWOW POND
Observed
0.289
0.218
0.096
0.214
0.49
0.113
0.215
0.175
0.098
0.194
0.072
0.188
0.231
0.123
0.299
0.175
0.349
0.141
0.166
0.295
0.475
0.201
0.067
0.52
0.318
0.305
0.243
0.18
0.23
0.165
0.136
0.329
0.217
0.141
0.187
0.456
0.162
1.813
1/2
0.06
0.03
0.04
0.02
0.07
0.07
0.02
0.03
0.02
0.03
0.02
0.03
0.01
0.03
0.04
0.04
0.01
0.02
0.03
0
0.04
0.03
0.02
0.04
0.06
0.03
0.03
0.02
0.05
0.02
0.02
0.03
0.02
0.03
0.03
0.03
0.05
0.01
1/3
0.11
0.04
0.05
0.02
0.09
0.1
0.03
0.06
0.02
0.04
0.03
0.04
0.01
0.06
0.05
0.07
0.01
0.03
0.03
0
0.05
0.03
0.02
0.05
0.09
0.04
0.04
0.02
0.08
0.02
0.03
0.03
0.03
0.03
0.03
0.04
0.08
0.01
1/4
0.15
0.04
0.07
0.02
0.11
0.12
0.03
0.07
0.02
0.05
0.03
0.05
0.02
0.07
0.06
0.09
0.02
0.03
0.04
0
0.06
0.03
0.02
0.06
0.1
0.05
0.05
0.02
0.1
0.03
0.03
0.03
0.03
0.04
0.04
0.05
0.09
0.01
1/5
0.18
0.05
0.07
0.03
0.13
0.13
0.03
0.09
0.02
0.06
0.03
0.06
0.02
0.08
0.07
0.11
0.02
0.03
0.04
0
0.06
0.03
0.02
0.07
0.11
0.05
0.05
0.02
0.11
0.03
0.03
0.04
0.03
0.04
0.04
0.05
0.1
0.01
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Lake Name
ROUND POND
SAB IN
SILVER LAKE
SOMERSET
SPRUCE POND
SUNCOOK POND- UPPER
SUNSET (BRKFLD)
TRIO PONDS- ONE AND TWO
WALKER POND
WILLOUGHBY
WOLCOTT
ZEPHYR LAKE
Observed
0.287
0.223
0.077
0.116
0.327
0.107
0.232
0.3
0.155
0.183
0.322
0.213
1/2
0.03
0.02
0.02
0.02
0.06
0.02
0
0.05
0.03
0.01
0.06
0.07
1/3
0.03
0.02
0.02
0.03
0.11
0.03
0.01
0.07
0.04
0.01
0.08
0.11
1/4
0.04
0.03
0.02
0.03
0.15
0.03
0.01
0.08
0.04
0.02
0.09
0.13
1/5
0.04
0.03
0.03
0.03
0.18
0.03
0.01
0.09
0.04
0.02
0.1
0.14
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-9. Predicted Epilimnetic Total Mercury Concentrations for the
Hypolimnion Surface Area Sensitivity Analysis. Data are presented in columns of
observed, and then for the fraction used to estimate the hypolimnion surface area (e.g.,
1/2 data corresponds to a hypolimnion area =1/2 lake surface area) in the model.
Lake Name
ADDER POND
ARMINGTON LAKE
BAKER (BARTON)
BEARCAMP POND
BRANCH
CHITTENDEN
CRANBERRY MEADOW
CURTIS
DUNMORE
EASTMAN POND
ECHO (CHARTN)
ECHO (HUBDTN)
FERN
GREAT HOSMER
HALL POND- UPPER
HIGH (SUDBRY)
HILDRETH DAM POND
HORN POND
HORTONIA
ISLAND POND
JENNESS POND
LITTLE AVERILL
LONG (WESTMR)
LOON LAKE
LOVELL LAKE- STN 1
LYFORD
MANSFIELD
MCCONNELL
MINARDS
NEWARK
NORTH (BRKFLD)
NOTCH
Observed
2.143
0.908
0.94
1.124
1.776
1.142
1.118
0.648
0.507
1.639
0.594
0.352
1.035
0.679
1.89
1.72
1.96
1.506
0.656
1.578
3.101
0.501
0.817
0.861
0.503
0.946
0.95
1.962
2.422
1.066
0.846
5.296
1/2
0.61
0.76
0.5
0.62
1.38
0.91
0.71
0.43
0.45
0.35
0.67
0.5
0.33
0.41
0.71
0.44
0.45
0.96
0.4
0.08
0.89
0.52
0.32
1.06
0.63
0.53
1.13
0.59
0.48
0.4
0.59
0.71
1/3
0.72
0.84
0.56
0.66
1.55
1.01
0.75
0.48
0.47
0.39
0.69
0.51
0.35
0.46
0.79
0.51
0.48
1.02
0.41
0.08
0.99
0.54
0.33
1.17
0.69
0.55
1.24
0.63
0.55
0.41
0.62
0.77
1/4
0.8
0.9
0.59
0.68
1.64
1.06
0.77
0.52
0.48
0.42
0.7
0.53
0.36
0.49
0.84
0.54
0.5
1.05
0.42
0.08
1.05
0.55
0.34
1.24
0.73
0.57
1.3
0.65
0.59
0.42
0.63
0.8
1/5
0.86
0.93
0.62
0.7
1.71
1.1
0.79
0.54
0.48
0.44
0.71
0.53
0.37
0.52
0.87
0.57
0.51
1.07
0.42
0.08
1.09
0.55
0.34
1.28
0.76
0.57
1.33
0.66
0.62
0.42
0.64
0.82
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Lake Name
PARKER
PAUGUS BAY- STN 1
PAWTUCKAWAY LAKE
PEMIGEWASSET LAKE
PERCH (BENSON)
POWWOW POND
ROUND POND
SAB IN
SILVER LAKE
SOMERSET
SPRUCE POND
SUNCOOK POND- UPPER
SUNSET (BRKFLD)
TRIO PONDS- ONE AND TWO
WALKER POND
WILLOUGHBY
WOLCOTT
ZEPHYR LAKE
Observed
1.741
3.12
2.257
2.057
0.586
5.872
0.69
1.216
2.224
1.885
1.456
1.38
0.451
2.267
0.932
1.116
1.132
0.899
1/2
0.39
0.97
0.6
0.72
0.55
0.5
0.61
0.67
0.3
0.37
0.46
0.7
0.21
1.08
0.73
0.16
1.07
0.9
1/3
0.41
1.06
0.65
0.79
0.6
0.49
0.67
0.71
0.31
0.41
0.55
0.74
0.22
1.19
0.79
0.16
1.15
1.01
1/4
0.41
1.11
0.69
0.82
0.63
0.49
0.71
0.74
0.31
0.43
0.62
0.76
0.23
1.26
0.82
0.16
1.2
1.07
1/5
0.42
1.14
0.71
0.85
0.65
0.49
0.73
0.75
0.31
0.45
0.68
0.77
0.23
1.3
0.85
0.17
1.24
1.11
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-10. Predicted Hypolimnetic Methylmercury Concentrations for the
Hypolimnion Surface Area Sensitivity Analysis. Data are presented in columns of
observed, and then for the fraction used to estimate the hypolimnion surface area (e.g.,
1/2 data corresponds to a hypolimnion area =1/2 lake surface area) in the model.
Lake Name
ADDER POND
BAKER (BARTON)
BEARCAMP POND
BRANCH
CHITTENDEN
CURTIS
EASTMAN POND
ECHO (CHARTN)
ECHO (HUBDTN)
FERN
GREAT HOSMER
HALL POND- UPPER
HIGH (SUDBRY)
HILDRETH DAM POND
HORN POND
HORTONIA
ISLAND POND
LITTLE AVERILL
LONG (WESTMR)
LOON LAKE
LOVELL LAKE- STN 1
LYFORD
MINARDS
NEWARK
NOTCH
PARKER
PAUGUS BAY- STN 1
PAWTUCKAWAY LAKE
PEMIGEWASSET LAKE
PERCH (BENSON)
POWWOW POND
ROUND POND
Observed
1.004
1.649
1.753
4.447
0.938
0.67
0.388
0.142
0.263
0.374
0.394
0.273
1.415
1.005
0.163
0.127
0.379
0.353
0.288
0.552
0.277
0.205
0.448
0.584
3.186
0.156
0.051
2.908
1.213
0.573
0.822
0.433
1/2
0.456
0.271
0.189
0.6
0.681
0.357
0.206
0.149
0.243
0.022
0.144
0.367
0.228
0.189
0.235
0.177
0.016
0.096
0.117
0.354
0.638
0.244
0.476
0.129
0.17
0.145
0.471
0.206
0.303
0.453
0.007
0.224
1/3
0.754
0.363
0.22
0.861
0.931
0.31
0.301
0.182
0.333
0.024
0.222
0.484
0.368
0.232
0.28
0.229
0.018
0.114
0.148
0.481
0.87
0.319
0.716
0.155
0.205
0.183
0.586
0.25
0.374
0.647
0.007
0.293
1/4
0.988
0.436
0.237
1.045
1.091
0.389
0.366
0.201
0.392
0.026
0.278
0.559
0.475
0.256
0.305
0.262
0.019
0.125
0.167
0.561
1.015
0.365
0.886
0.171
0.225
0.205
0.653
0.277
0.417
0.777
0.007
0.336
1/5
1.172
0.487
0.248
1.18
1.2
0.448
0.414
0.213
0.433
0.027
0.321
0.61
0.557
0.272
0.321
0.284
0.02
0.131
0.179
0.617
1.115
0.395
1.012
0.181
0.238
0.219
0.697
0.294
0.444
0.868
0.007
0.365
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Lake Name
SABIN
SILVER LAKE
SOMERSET
SPRUCE POND
SUNCOOK POND- UPPER
SUNSET (BRKFLD)
TRIO PONDS- ONE AND TWO
WALKER POND
WILLOUGHBY
WOLCOTT
ZEPHYR LAKE
Observed
0.326
0.195
0.165
4.454
0.733
0.344
0.178
0.892
0.265
0.303
0.221
1/2
0.189
0.137
0.205
0.469
0.168
0.079
0.438
0.265
0.09
0.476
0.768
1/3
0.239
0.17
0.275
0.094
0.195
0.094
0.582
0.319
0.105
0.592
1.065
1/4
0.269
0.19
0.32
1.108
0.209
0.103
0.673
0.35
0.114
0.66
1.257
1/5
0.289
0.203
0.352
1.36
0.219
0.108
0.735
0.37
0.119
0.704
1.39
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-ll. Predicted Hypolimnetic Total Mercury Concentrations for the
Hypolimnion Surface Area Sensitivity Analysis. Data are presented in columns of
observed, and then for the fraction used to estimate the hypolimnion surface area (e.g.,
1/2 data corresponds to a hypolimnion area =1/2 lake surface area) in the model.
Lake Name
ADDER POND
BAKER (BARTON)
BEARCAMP POND
BRANCH
CHITTENDEN
CURTIS
EASTMAN POND
ECHO (CHARTN)
ECHO (HUBDTN)
FERN
GREAT HOSMER
HALL POND- UPPER
HIGH (SUDBRY)
HILDRETH DAM POND
HORN POND
HORTONIA
ISLAND POND
LITTLE AVERILL
LONG (WESTMR)
LOON LAKE
LOVELL LAKE- STN 1
LYFORD
MINARDS
NEWARK
NOTCH
PARKER
PAUGUS BAY- STN 1
PAWTUCKAWAY LAKE
PEMIGEWASSET LAKE
PERCH (BENSON)
POWWOW POND
ROUND POND
Observed
15.191
8.83
8.503
25.843
13.43
2.569
4.685
2.094
2.727
5.625
9.763
7.88
8.522
8.775
5.58
1.266
2.421
6.678
2.015
29.786
4.711
7.992
3.784
13.346
13.66
4.911
5.21
20.741
8.703
9.468
4.54
29.604
1/2
2.467
1.465
1.683
2.65
2.81
3.292
1.858
2.164
1.678
0.493
2.029
1.486
2.915
2.038
1.77
1.472
12.316
1.143
1.098
1.562
2.193
1.493
2.135
0.908
1.175
1.381
4.184
1.412
1.51
2.065
0.385
1.137
1/3
3.696
2.566
1.759
3.161
3.364
3.284
2.369
2.243
1.872
0.523
2.591
1.719
3.896
2.095
1.866
1.588
11.718
1.193
1.153
1.822
2.639
1.655
2.719
0.957
1.248
1.468
4.463
1.533
1.672
2.484
0.396
1.295
1/4
4.644
2.892
1.803
3.501
3.707
3.836
2.714
2.287
1.995
0.539
2.984
1.864
4.614
2.131
1.921
1.659
11.419
1.22
1.185
1.983
2.914
1.752
3.121
0.984
1.291
1.518
4.633
1.604
1.767
2.755
0.403
1.389
1/5
5.385
3.121
1.831
3.744
3.94
4.242
2.96
2.315
2.08
0.55
3.274
1.962
5.159
2.155
1.955
1.706
11.24
1.237
1.206
2.092
3.1
1.816
3.411
1.001
1.319
1.551
4.748
1.651
1.829
2.944
0.406
1.452
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Lake Name
SABIN
SILVER LAKE
SOMERSET
SPRUCE POND
SUNCOOK POND- UPPER
SUNSET (BRKFLD)
TRIO PONDS- ONE AND TWO
WALKER POND
WILLOUGHBY
WOLCOTT
ZEPHYR LAKE
Observed
6.54
27.134
6.402
11.518
12.303
10.277
4.75
14.282
5.267
5.659
19.048
1/2
3.688
1.032
1.157
1.871
2.58
3.63
1.804
1.892
0.443
1.893
2.419
1/3
3.901
1.091
1.314
3.492
2.657
3.492
2.085
2.041
0.459
2.121
3.029
1/4
4.024
1.125
1.415
3.757
2.702
3.426
2.257
2.128
0.468
2.253
3.416
1/5
4.104
1.147
1.485
4.485
2.73
3.387
2.374
2.184
0.475
2.338
3.681
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-12. Predicted Sediment Methylmercury Concentrations for the
Hypolimnion Surface Area Sensitivity Analysis. Data are presented in columns of
observed, and then for the fraction used to estimate the hypolimnion surface area (e.g.,
1/2 data corresponds to a hypolimnion area =1/2 lake surface area) in the model.
Lake Name
ADDER POND
ARMINGTON LAKE
BEARCAMP POND
BRANCH
CHITTENDEN
CRANBERRY MEADOW
CURTIS
DUNMORE
EASTMAN POND
ECHO (CHARTN)
ECHO (HUBDTN)
FERN
GREAT HOSMER
HIGH (SUDBRY)
HORN POND
HORTONIA
ISLAND POND
JENNESS POND
LONG (WESTMR)
LOON LAKE
LOVELL LAKE- STN 1
LYFORD
MANSFIELD
MINARDS
NEWARK
NORTH (BRKFLD)
NOTCH
PARKER
PAUGUS BAY- STN 1
PEMIGEWASSET LAKE
PERCH (BENSON)
ROUND POND
Observed
0.0032
0.0041
0.001
0.003
0.0004
0.002
0.01
0.005
0.003
0.0014
0.008
0.012
0.007
0.0018
0.0031
0.021
0.00345
0.0023
0.002
0.002
0.001
0.001
0.002
0.0044
0.002
0.004
0.0025
0.001
0.003
0.007
0.0031
0.002
1/2
0.002
0.002
0.001
0.012
0.003
0.001
0.001
0.001
0.001
0.001
0.001
0
0.001
0.001
0.002
0.001
0
0.003
0.001
0.004
0.002
0.001
0.002
0.003
0.001
0.001
0.001
0.001
0.004
0.001
0.002
0.002
1/3
0.003
0.001
0.001
0.119
0.004
0.001
0.002
0.001
0.001
0.001
0.002
0
0.009
0.002
0.003
0.001
0
0.004
0.001
0.006
0.003
0.001
0.002
0.004
0.001
0.001
0.001
0.001
0.005
0.001
0.003
0.002
1/4
0.004
0.001
0.001
0.021
0.004
0.002
0.002
0.001
0.001
0.001
0.002
0
0.001
0.002
0.003
0.001
0
0.005
0.001
0.007
0.003
0.002
0.002
0.005
0.001
0.001
0.001
0.001
0.005
0.002
0.003
0.003
1/5
0.005
0.001
0.001
0.023
0.005
0.002
0.002
0.001
0.001
0.002
0.002
0
0.001
0.003
0.003
0.001
0
0.005
0.002
0.008
0.003
0.002
0.003
0.006
0.001
0.001
0.001
0.001
0.006
0.001
0.002
0.003
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Lake Name
SABIN
SILVER LAKE
SOMERSET
SUNSET (BRKFLD)
TRIO PONDS- ONE AND
TWO
WALKER POND
WILLOUGHBY
WOLCOTT
ZEPHYR LAKE
Observed
0.001
0.001
0.001
0.008
0.002
0.003
0.005
0.003
0.0055
1/2
0.001
0.001
0.001
0
0.003
0.001
0.001
0.002
0.004
1/3
0.001
0.001
0.002
0.001
0.004
0.001
0.001
0.002
0.005
1/4
0.001
0.002
0.002
0.001
0.005
0.001
0.001
0.002
0.006
1/5
0.002
0.002
0.002
0.001
0.006
0.001
0.001
0.003
0.007
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-13. Predicted Sediment Total Mercury Concentrations for the Hypolimnion
Surface Area Sensitivity Analysis. Data are presented in columns of observed, and then
for the fraction used to estimate the hypolimnion surface area (e.g., 1/2 data corresponds
to a hypolimnion area =1/2 lake surface area) in the model.
Lake Name
ADDER POND
ARMINGTON LAKE
BAKER (BARTON)
BEARCAMP POND
BRANCH
CHITTENDEN
CRANBERRY MEADOW
CURTIS
DUNMORE
EASTMAN POND
ECHO (CHARTN)
ECHO (HUBDTN)
FERN
GREAT HOSMER
HALL POND- UPPER
HIGH (SUDBRY)
HILDRETH DAM POND
HORN POND
HORTONIA
ISLAND POND
JENNESS POND
LITTLE AVERILL
LONG (WESTMR)
LOON LAKE
LOVELL LAKE- STN 1
LYFORD
MANSFIELD
MCCONNELL
MINARDS
NEWARK
NORTH (BRKFLD)
NOTCH
Observed
0.166
0.296
0.09
0.29
0.48
0.17
0.175
0.164
0.27
0.23
0.174
0.282
0.283
0.2
0.39
0.315
0.195
0.166
0.268
0.363
0.195
0.24
0.14
0.14
0.19
0.19
0.29
0.28
0.237
0.256
0.142
0.622
1/2
0.02
0.08
0.01
0.04
0.03
0.08
0.06
0.06
0.06
0.13
0.06
0.06
0.04
0.01
0.06
0.03
0.21
0.01
0.03
0.01
0.04
0.01
0.03
0.01
0.04
0.01
0.01
0.04
0.02
0.1
0.01
0.03
1/3
0.02
0.09
0.01
0.04
0.04
0.08
0.06
0.06
0.06
0.13
0.06
0.06
0.04
0.01
0.06
0.03
0.21
0.01
0.03
0.01
0.04
0.01
0.03
0.01
0.04
0.01
0.01
0.04
0.02
0.1
0.01
0.03
1/4
0.02
0.1
0.01
0.04
0.04
0.09
0.06
0.06
0.06
0.13
0.07
0.06
0.04
0.01
0.06
0.03
0.21
0.01
0.03
0.01
0.04
0.01
0.03
0.01
0.04
0.01
0.01
0.05
0.02
0.1
0.01
0.03
1/5
0.02
0.1
0.01
0.04
0.04
0.09
0.06
0.06
0.06
0.13
0.07
0.06
0.04
0.01
0.06
0.03
0.21
0.02
0.03
0.01
0.04
0.01
0.03
0.01
0.04
0.01
0.01
0.05
0.02
0.1
0.01
0.03
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Lake Name
PARKER
PAUGUS BAY- STN 1
PAWTUCKAWAY LAKE
PEMIGEWASSET LAKE
PERCH (BENSON)
POWWOW POND
ROUND POND
SABIN
SILVER LAKE
SOMERSET
SPRUCE POND
SUNCOOK POND- UPPER
SUNSET (BRKFLD)
TRIO PONDS- ONE AND TWO
WALKER POND
WILLOUGHBY
WOLCOTT
ZEPHYR LAKE
Observed
0.11
0.21
0.29
0.163
0.3
0.175
0.37
0.135
0.237
0.256
0.37
0.257
0.19
0.608
0.225
0.128
0.215
0.14
1/2
0.09
0.01
0.07
0
0.2
0.06
0.1
0.03
0.02
0.1
0.02
0.02
0.01
0.1
0.04
0.01
0.07
0.02
1/3
0.1
0.01
0.07
0
0.2
0.06
0.11
0.03
0.02
0.1
0.02
0.02
0.02
0.11
0.04
0.01
0.08
0.02
1/4
0.1
0.01
0.07
0
0.2
0.06
0.12
0.03
0.02
0.1
0.02
0.02
0.02
0.12
0.04
0.01
0.08
0.02
1/5
0.1
0.01
0.07
0
0.2
0.06
0.12
0.03
0.02
0.1
0.02
0.02
0.02
0.12
0.05
0.01
0.08
0.02
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-14. Predicted Fish Mercury Concentrations for the Hypolimnion Surface
Area Sensitivity Analysis. Data are presented in columns of observed, and then for the
fraction used to estimate the hypolimnion surface area (e.g., 1/2 data corresponds to a
hypolimnion area =1/2 lake surface area) in the model.
Lake Name
CURTIS
GREAT HOSMER
WILLOUGHBY
CRANBERRY MEADOW
NEWARK
DUNMORE
ZEPHYR LAKE
ECHO (HUBDTN)
LOVELL LAKE- STN 1
SABIN
HORTONIA
LYFORD
SUNSET (BRKFLD)
FERN
WALKER POND
LOON LAKE
PERCH (BENSON)
PAWTUCKAWAY LAKE
CHITTENDEN
HORN POND
JENNESS POND
NORTH (BRKFLD)
ARMINGTON LAKE
PARKER
ROUND POND
ECHO (CHARTN)
POWWOW POND
BEARCAMP POND
PEMIGEWASSET LAKE
SUNCOOK POND- UPPER
ISLAND POND
SILVER LAKE
Observed
0.046
0.049
0.068
0.074
0.076
0.111
0.112
0.117
0.117
0.128
0.132
0.156
0.156
0.163
0.172
0.189
0.19
0.204
0.21
0.215
0.215
0.225
0.231
0.233
0.233
0.276
0.308
0.321
0.331
0.331
0.382
0.462
1/2
0.04
0.03
0.04
0.08
0.05
0.06
0.18
0.07
0.13
0.06
0.05
0.06
0.02
0.03
0.08
0.16
0.11
0.06
0.16
0.1
0.14
0.06
0.11
0.04
0.09
0.07
0.01
0.06
0.09
0.06
0.01
0.06
1/3
0.07
0.05
0.05
0.11
0.07
0.08
0.27
0.1
0.19
0.08
0.07
0.08
0.03
0.03
0.11
0.24
0.17
0.08
0.23
0.13
0.21
0.08
0.16
0.05
0.13
0.09
0.01
0.07
0.12
0.07
0.01
0.08
1/4
0.08
0.06
0.05
0.12
0.08
0.09
0.33
0.11
0.23
0.1
0.08
0.1
0.03
0.04
0.12
0.29
0.21
0.09
0.28
0.14
0.25
0.09
0.2
0.06
0.15
0.1
0.01
0.08
0.14
0.08
0.01
0.09
1/5
0.1
0.07
0.06
0.13
0.08
0.09
0.37
0.13
0.25
0.1
0.08
0.11
0.03
0.04
0.13
0.33
0.24
0.1
0.31
0.16
0.27
0.09
0.22
0.07
0.17
0.11
0.01
0.08
0.15
0.08
0.01
0.1
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Lake Name
SOMERSET
WOLCOTT
Observed
0.463
0.696
1/2
0.06
0.14
1/3
0.08
0.19
1/4
0.1
0.22
1/5
0.11
0.23
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-15. Epilimnetic Methylmercury Concentration Sensitivity to Change in
Hypolimnion Surface Area/Epilimnion Epilimnion Area. Percent change is calculated
from the Tier 1 model results with the hypolimnion surface area to epilimnion surface
area ratio, R = 1/3. The percent changes presented are for the other area ratios. The
averages for all other ratios are presented in the fourth column (average percent change
for that given lake). Also, the overall average for each ratio is presented at the bottom of
each column, along with the median and largest absolute change.
A(C,l/2)
-90.91%
-50.00%
-40.00%
0.00%
-44.44%
-60.00%
-66.67%
-100.00%
0.00%
-50.00%
-66.67%
-50.00%
0.00%
-100.00%
-40.00%
-85.71%
0.00%
-66.67%
0.00%
0.00%
-40.00%
0.00%
0.00%
-40.00%
-66.67%
-50.00%
-50.00%
0.00%
-75.00%
0.00%
-66.67%
A (C,l/4)
-145.45%
0.00%
-160.00%
0.00%
-88.89%
-80.00%
0.00%
-66.67%
0.00%
-100.00%
0.00%
-100.00%
-400.00%
-66.67%
-80.00%
-114.29%
-400.00%
0.00%
-133.33%
0.00%
-80.00%
0.00%
0.00%
-80.00%
-44.44%
-100.00%
-100.00%
0.00%
-100.00%
-200.00%
0.00%
A (C,l/5)
-159.09%
-62.50%
-100.00%
-125.00%
-111.11%
-75.00%
0.00%
-125.00%
0.00%
-125.00%
0.00%
-125.00%
-250.00%
-83.33%
-100.00%
-142.86%
-250.00%
0.00%
-83.33%
0.00%
-50.00%
0.00%
0.00%
-100.00%
-55.56%
-62.50%
-62.50%
0.00%
-93.75%
-125.00%
0.00%
Average A
-131.82%
-37.50%
-100.00%
-41.67%
-81.48%
-71.67%
-22.22%
-97.22%
0.00%
-91.67%
-22.22%
-91.67%
-216.67%
-83.33%
-73.33%
-114.29%
-216.67%
-22.22%
-72.22%
0.00%
-56.67%
0.00%
0.00%
-73.33%
-55.56%
-70.83%
-70.83%
0.00%
-89.58%
-108.33%
-22.22%
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Mean
Abs. Max. A
Median
0.00%
-66.67%
0.00%
0.00%
-50.00%
-75.00%
0.00%
0.00%
0.00%
0.00%
-66.67%
-90.91%
-66.67%
-200.00%
-57.14%
-50.00%
0.00%
-50.00%
-72.73%
-42.90%
-200.00%
-50.00%
0.00%
0.00%
-133.33%
-133.33%
-100.00%
-50.00%
0.00%
-133.33%
-200.00%
0.00%
0.00%
-145.45%
0.00%
0.00%
-57.14%
0.00%
-400.00%
-50.00%
-72.73%
-82.30%
-400.00%
-69.70%
-83.33%
0.00%
-83.33%
-83.33%
-62.50%
-62.50%
0.00%
-83.33%
-125.00%
-125.00%
0.00%
-159.09%
0.00%
0.00%
-71.43%
0.00%
-250.00%
-62.50%
-68.18%
-75.72%
-250.00%
-73.21%
-27.78%
-22.22%
-72.22%
-72.22%
-70.83%
-62.50%
0.00%
-72.22%
-108.33%
-41.67%
-22.22%
-131.82%
-22.22%
-66.67%
-61.90%
-16.67%
-216.67%
-54.17%
-71.21%
-66.98%
-283.33%
-70.83%
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-16. Epilimnetic Total Mercury Concentration Sensitivity to Change in
Hypolimnion Surface Area/Epilimnion Epilimnion Area. Percent change is calculated
from the Tier 1 model results with the hypolimnion surface area to epilimnion surface
area ratio, R = 1/3. The percent changes presented are for the other area ratios. The
averages for all other ratios are presented in the fourth column (average percent change
for that given lake). Also, the overall average for each ratio is presented at the bottom of
each column, along with the median and largest absolute change.
A(C,l/2)
-3.92%
-9.09%
-7.41%
-17.39%
-8.51%
-16.67%
-5.80%
-20.83%
-4.88%
-21.74%
-17.91%
-6.06%
-9.68%
-18.80%
-21.78%
-19.05%
-15.19%
-21.43%
-7.27%
-17.74%
-11.43%
-4.88%
0.00%
-10.67%
-12.12%
-13.91%
-19.80%
-11.27%
-10.81%
-32.73%
A (C,l/4)
-15.69%
-18.18%
-7.41%
-23.19%
-8.51%
-20.00%
-5.80%
-33.33%
-9.76%
-26.09%
-23.88%
-12.12%
-6.45%
-23.93%
-23.76%
-28.57%
-15.19%
-21.43%
-14.55%
-19.35%
-11.43%
-9.76%
0.00%
-10.67%
-12.12%
-17.39%
-19.80%
-16.90%
-10.81%
-50.91%
A (C,l/5)
-9.80%
-11.36%
-4.63%
-25.36%
-5.32%
-20.83%
-7.25%
-31.25%
-6.10%
-32.61%
-22.39%
-7.58%
-8.06%
-23.50%
-24.75%
-26.79%
-18.99%
-26.79%
-9.09%
-18.15%
-14.29%
-6.10%
-15.63%
-13.33%
-15.15%
-19.57%
-22.28%
-14.08%
-10.14%
-59.09%
Average A
-9.80%
-12.88%
-6.48%
-21.98%
-7.45%
-19.17%
-6.28%
-28.47%
-6.91%
-26.81%
-21.39%
-8.59%
-8.06%
-22.08%
-23.43%
-24.80%
-16.46%
-23.21%
-10.30%
-18.41%
-12.38%
-6.91%
-5.21%
-11.56%
-13.13%
-16.96%
-20.63%
-14.08%
-10.59%
-47.58%
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Mean
Abs. Max. A
Median
-11.76%
0.00%
-20.51%
-27.45%
-9.76%
-21.94%
-19.51%
-20.25%
-12.50%
-12.70%
-17.72%
-30.56%
-6.45%
-15.38%
-18.49%
-25.45%
-20.20%
-16.98%
-15.58%
4.08%
-14.36%
-32.73%
-15.29%
-11.76%
0.00%
-30.77%
-23.53%
0.00%
-23.23%
-19.51%
-25.32%
-16.67%
-12.70%
-15.19%
-44.44%
0.00%
-24.62%
-23.53%
-29.09%
-24.24%
-18.87%
-15.58%
0.00%
-17.52%
-50.91%
-17.15%
-12.25%
0.00%
-32.05%
-29.41%
-6.10%
-25.81%
-24.39%
-25.32%
-15.63%
-11.90%
-18.99%
-48.61%
0.00%
-23.08%
-23.11%
-31.82%
-25.25%
-18.87%
-16.23%
0.00%
-18.38%
-59.09%
-18.51%
-11.93%
0.00%
-27.78%
-26.80%
-5.28%
-23.66%
-21.14%
-23.63%
-14.93%
-12.43%
-17.30%
-41.20%
-2.15%
-21.03%
-21.71%
-28.79%
-23.23%
-18.24%
-15.80%
1.36%
-16.75%
-47.58%
-16.71%
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-17. Hypolimnetic Methylmercury Concentration Sensitivity to Change in
Hypolimnion Surface Area/Epilimnion Epilimnion Area. Percent change is calculated
from the Tier 1 model results with the hypolimnion surface area to epilimnion surface
area ratio, R = 1/3. The percent changes presented are for the other area ratios. The
averages for all other ratios are presented in the fourth column (average percent change
for that given lake). Also, the overall average for each ratio is presented at the bottom of
each column, along with the median and largest absolute change.
A (C,l/2)
-79.05%
-50.69%
-28.18%
-60.63%
-53.71%
30.32%
-63.12%
-36.26%
-54.05%
-16.67%
-70.27%
-48.35%
-76.09%
-37.07%
-32.14%
-45.41%
-22.22%
-31.58%
-41.89%
-52.81%
-53.33%
-47.02%
-67.04%
-33.55%
-34.15%
-41.53%
-39.25%
-35.20%
-37.97%
-59.97%
A(C,l/4)
-124.14%
-80.44%
-30.91%
-85.48%
-68.74%
-101.94%
-86.38%
-41.76%
-70.87%
-33.33%
-100.90%
-61.98%
-116.30%
-41.38%
-35.71%
-57.64%
-22.22%
-38.60%
-51.35%
-66.53%
-66.67%
-57.68%
-94.97%
-41.29%
-39.02%
-48.09%
-45.73%
-43.20%
-45.99%
-80.37%
A (C,l/5)
-138.59%
-85.40%
-31.82%
-92.62%
-72.23%
-111.29%
-93.85%
-42.58%
-75.08%
-31.25%
-111.49%
-65.08%
-128.40%
-43.10%
-36.61%
-60.04%
-27.78%
-37.28%
-52.36%
-70.69%
-70.40%
-59.56%
-103.35%
-41.94%
-40.24%
-49.18%
-47.35%
-44.00%
-46.79%
-85.39%
Average
-113.93%
-72.18%
-30.30%
-79.58%
-64.89%
-60.97%
-81.12%
-40.20%
-66.67%
-27.08%
-94.22%
-58.47%
-106.93%
-40.52%
-34.82%
-54.37%
-24.07%
-35.82%
-48.54%
-63.34%
-63.47%
-54.75%
-88.45%
-38.92%
-37.80%
-46.27%
-44.11%
-40.80%
-43.58%
-75.24%
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Mean
Abs. Max. A
Median
A (C,l/2)
0.00%
-47.10%
-41.84%
-38.82%
-50.91%
-84.48%
-27.69%
-31.91%
-49.48%
-33.86%
-28.57%
-39.19%
-55.77%
-42.99%
-84.48%
-41.84%
A(C,l/4)
0.00%
-58.70%
-50.21%
-47.06%
-65.45%
-145.81%
-28.72%
-38.30%
-62.54%
-38.87%
-34.29%
-45.95%
-72.11%
-59.71%
-145.81%
-51.35%
A(C,l/5)
0.00%
-61.43%
-52.30%
-48.53%
-70.00%
-168.72%
-30.77%
-37.23%
-65.72%
-39.97%
-33.33%
-47.30%
-76.29%
-63.43%
-168.72%
-52.36%
Average
0.00%
-55.75%
-48.12%
-44.80%
-62.12%
-133.00%
-29.06%
-35.82%
-59.25%
-37.57%
-32.06%
-44.14%
-68.06%
-55.38%
-133.00%
-48.54%
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-18. Hypolimnetic Total Mercury Concentration Sensitivity to Change in
Hypolimnion Surface Area/Epilimnion Epilimnion Area. Percent change is calculated
from the Tier 1 model results with the hypolimnion surface area to epilimnion surface
area ratio, R = 1/3. The percent changes presented are for the other area ratios. The
averages for all other ratios are presented in the fourth column (average percent change
for that given lake). Also, the overall average for each ratio is presented at the bottom of
each column, along with the median and largest absolute change.
A(C,l/2)
-66.50%
-85.81%
-8.64%
-32.33%
-32.94%
0.49%
-43.14%
-7.04%
-20.73%
-11.47%
-43.38%
-27.11%
-50.36%
-5.44%
-10.29%
-14.61%
10.21%
-8.38%
-9.54%
-28.54%
-33.80%
-19.58%
-42.96%
-10.24%
-11.70%
-11.85%
-12.50%
-15.79%
-19.38%
-33.74%
A(C,l/4)
-102.60%
-50.82%
-10.01%
-43.02%
-40.78%
-67.24%
-58.25%
-7.85%
-26.28%
-12.24%
-60.67%
-33.74%
-73.72%
-6.87%
-11.79%
-17.88%
10.21%
-9.05%
-11.10%
-35.35%
-41.68%
-23.44%
-59.14%
-11.29%
-13.78%
-13.62%
-15.24%
-18.53%
-22.73%
-43.64%
A(C,l/5)
-114.25%
-54.07%
-10.23%
-46.11%
-42.81%
-72.93%
-62.37%
-8.02%
-27.78%
-12.91%
-65.90%
-35.34%
-81.04%
-7.16%
-11.92%
-18.58%
10.20%
-9.22%
-11.49%
-37.05%
-43.67%
-24.32%
-63.63%
-11.49%
-14.22%
-14.13%
-15.96%
-19.24%
-23.47%
-46.30%
Average
-94.45%
-63.57%
-9.63%
-40.49%
-38.84%
-46.56%
-54.59%
-7.64%
-24.93%
-12.21%
-56.65%
-32.06%
-68.37%
-6.49%
-11.33%
-17.02%
10.20%
-8.89%
-10.71%
-33.64%
-39.72%
-22.45%
-55.24%
-11.01%
-13.23%
-13.20%
-14.57%
-17.85%
-21.86%
-41.22%
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Mean
Abs. Max. A
Median
A(C,l/2)
-5.56%
-24.40%
-10.92%
-10.82%
-23.90%
-92.84%
-5.80%
7.90%
-26.95%
-14.60%
-6.97%
-21.50%
-40.28%
-22.88%
-92.84%
-15.79%
A(C,l/4)
-7.07%
-29.03%
-12.61%
-12.47%
-30.75%
-30.36%
-6.77%
7.56%
-33.00%
-17.05%
-7.84%
-24.89%
-51.11%
-27.62%
-102.60%
-22.73%
A(C,l/5)
-6.31%
-30.31%
-13.01%
-12.83%
-32.53%
-71.09%
-6.87%
7.52%
-34.65%
-17.52%
-8.71%
-25.58%
-53.81%
-30.26%
-114.25%
-23.47%
Average
-6.31%
-27.92%
-12.18%
-12.04%
-29.06%
-64.76%
-6.48%
7.66%
-31.53%
-16.39%
-7.84%
-23.99%
-48.40%
-26.92%
-103.23%
-21.86%
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-19. Sediment Methylmercury Concentration Sensitivity to Change in
Hypolimnion Surface Area/Epilimnion Epilimnion Area. Percent change is calculated
from the Tier 1 model results with the hypolimnion surface area to epilimnion surface
area ratio, R = 1/3. The percent changes presented are for the other area ratios. The
averages for all other ratios are presented in the fourth column (average percent change
for that given lake). Also, the overall average for each ratio is presented at the bottom of
each column, along with the median and largest absolute change.
Mean
Abs. Max. A
Median
A (C,l/2)
-66.67%
200.00%
0.00%
-58.82%
-50.00%
0.00%
-100.00%
0.00%
0.00%
0.00%
-100.00%
0.00%
-177.78%
-100.00%
-66.67%
0.00%
0.00%
-50.00%
0.00%
-66.67%
-66.67%
0.00%
0.00%
-50.00%
0.00%
0.00%
0.00%
0.00%
-40.00%
0.00%
-66.67%
0.00%
0.00%
0.00%
-100.00%
-200.00%
-50.00%
0.00%
0.00%
0.00%
-40.00%
-30.49%
-200.00%
0.00%
A (C,l/4)
-133.33%
0.00%
0.00%
-94.12%
0.00%
-400.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
355.56%
0.00%
0.00%
0.00%
0.00%
-100.00%
0.00%
-66.67%
0.00%
-400.00%
0.00%
-100.00%
0.00%
0.00%
0.00%
0.00%
0.00%
-400.00%
0.00%
-200.00%
0.00%
-400.00%
0.00%
0.00%
-100.00%
0.00%
0.00%
0.00%
-80.00%
-51.67%
-400.00%
0.00%
A (C,l/5)
-166.67%
0.00%
0.00%
-88.24%
-62.50%
-250.00%
0.00%
0.00%
0.00%
-250.00%
0.00%
0.00%
222.22%
-125.00%
0.00%
0.00%
0.00%
-62.50%
-250.00%
-83.33%
0.00%
-250.00%
-125.00%
-125.00%
0.00%
0.00%
0.00%
0.00%
-50.00%
0.00%
83.33%
-125.00%
-250.00%
-250.00%
0.00%
0.00%
-125.00%
0.00%
0.00%
-125.00%
-100.00%
-62.38%
-250.00%
0.00%
Average
-122.22%
66.67%
0.00%
-80.39%
-37.50%
-216.67%
-33.33%
0.00%
0.00%
-83.33%
-33.33%
0.00%
133.33%
-75.00%
-22.22%
0.00%
0.00%
-70.83%
-83.33%
-72.22%
-22.22%
-216.67%
-41.67%
-91.67%
0.00%
0.00%
0.00%
0.00%
-30.00%
-133.33%
5.56%
-108.33%
-83.33%
-216.67%
-33.33%
-66.67%
-91.67%
0.00%
0.00%
-41.67%
-73.33%
-48.18%
-283.33%
-33.33%
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-20. Sediment Total Mercury Concentration Sensitivity to Change in
Hypolimnion Surface Area/Epilimnion Epilimnion Area. Percent change is calculated
from the Tier 1 model results with the hypolimnion surface area to epilimnion surface
area ratio, R = 1/3. The percent changes presented are for the other area ratios. The
averages for all other ratios are presented in the fourth column (average percent change
for that given lake). Also, the overall average for each ratio is presented at the bottom of
each column, along with the median and largest absolute change.
A (C,l/2)
0.00%
-22.22%
0.00%
0.00%
-50.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
A(C,l/4)
0.00%
-44.44%
0.00%
0.00%
0.00%
-50.00%
0.00%
0.00%
0.00%
0.00%
-66.67%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
-100.00%
0.00%
0.00%
A (C,l/5)
0.00%
-27.78%
0.00%
0.00%
0.00%
-31.25%
0.00%
0.00%
0.00%
0.00%
-41.67%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
-250.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
-62.50%
0.00%
0.00%
Average
0.00%
-31.48%
0.00%
0.00%
-16.67%
-27.08%
0.00%
0.00%
0.00%
0.00%
-36.11%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
-83.33%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
-54.17%
0.00%
0.00%
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Mean
Abs. Max. A
Median
0.00%
0.00%
-20.00%
0.00%
0.00%
0.00%
0.00%
0.00%
-18.18%
0.00%
0.00%
0.00%
0.00%
0.00%
-100.00%
-18.18%
0.00%
0.00%
-25.00%
0.00%
-5.07%
-100.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
-36.36%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
-36.36%
0.00%
0.00%
0.00%
0.00%
-6.68%
-100.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
-22.73%
0.00%
0.00%
0.00%
0.00%
0.00%
0.00%
-22.73%
-62.50%
0.00%
0.00%
0.00%
-10.42%
-250.00%
0.00%
0.00%
0.00%
-6.67%
0.00%
0.00%
0.00%
0.00%
0.00%
-25.76%
0.00%
0.00%
0.00%
0.00%
0.00%
-33.33%
-25.76%
-20.83%
0.00%
-8.33%
0.00%
-7.39%
-150.00%
0.00%
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-21. Fish Tissue Mercury Concentration Sensitivity to Change in
Hypolimnion Surface Area/Epilimnion Epilimnion Area. Percent change is calculated
from the Tier 1 model results with the hypolimnion surface area to epilimnion surface
area ratio, R = 1/3. The percent changes presented are for the other area ratios. The
averages for all other ratios are presented in the fourth column (average percent change
for that given lake). Also, the overall average for each ratio is presented at the bottom of
each column, along with the median and largest absolute change.
A(C,l/2)
-85.71%
-80.00%
-40.00%
-54.55%
-57.14%
-50.00%
-66.67%
-60.00%
-63.16%
-50.00%
-57.14%
-50.00%
-66.67%
0.00%
-54.55%
-66.67%
-70.59%
-50.00%
-60.87%
-46.15%
-66.67%
-50.00%
-62.50%
-40.00%
-61.54%
-44.44%
0.00%
-28.57%
-50.00%
-28.57%
A(C,l/4)
-57.14%
-80.00%
0.00%
-36.36%
-57.14%
-50.00%
-88.89%
-40.00%
-84.21%
-100.00%
-57.14%
-100.00%
0.00%
-133.33%
-36.36%
-83.33%
-94.12%
-50.00%
-86.96%
-30.77%
-76.19%
-50.00%
-100.00%
-80.00%
-61.54%
-44.44%
0.00%
-57.14%
-66.67%
-57.14%
A (C,l/5)
-107.14%
-100.00%
-50.00%
-45.45%
-35.71%
-31.25%
-92.59%
-75.00%
-78.95%
-62.50%
-35.71%
-93.75%
0.00%
-83.33%
-45.45%
-93.75%
-102.94%
-62.50%
-86.96%
-57.69%
-71.43%
-31.25%
-93.75%
-100.00%
-76.92%
-55.56%
0.00%
-35.71%
-62.50%
-35.71%
Average
-83.33%
-86.67%
-30.00%
-45.45%
-50.00%
-43.75%
-82.72%
-58.33%
-75.44%
-70.83%
-50.00%
-81.25%
-22.22%
-72.22%
-45.45%
-81.25%
-89.22%
-54.17%
-78.26%
-44.87%
-71.43%
-43.75%
-85.42%
-73.33%
-66.67%
-48.15%
0.00%
-40.48%
-59.72%
-40.48%
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Mean
Abs. Max. A
Median
0.00%
-50.00%
-50.00%
-52.63%
-50.43%
-85.71%
-51.32%
0.00%
-50.00%
-100.00%
-63.16%
-60.94%
-133.33%
-57.14%
0.00%
-62.50%
-93.75%
-52.63%
-62.13%
-107.14%
-62.50%
0.00%
-54.17%
-81.25%
-56.14%
-57.84%
-108.73%
-57.24%
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-22. Epilimnion MeHg and HgT Concentrations for Range of RlUp and R2Up
Values.
R1Up
R2Up
Lake
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
0.1
0.1
MeHg
0.11
0.04
0.05
0.06
0.02
0.09
0.11
0.02
0.02
0.51
0.1
0.07
0.03
0.06
0.1
0.02
0.22
0.04
0.03
0.04
0.02
0.01
0.04
0.03
0.01
0.06
0.14
0.05
0.03
0.31
0.07
0.01
0.03
0.03
0.08
0
0.17
0.04
0.05
0.05
0.09
0.1
0.42
0.03
0.02
0.05
0.1
0.1
HgT
0.72
0.84
0.56
0.72
0.66
1.55
1.34
0.84
1.02
1.9
1.01
1.37
0.75
0.48
1.28
0.47
2.18
0.39
0.69
0.51
0.63
0.35
0.95
0.43
0.42
0.46
0.57
0.79
0.74
2.13
0.51
0.48
1.02
0.41
1.39
0.08
1.81
1.11
0.99
0.74
0.77
1.32
0.98
0.54
0.33
1.17
0.1
1
MeHg
0.33
0.09
0.21
0.24
0.07
0.19
0.28
0.08
0.05
1.47
0.29
0.23
0.07
0.16
0.33
0.04
0.43
0.13
0.04
0.09
0.02
0.02
0.14
0.1
0.02
0.11
0.32
0.14
0.09
0.62
0.18
0.06
0.06
0.06
0.27
0
0.32
0.08
0.1
0.16
0.46
0.4
0.8
0.05
0.03
0.18
0.1
1
HgT
2.25
2.58
2.54
3.74
3.79
3.72
3.98
5.44
5.28
5.58
3.3
7.09
3.15
1.44
5.86
1.42
4.35
1.37
1.78
1.47
2.21
0.84
5.5
2.59
3.02
1.04
1.34
2.44
4.49
4.51
1.41
3.51
3.71
0.98
5.9
0.37
3.63
6.38
2.46
3
4.53
6.61
1.91
1.23
0.91
5.13
0.1
2
MeHg
0.58
0.16
0.38
0.44
0.12
0.31
0.46
0.14
0.09
2.54
0.5
0.41
0.12
0.27
0.58
0.06
0.66
0.22
0.05
0.15
0.02
0.03
0.25
0.18
0.02
0.18
0.52
0.23
0.17
0.97
0.31
0.11
0.09
0.08
0.49
0
0.5
0.13
0.15
0.29
0.87
0.74
1.23
0.08
0.04
0.32
0.1
2
HgT
3.95
4.51
4.75
7.1
7.27
6.13
6.92
10.56
10.02
9.67
5.86
13.45
5.81
2.49
10.96
2.49
6.76
2.45
2.98
2.52
3.97
1.34
10.57
4.98
5.9
1.67
2.19
4.28
8.65
7.15
2.41
6.87
6.69
1.61
10.9
0.68
5.66
12.24
4.09
5.52
8.71
12.49
2.94
2
1.56
9.52
1
0.1
MeHg
0.13
0.05
0.08
0.09
0.06
0.12
0.14
0.08
0.08
0.55
0.13
0.13
0.06
0.07
0.17
0.04
0.24
0.05
0.05
0.07
0.04
0.03
0.09
0.05
0.04
0.06
0.15
0.08
0.09
0.33
0.08
0.05
0.07
0.05
0.13
0.01
0.18
0.1
0.06
0.08
0.13
0.15
0.43
0.04
0.03
0.09
1
0.1
HgT
0.74
0.86
0.59
0.76
0.7
1.57
1.37
0.89
1.07
1.94
1.04
1.43
0.78
0.5
1.34
0.49
2.19
0.41
0.72
0.54
0.66
0.39
1.01
0.46
0.45
0.47
0.57
0.81
0.8
2.16
0.52
0.52
1.06
0.43
1.44
0.08
1.82
1.17
1.01
0.77
0.82
1.37
0.99
0.55
0.35
1.21
-------
Evaluating R-MCMfor 91 VT/NH Lakes
R1Up
R2Up
Lake
Number
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
0.1
0.1
MeHg
0.09
0.04
0.04
0.02
0.06
0.23
0.08
0.31
0.02
0.02
0.02
0.03
0.03
0.03
0.06
0.03
0.03
0.03
0.04
0.08
0.07
0.39
0.01
0.03
0.03
0.02
0.12
0.02
0.03
0.12
0.11
0.49
0.03
0.35
0.01
0.07
0.11
0.07
0.04
0.17
0.08
0.01
0.18
0.08
0.11
0.1
0.1
HgT
0.69
0.55
1.24
0.63
1.28
1.37
0.55
4.67
0.57
0.66
0.41
0.62
0.77
1.09
1.47
0.41
1.06
0.65
0.79
0.6
0.75
0.94
0.49
0.93
0.67
0.71
1.67
0.31
0.41
1.69
0.55
2.03
0.74
1.86
0.22
1.19
0.57
1.03
0.79
0.85
0.65
0.16
1.04
1.15
1.01
0.1
1
MeHg
0.19
0.09
0.12
0.04
0.27
0.66
0.15
1.4
0.07
0.09
0.03
0.07
0.08
0.06
0.16
0.06
0.07
0.08
0.12
0.14
0.23
0.87
0.01
0.1
0.06
0.05
0.39
0.03
0.06
0.4
0.17
1.13
0.05
0.94
0.01
0.22
0.43
0.21
0.11
0.54
0.14
0.02
0.59
0.21
0.35
0.1
1
HgT
1.6
1.49
4.84
2.54
7.54
4.06
1.14
29.63
2.64
4.47
0.76
2.6
3.15
3.73
8.58
1.61
3.83
2.15
2.76
1.18
3.04
2.12
2.93
5.3
1.35
2.75
6.14
0.6
1.17
6.64
0.9
4.82
4.18
5.22
1.22
4.34
2.39
3.9
3.87
2.74
1.86
0.38
3.73
3.7
3.69
0.1
2
MeHg
0.29
0.14
0.2
0.06
0.5
1.13
0.23
2.62
0.12
0.17
0.04
0.12
0.13
0.1
0.27
0.1
0.11
0.12
0.2
0.21
0.42
1.4
0.01
0.18
0.08
0.08
0.68
0.04
0.1
0.72
0.24
1.84
0.08
1.59
0.02
0.38
0.78
0.37
0.19
0.95
0.22
0.02
1.05
0.36
0.61
0.1
2
HgT
2.6
2.54
8.85
4.66
14.51
7.05
1.8
57.36
4.93
8.71
1.15
4.8
5.79
6.65
16.48
2.94
6.91
3.81
4.96
1.83
5.58
3.42
5.51
10.17
2.1
5.02
11.11
0.93
2.02
12.14
1.29
7.91
8
8.94
2.33
7.83
4.41
7.09
7.29
4.84
3.21
0.61
6.73
6.53
6.68
1
0.1
MeHg
0.1
0.06
0.08
0.04
0.12
0.26
0.09
0.79
0.04
0.07
0.03
0.05
0.06
0.06
0.14
0.06
0.07
0.06
0.07
0.09
0.1
0.41
0.05
0.08
0.04
0.05
0.16
0.03
0.04
0.17
0.11
0.52
0.07
0.38
0.02
0.1
0.12
0.1
0.08
0.2
0.12
0.03
0.21
0.11
0.14
1
0.1
HgT
0.71
0.57
1.27
0.65
1.34
1.4
0.56
5.19
0.59
0.7
0.42
0.65
0.79
1.12
1.56
0.44
1.1
0.68
0.81
0.61
0.78
0.96
0.65
0.97
0.68
0.74
1.71
0.32
0.42
1.74
0.56
2.06
0.79
1.9
0.23
1.22
0.59
1.06
0.84
0.87
0.69
0.18
1.08
1.18
1.04
-------
Evaluating R-MCMfor 91 VT/NH Lakes
FMUp
R2Up
Lake
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
1
1
MeHg
0.35
0.11
0.24
0.28
0.11
0.22
0.31
0.14
0.11
1.51
0.32
0.29
0.1
0.17
0.4
0.06
0.44
0.14
0.06
0.12
0.04
0.03
0.19
0.12
0.05
0.12
0.33
0.16
0.15
0.65
0.2
0.09
0.09
0.07
0.32
0.01
0.34
0.14
0.11
0.19
0.5
0.45
0.82
0.07
0.05
0.21
0.2
0.11
1
1
HgT
2.27
2.6
2.57
3.78
3.83
3.74
4.01
5.5
5.34
5.62
3.34
7.15
3.18
1.45
5.93
1.44
4.36
1.38
1.8
1.49
2.24
0.85
5.56
2.61
3.05
1.05
1.34
2.46
4.54
4.54
1.42
3.54
3.75
1
5.95
0.37
3.64
6.44
2.48
3.03
4.58
6.67
1.92
1.25
0.93
5.17
1.61
1.51
1
2
MeHg
0.59
0.17
0.41
0.48
0.16
0.33
0.49
0.2
0.14
2.58
0.53
0.47
0.15
0.28
0.65
0.08
0.67
0.24
0.08
0.17
0.05
0.04
0.3
0.2
0.06
0.19
0.53
0.25
0.22
1
0.32
0.14
0.12
0.1
0.54
0.01
0.51
0.2
0.17
0.32
0.91
0.79
1.25
0.09
0.06
0.35
0.31
0.16
1
2
HgT
3.97
4.52
4.78
7.14
7.31
6.16
6.95
10.62
10.07
9.71
5.89
13.51
5.85
2.51
11.03
2.5
6.78
2.46
3.01
2.55
4
1.36
10.62
5
5.93
1.68
2.2
4.3
8.71
7.18
2.42
6.9
6.73
1.63
10.95
0.69
5.67
12.3
4.11
5.55
8.75
12.55
2.95
2.02
1.57
9.56
2.62
2.56
2
0.1
MeHg
0.15
0.07
0.12
0.14
0.11
0.14
0.18
0.14
0.14
0.59
0.16
0.2
0.09
0.08
0.24
0.06
0.25
0.07
0.08
0.1
0.07
0.04
0.15
0.08
0.08
0.07
0.16
0.1
0.15
0.36
0.09
0.09
0.11
0.07
0.19
0.01
0.19
0.17
0.08
0.11
0.18
0.21
0.44
0.06
0.05
0.13
0.12
0.08
2
0.1
HgT
0.76
0.88
0.63
0.8
0.75
1.59
1.41
0.95
1.13
1.98
1.07
1.5
0.81
0.51
1.42
0.51
2.21
0.42
0.75
0.57
0.69
0.4
1.07
0.49
0.49
0.48
0.58
0.84
0.86
2.19
0.53
0.56
1.1
0.45
1.5
0.09
1.83
1.24
1.02
0.8
0.87
1.44
1.01
0.57
0.37
1.25
0.73
0.59
2
1
MeHg
0.37
0.13
0.27
0.32
0.15
0.24
0.35
0.2
0.17
1.55
0.35
0.36
0.13
0.18
0.47
0.08
0.46
0.15
0.09
0.15
0.07
0.04
0.25
0.15
0.08
0.13
0.34
0.18
0.21
0.68
0.21
0.13
0.14
0.09
0.38
0.01
0.35
0.21
0.13
0.22
0.55
0.51
0.83
0.08
0.06
0.25
0.22
0.13
2
1
HgT
2.29
2.62
2.61
3.82
3.88
3.76
4.05
5.56
5.4
5.66
3.37
7.22
3.21
1.46
6
1.46
4.38
1.4
1.83
1.52
2.27
0.86
5.62
2.64
3.08
1.06
1.35
2.49
4.61
4.57
1.44
3.58
3.79
1.02
6
0.37
3.66
6.51
2.49
3.06
4.62
6.73
1.94
1.26
0.95
5.21
1.63
1.53
-------
Evaluating R-MCMfor 91 VT/NH Lakes
R1Up
R2Up
Lake
Number
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
1
1
MeHg
0.15
0.06
0.33
0.69
0.16
1.88
0.09
0.14
0.04
0.1
0.1
0.09
0.25
0.09
0.11
0.1
0.14
0.15
0.26
0.89
0.05
0.15
0.07
0.08
0.43
0.04
0.07
0.45
0.17
1.16
0.1
0.97
0.02
0.25
0.44
0.24
0.15
0.56
0.18
0.03
0.62
0.24
0.38
1
1
HgT
4.88
2.56
7.61
4.09
1.15
30.15
2.66
4.52
0.77
2.63
3.17
3.75
8.67
1.64
3.87
2.17
2.79
1.19
3.07
2.13
2.97
5.35
1.36
2.78
6.19
0.61
1.18
6.69
0.91
4.84
4.23
5.25
1.23
4.37
2.41
3.93
3.92
2.76
1.9
0.39
3.77
3.73
3.73
1
2
MeHg
0.24
0.08
0.57
1.16
0.24
3.1
0.14
0.21
0.05
0.15
0.16
0.13
0.36
0.13
0.15
0.14
0.23
0.22
0.45
1.42
0.05
0.23
0.09
0.11
0.72
0.05
0.11
0.77
0.24
1.86
0.13
1.62
0.03
0.41
0.8
0.4
0.24
0.97
0.26
0.03
1.08
0.39
0.64
1
2
HgT
8.88
4.68
14.57
7.08
1.81
57.88
4.95
8.75
1.16
4.83
5.81
6.68
16.57
2.98
6.95
3.83
4.98
1.85
5.62
3.44
5.54
10.21
2.11
5.05
11.15
0.94
2.04
12.19
1.29
7.94
8.05
8.97
2.34
7.86
4.43
7.12
7.34
4.87
3.25
0.63
6.76
6.56
6.72
2
0.1
MeHg
0.12
0.06
0.2
0.3
0.1
1.32
0.07
0.11
0.04
0.08
0.08
0.09
0.24
0.1
0.12
0.08
0.1
0.1
0.14
0.42
0.08
0.13
0.05
0.08
0.21
0.04
0.05
0.22
0.12
0.55
0.13
0.41
0.03
0.14
0.14
0.13
0.13
0.22
0.16
0.05
0.24
0.14
0.17
2
0.1
HgT
1.32
0.68
1.42
1.44
0.57
5.76
0.62
0.75
0.43
0.68
0.82
1.15
1.66
0.48
1.15
0.7
0.84
0.62
0.82
0.98
0.69
1.02
0.69
0.77
1.76
0.33
0.43
1.79
0.56
2.09
0.84
1.93
0.25
1.26
0.61
1.1
0.88
0.9
0.74
0.2
1.12
1.22
1.08
2
1
MeHg
0.2
0.08
0.41
0.72
0.17
2.42
0.11
0.18
0.05
0.12
0.13
0.12
0.34
0.13
0.15
0.12
0.17
0.16
0.3
0.9
0.09
0.2
0.07
0.11
0.47
0.05
0.08
0.5
0.18
1.18
0.15
1
0.04
0.28
0.46
0.27
0.2
0.59
0.23
0.05
0.66
0.27
0.41
2
1
HgT
4.92
2.59
7.68
4.13
1.17
30.72
2.68
4.56
0.78
2.66
3.2
3.78
8.77
1.68
3.91
2.19
2.82
1.21
3.11
2.15
3.01
5.4
1.37
2.81
6.23
0.63
1.2
6.74
0.91
4.87
4.28
5.28
1.25
4.4
2.43
3.97
3.96
2.79
1.95
0.41
3.81
3.76
3.76
-------
Evaluating R-MCMfor 91 VT/NH Lakes
FMUp
R2Up
Lake
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
2
2
MeHg
0.61
0.19
0.44
0.52
0.21
0.35
0.53
0.26
0.2
2.62
0.56
0.54
0.18
0.29
0.72
0.1
0.69
0.25
0.1
0.2
0.07
0.05
0.36
0.23
0.09
0.2
0.54
0.28
0.28
1.03
0.34
0.18
0.17
0.12
0.59
0.01
0.52
0.27
0.18
0.35
0.96
0.85
1.26
0.11
0.08
0.39
0.33
0.18
2
2
HgT
3.99
4.54
4.81
7.18
7.36
6.18
6.99
10.68
10.13
9.75
5.93
13.58
5.88
2.52
11.1
2.53
6.8
2.48
3.04
2.58
4.02
1.37
10.69
5.03
5.97
1.7
2.21
4.32
8.77
7.21
2.44
6.94
6.78
1.65
11.01
0.69
5.69
12.37
4.13
5.58
8.8
12.61
2.97
2.03
1.59
9.6
2.63
2.58
5
0.1
MeHg
0.2
0.13
0.21
0.27
0.25
0.21
0.29
0.32
0.31
0.7
0.27
0.4
0.19
0.12
0.45
0.12
0.3
0.11
0.16
0.18
0.15
0.07
0.34
0.17
0.18
0.1
0.18
0.17
0.33
0.45
0.13
0.21
0.23
0.13
0.35
0.02
0.24
0.38
0.12
0.2
0.32
0.38
0.49
0.1
0.1
0.24
0.17
0.14
5
0.1
HgT
0.82
0.94
0.73
0.94
0.89
1.66
1.52
1.14
1.31
2.1
1.18
1.7
0.92
0.56
1.63
0.57
2.26
0.47
0.84
0.66
0.77
0.44
1.26
0.58
0.6
0.52
0.61
0.91
1.05
2.28
0.58
0.68
1.24
0.51
1.66
0.1
1.88
1.45
1.07
0.9
1.01
1.62
1.06
0.61
0.42
1.38
0.78
0.66
5
1
MeHg
0.42
0.18
0.37
0.45
0.29
0.31
0.46
0.38
0.34
1.66
0.45
0.57
0.23
0.22
0.68
0.14
0.51
0.19
0.17
0.23
0.15
0.08
0.44
0.24
0.19
0.16
0.36
0.26
0.4
0.77
0.25
0.25
0.26
0.15
0.54
0.02
0.39
0.42
0.17
0.31
0.69
0.69
0.87
0.12
0.11
0.37
0.27
0.19
5
1
HgT
2.35
2.67
2.72
3.96
4.02
3.84
4.16
5.75
5.57
5.78
3.48
7.43
3.32
1.51
6.22
1.53
4.43
1.44
1.92
1.61
2.35
0.89
5.81
2.73
3.19
1.09
1.38
2.56
4.8
4.66
1.48
3.7
3.93
1.08
6.17
0.39
3.7
6.73
2.54
3.16
4.77
6.91
1.98
1.31
1
5.33
1.68
1.6
5
2
MeHg
0.67
0.24
0.54
0.65
0.34
0.42
0.64
0.44
0.38
2.73
0.67
0.74
0.28
0.33
0.93
0.16
0.74
0.29
0.19
0.29
0.16
0.08
0.55
0.32
0.2
0.23
0.57
0.35
0.47
1.12
0.37
0.3
0.29
0.17
0.75
0.03
0.56
0.48
0.23
0.44
1.1
1.03
1.3
0.15
0.13
0.51
0.38
0.24
5
2
HgT
4.05
4.6
4.92
7.32
7.5
6.25
7.1
10.87
10.31
9.87
6.04
13.79
5.98
2.57
11.32
2.59
6.84
2.52
3.13
2.67
4.11
1.4
10.88
5.12
6.08
1.73
2.24
4.4
8.96
7.3
2.48
7.07
6.91
1.71
11.17
0.71
5.73
12.59
4.18
5.68
8.95
12.79
3.01
2.08
1.65
9.73
2.69
2.64
-------
Evaluating R-MCMfor 91 VT/NH Lakes
R1Up
R2Up
Lake
Number
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
2
2
MeHg
0.28
0.1
0.64
1.2
0.26
3.64
0.16
0.26
0.06
0.17
0.18
0.16
0.46
0.17
0.2
0.17
0.25
0.23
0.48
1.43
0.09
0.28
0.1
0.14
0.77
0.06
0.12
0.82
0.25
1.89
0.18
1.65
0.04
0.45
0.82
0.43
0.29
0.99
0.3
0.05
1.12
0.42
0.68
2
2
HgT
8.93
4.71
14.65
7.12
1.82
58.46
4.98
8.8
1.17
4.86
5.84
6.71
16.67
3.02
6.99
3.86
5.01
1.86
5.65
3.45
5.58
10.26
2.12
5.08
11.2
0.95
2.05
12.24
1.3
7.97
8.11
9.01
2.36
7.9
4.45
7.15
7.39
4.89
3.3
0.64
6.8
6.59
6.75
5
0.1
MeHg
0.24
0.13
0.41
0.4
0.13
2.92
0.14
0.26
0.07
0.17
0.17
0.18
0.53
0.21
0.25
0.15
0.18
0.14
0.24
0.47
0.2
0.28
0.08
0.17
0.35
0.08
0.09
0.37
0.14
0.63
0.29
0.51
0.07
0.24
0.2
0.23
0.27
0.3
0.3
0.09
0.35
0.24
0.27
5
0.1
HgT
1.44
0.74
1.64
1.55
0.61
7.49
0.69
0.9
0.46
0.77
0.91
1.24
1.96
0.6
1.28
0.77
0.93
0.66
0.94
1.03
0.81
1.18
0.72
0.87
1.91
0.37
0.47
1.95
0.58
2.18
1.01
2.04
0.29
1.36
0.67
1.2
1.03
0.98
0.88
0.25
1.23
1.32
1.18
5
1
MeHg
0.32
0.15
0.62
0.83
0.21
4.02
0.18
0.33
0.08
0.21
0.22
0.21
0.63
0.24
0.28
0.19
0.26
0.2
0.41
0.95
0.2
0.35
0.1
0.2
0.61
0.09
0.12
0.66
0.2
1.27
0.32
1.1
0.08
0.39
0.52
0.38
0.35
0.66
0.37
0.1
0.77
0.37
0.51
5
1
HgT
5.05
2.66
7.9
4.24
1.2
32.45
2.76
4.71
0.81
2.75
3.28
3.88
9.07
1.8
4.05
2.27
2.9
1.25
3.22
2.2
3.12
5.56
1.39
2.91
6.38
0.66
1.24
6.9
0.93
4.96
4.45
5.39
1.29
4.51
2.49
4.07
4.11
2.87
2.1
0.46
3.92
3.87
3.87
5
2
MeHg
0.4
0.17
0.86
1.3
0.29
5.24
0.23
0.41
0.09
0.26
0.27
0.25
0.75
0.28
0.33
0.24
0.34
0.27
0.59
1.48
0.2
0.43
0.12
0.23
0.9
0.1
0.16
0.97
0.26
1.98
0.35
1.75
0.09
0.55
0.88
0.54
0.43
1.07
0.44
0.1
1.23
0.52
0.78
5
2
HgT
9.05
4.78
14.87
7.22
1.86
60.19
5.05
8.95
1.2
4.95
5.93
6.81
16.97
3.14
7.13
3.93
5.1
1.9
5.77
3.5
5.7
10.42
2.15
5.17
11.35
0.99
2.09
12.39
1.32
8.06
8.28
9.11
2.4
8
4.51
7.25
7.54
4.97
3.44
0.7
6.91
6.7
6.86
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-23. Hypolimnion MeHg and HgT Concentrations for Range of RlUp and R2Up
Values.
R1Up
R2Up
Lake
Number
1
3
5
6
11
14
18
19
20
22
26
28
31
32
33
34
36
44
45
46
47
48
53
57
59
62
63
64
65
66
69
71
72
74
75
77
79
81
82
85
88
90
91
0.1
0.1
MeHg
0.754
0.363
0.22
0.861
0.931
0.31
0.301
0.182
0.333
0.024
0.222
0.484
0.368
0.232
0.28
0.229
0.018
0.114
0.148
0.481
0.87
0.319
0.716
0.155
0.205
0.183
0.586
0.25
0.374
0.647
0.007
0.293
0.239
0.17
0.275
0.812
0.195
0.094
0.582
0.319
0.105
0.592
1.065
0.1
0.1
HgT
3.696
2.566
1.759
3.161
3.364
3.284
2.369
2.243
1.872
0.523
2.591
1.719
3.896
2.095
1.866
1.588
11.718
1.193
1.153
1.822
2.639
1.655
2.719
0.957
1.248
1.468
4.463
1.533
1.672
2.484
0.396
1.295
3.901
1.091
1.314
2.893
2.657
3.492
2.085
2.041
0.459
2.121
3.029
0.1
1
MeHg
2.284
1.516
0.983
1.888
2.829
0.883
0.992
0.305
0.743
0.038
0.472
1.355
0.992
1.458
0.792
0.406
0.071
0.21
0.268
1.855
1.874
0.727
1.416
0.222
0.678
0.48
1.828
0.661
1.143
1.2
0.007
0.527
0.704
0.24
0.714
1.286
0.574
0.447
1.942
1.219
0.125
1.701
3.655
0.1
1
HgT
11.168
10.582
7.766
7.257
10.601
9.369
7.538
5.337
5
2.224
5.675
4.885
10.696
8.67
5.311
3.532
12.127
2.639
2.909
7.323
5.882
4.168
5.525
1.69
4.054
5.136
12.112
4.43
5.319
4.803
2.904
2.45
13.068
2.044
3.18
4.628
1 1 .775
5.653
7.083
8.693
0.859
6.365
10.483
0.1
2
MeHg
3.984
2.797
1.831
3.03
4.938
1.52
1.76
0.441
1.199
0.053
0.75
2.322
1.685
2.821
1.361
0.602
0.13
0.361
0.401
3.382
2.99
1.179
2.194
0.296
1.205
0.811
3.209
1.118
1.996
1.814
0.008
0.788
1.221
0.318
1.201
1.812
0.996
0.839
3.452
2.218
0.147
2.933
6.532
0.1
2
HgT
19.47
19.488
14.44
1 1 .809
18.642
16.13
13.281
8.776
8.476
4.014
9.101
8.402
18.251
15.975
9.137
5.691
12.581
4.247
4.86
13.434
9.486
6.96
8.643
2.504
7.172
9.211
20.611
7.649
9.37
7.379
5.543
3.734
23.254
3.103
5.254
6.555
21.907
8.054
12.637
16.083
1.304
11.08
18.766
1
0.1
MeHg
0.836
0.492
0.427
0.991
1.139
0.357
0.36
0.332
0.502
0.058
0.249
0.607
0.414
0.444
0.479
0.333
0.039
0.153
0.258
0.682
0.984
0.432
0.787
0.206
0.291
0.386
0.843
0.35
0.51
0.725
0.046
0.336
0.414
0.242
0.344
0.842
0.485
0.168
0.74
0.545
0.206
0.739
1.275
1
0.1
HgT
3.983
3.034
2.02
3.325
3.709
3.554
2.624
2.445
2.189
0.646
2.746
1.876
4.154
2.448
2.131
1.777
1 1 .744
1.244
1.306
2.089
2.811
1.849
2.859
1.025
1.354
1.831
4.95
1.657
1.842
2.622
0.567
1.353
4.227
1.191
1.431
2.969
3.024
3.625
2.285
2.329
0.596
2.307
3.382
-------
Evaluating R-MCMfor 91 VT/NH Lakes
FMUp
R2Up
Lake
Number
1
3
5
6
11
14
18
19
20
22
26
28
31
32
33
34
36
44
45
46
47
48
53
57
59
62
63
64
65
66
69
71
72
74
75
77
79
81
82
85
88
90
91
1
1
MeHg
2.366
1.645
1.19
2.018
3.037
0.93
1.051
0.455
0.912
0.071
0.499
1.477
1.037
1.67
0.991
0.51
0.092
0.248
0.378
2.056
1.988
0.839
1.487
0.273
0.764
0.683
2.086
0.76
1.278
1.278
0.046
0.57
0.879
0.313
0.782
1.315
0.864
0.521
2.099
1.445
0.226
1.847
3.864
1
1
HgT
11.455
1 1 .049
8.026
7.422
10.946
9.639
7.793
5.539
5.317
2.257
5.83
5.042
10.954
9.023
5.576
3.72
12.153
2.691
3.062
7.59
6.054
4.362
5.665
1.758
4.16
5.499
12.599
4.554
5.488
4.941
2.943
2.509
13.394
2.144
3.297
4.705
12.142
5.786
7.283
8.98
0.997
6.55
10.837
1
2
MeHg
4.065
2.927
2.038
3.16
5.146
1.567
1.819
0.592
1.368
0.086
0.777
2.445
1.73
3.033
1.56
0.706
0.151
0.354
0.511
3.583
3.104
1.292
2.265
0.347
1.291
1.014
3.467
1.217
2.131
1.892
0.047
0.83
1.395
0.391
1.269
1.842
1.286
0.912
3.609
2.444
0.248
3.079
6.741
1
2
HgT
19.757
19.955
14.7
1 1 .974
18.987
16.401
13.536
8.978
8.793
4.047
9.256
8.559
18.509
16.329
9.403
5.879
12.607
4.298
5.014
13.702
9.658
7.154
8.782
2.572
7.278
9.574
21.098
7.772
9.54
7.517
5.582
3.793
23.579
3.203
5.371
6.631
22.273
8.187
12.837
16.37
1.441
11.265
19.119
2
0.1
MeHg
0.926
0.636
0.657
1.136
1.37
0.409
0.426
0.5
0.69
0.092
0.28
0.742
0.464
0.68
0.7
0.448
0.062
0.196
0.381
0.905
1.11
0.556
0.866
0.263
0.386
0.612
1.13
0.46
0.66
0.811
0.087
0.383
0.608
0.323
0.419
0.875
0.807
0.25
0.915
0.796
0.318
0.901
1.508
2
0.1
HgT
4.301
3.553
2.309
3.508
4.093
3.855
2.907
2.669
2.542
0.681
2.918
2.051
4.441
2.841
2.426
1.986
1 1 .773
1.301
1.477
2.387
3.001
2.065
3.013
1.101
1.471
2.235
5.492
1.794
2.031
2.776
0.608
1.419
4.589
1.303
1.562
3.052
3.432
3.772
2.506
2.648
0.749
2.513
3.775
2
1
MeHg
2.456
1.789
1.42
2.163
3.268
0.982
1.117
0.622
1.1
0.106
0.53
1.613
1.088
1.906
1.212
0.625
0.115
0.291
0.5
2.279
2.115
0.964
1.566
0.33
0.86
0.909
2.372
0.871
1.428
1.364
0.087
0.618
1.073
0.393
0.858
1.348
1.187
0.603
2.275
1.695
0.338
2.01
4.098
2
1
HgT
1 1 .773
1 1 .569
8.316
7.605
11.33
9.94
8.076
5.764
5.67
2.292
6.002
5.217
11.241
9.416
5.87
3.929
12.182
2.748
3.233
7.887
6.245
4.578
5.819
1.834
4.277
5.903
13.141
4.691
5.677
5.095
2.984
2.574
13.756
2.256
3.428
4.787
12.55
5.933
7.505
9.299
1.15
6.756
11.229
-------
Evaluating R-MCMfor 91 VT/NH Lakes
FMUp
R2Up
Lake
Number
1
3
5
6
11
14
18
19
20
22
26
28
31
32
33
34
36
44
45
46
47
48
53
57
59
62
63
64
65
66
69
71
72
74
75
77
79
81
82
85
88
90
91
2
2
MeHg
4.156
3.07
2.267
3.304
5.377
1.618
1.885
0.759
1.556
0.12
0.808
2.58
1.781
3.269
1.781
0.821
0.175
0.397
0.633
3.806
3.23
1.416
2.344
0.404
1.386
1.24
3.753
1.327
2.281
1.978
0.088
0.878
1.589
0.471
1.345
1.875
1.608
0.995
3.785
2.695
0.361
3.242
6.974
2
2
HgT
20.075
20.474
14.99
12.156
19.371
16.701
13.819
9.202
9.146
4.082
9.428
8.734
18.796
16.721
9.697
6.088
12.636
4.355
5.185
13.999
9.849
7.369
8.937
2.648
7.395
9.978
21.64
7.91
9.729
7.671
5.623
3.858
23.942
3.315
5.501
6.714
22.681
8.334
13.058
16.689
1.594
11.471
19.512
5
0.1
MeHg
1.197
1.067
1.346
1.569
2.062
0.564
0.623
1.001
1.254
0.196
0.371
1.15
0.616
1.387
1.362
0.793
0.132
0.325
0.747
1.574
1.49
0.931
1.104
0.433
0.672
1.29
1.989
0.791
1.11
1.07
0.21
0.526
1.189
0.564
0.647
0.974
1.774
0.497
1.441
1.549
0.656
1.39
2.207
5
0.1
HgT
5.257
5.111
3.178
4.056
5.244
4.757
3.756
3.342
3.599
0.785
3.435
2.575
5.301
4.02
3.31
2.613
11.86
1.473
1.99
3.278
3.574
2.712
3.478
1.328
1.823
3.447
7.117
2.206
2.597
3.236
0.732
1.614
5.675
1.637
1.952
3.303
4.655
4.213
3.171
3.605
1.208
3.13
4.952
5
1
MeHg
2.727
2.22
2.108
2.596
3.96
1.137
1.314
1.124
1.664
0.209
0.622
2.02
1.239
2.613
1.874
0.97
0.185
0.42
0.867
2.949
2.495
1.338
1.804
0.5
1.146
1.587
3.232
1.202
1.878
1.623
0.211
0.76
1.654
0.634
1.085
1.447
2.154
0.849
2.8
2.448
0.675
2.499
4.797
5
1
HgT
12.728
13.127
9.184
8.153
12.481
10.842
8.925
6.436
6.727
2.397
6.519
5.741
12.101
10.595
6.754
4.556
12.269
2.919
3.747
8.779
6.818
5.224
6.284
2.061
4.629
7.115
14.766
5.103
6.243
5.555
3.107
2.77
14.842
2.59
3.819
5.038
13.773
6.374
8.169
10.256
1.608
7.374
0.442
5
2
MeHg
4.427
3.501
2.956
3.737
6.069
1.773
2.082
1.26
2.12
0.223
0.899
2.988
1.932
3.976
2.443
1.166
0.245
0.526
1
4.475
3.61
1.791
2.582
0.574
1.672
1.918
4.612
1.658
2.732
2.237
0.211
1.02
2.171
0.712
0.007
1.973
2.575
1.241
4.311
3.488
0.698
3.73
7.673
5
2
HgT
21.03
22.033
15.858
12.705
20.522
17.603
14.668
9.875
10.203
4.187
9.945
9.258
19.656
17.9
10.581
6.716
12.723
4.526
5.698
14.891
10.422
8.016
9.401
2.876
7.747
11.19
23.265
8.322
10.295
8.132
5.746
4.054
25.028
3.649
5.892
6.965
23.904
8.775
13.723
17.646
2.053
12.089
20.689
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-24. Sediment HgT Concentrations for Range of RlUp and R2Up Values.
R1Up
R2Up
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
0.1
0.1
0.02
0.09
0.01
0.04
0.04
0.08
0.06
0.06
0.06
0.13
0.06
0.06
0.04
0.01
0.06
0.03
0.21
0.01
0.03
0.01
0.04
0.01
0.03
0.01
0.04
0.01
0.01
0.04
0.02
0.1
0.01
0.03
0.1
0.01
0.07
0
0.2
0.06
0.11
0.03
0.02
0.1
0.02
0.02
0.02
0.11
0.04
0.01
0.08
0.1
1
0.06
0.29
0.05
0.24
0.21
0.21
0.17
0.39
0.29
0.39
0.21
0.34
0.16
0.04
0.28
0.1
0.42
0.05
0.08
0.03
0.13
0.02
0.18
0.09
0.33
0.02
0.03
0.14
0.15
0.22
0.04
0.22
0.36
0.02
0.32
0.02
0.4
0.34
0.28
0.12
0.14
0.52
0.04
0.05
0.04
0.51
0.1
0.04
0.3
0.1
2
0.11
0.51
0.1
0.45
0.4
0.34
0.3
0.77
0.56
0.68
0.38
0.65
0.31
0.07
0.53
0.18
0.65
0.09
0.14
0.06
0.23
0.03
0.34
0.18
0.64
0.04
0.05
0.25
0.29
0.35
0.06
0.43
0.66
0.04
0.59
0.04
0.62
0.65
0.47
0.23
0.27
0.99
0.06
0.07
0.07
0.95
0.16
0.07
0.55
1
0.1
0.02
0.09
0.01
0.04
0.04
0.08
0.06
0.06
0.06
0.13
0.06
0.06
0.04
0.01
0.06
0.03
0.21
0.01
0.03
0.01
0.04
0.01
0.03
0.01
0.04
0.01
0.01
0.05
0.02
0.1
0.01
0.03
0.1
0.01
0.07
0
0.2
0.06
0.11
0.03
0.02
0.1
0.02
0.02
0.02
0.11
0.04
0.01
0.08
1
1
0.06
0.29
0.05
0.24
0.21
0.21
0.17
0.39
0.29
0.39
0.21
0.34
0.16
0.04
0.28
0.1
0.42
0.05
0.08
0.03
0.13
0.02
0.18
0.09
0.33
0.02
0.03
0.14
0.15
0.22
0.04
0.22
0.36
0.03
0.32
0.02
0.4
0.34
0.28
0.12
0.14
0.52
0.04
0.05
0.04
0.51
0.1
0.04
0.3
1
2
0.11
0.51
0.1
0.45
0.4
0.34
0.3
0.77
0.56
0.68
0.38
0.65
0.31
0.07
0.53
0.18
0.65
0.09
0.14
0.06
0.23
0.03
0.34
0.18
0.64
0.04
0.05
0.25
0.29
0.35
0.06
0.43
0.66
0.04
0.6
0.04
0.62
0.66
0.47
0.23
0.27
0.99
0.06
0.07
0.07
0.95
0.16
0.07
0.55
2
0.1
0.02
0.09
0.01
0.04
0.04
0.08
0.06
0.06
0.06
0.13
0.06
0.06
0.04
0.01
0.06
0.03
0.21
0.01
0.03
0.01
0.04
0.01
0.03
0.02
0.04
0.01
0.01
0.05
0.02
0.1
0.01
0.03
0.1
0.01
0.08
0
0.2
0.06
0.11
0.03
0.02
0.1
0.02
0.02
0.02
0.12
0.04
0.01
0.08
2
1
0.06
0.29
0.05
0.24
0.21
0.21
0.17
0.39
0.29
0.39
0.21
0.34
0.17
0.04
0.28
0.1
0.42
0.05
0.08
0.03
0.13
0.02
0.18
0.09
0.33
0.02
0.03
0.14
0.15
0.22
0.04
0.22
0.36
0.03
0.32
0.02
0.4
0.34
0.28
0.12
0.14
0.52
0.04
0.05
0.04
0.51
0.1
0.04
0.3
2
2
0.11
0.51
0.1
0.45
0.4
0.34
0.3
0.77
0.56
0.68
0.38
0.65
0.31
0.07
0.53
0.18
0.65
0.09
0.14
0.06
0.23
0.03
0.34
0.18
0.64
0.04
0.05
0.25
0.29
0.35
0.06
0.43
0.66
0.04
0.6
0.04
0.62
0.66
0.47
0.23
0.27
0.99
0.06
0.07
0.07
0.95
0.16
0.07
0.55
5
0.1
0.02
0.09
0.01
0.05
0.04
0.09
0.06
0.06
0.06
0.13
0.06
0.07
0.04
0.01
0.06
0.03
0.21
0.01
0.03
0.01
0.04
0.01
0.03
0.02
0.05
0.01
0.01
0.05
0.02
0.11
0.01
0.03
0.1
0.01
0.08
0
0.2
0.06
0.11
0.03
0.02
0.1
0.02
0.02
0.02
0.12
0.04
0.01
0.08
5
1
0.06
0.29
0.05
0.24
0.21
0.21
0.17
0.4
0.3
0.39
0.21
0.34
0.17
0.04
0.28
0.1
0.42
0.05
0.09
0.03
0.13
0.02
0.18
0.09
0.33
0.02
0.03
0.14
0.15
0.22
0.04
0.22
0.36
0.03
0.32
0.02
0.4
0.34
0.28
0.12
0.14
0.52
0.04
0.05
0.04
0.51
0.1
0.04
0.3
5
2
0.11
0.51
0.1
0.45
0.4
0.34
0.3
0.77
0.56
0.68
0.38
0.65
0.31
0.07
0.53
0.18
0.65
0.09
0.14
0.06
0.24
0.03
0.34
0.18
0.64
0.04
0.05
0.25
0.29
0.35
0.06
0.43
0.66
0.04
0.6
0.04
0.62
0.66
0.47
0.23
0.27
0.99
0.06
0.07
0.08
0.95
0.16
0.07
0.55
-------
Evaluating R-MCMfor 91 VT/NH Lakes
R1Up
R2Up
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
0.1
0.1
0.02
0.06
0.07
0.04
0.16
0.02
0.03
0.01
0.02
0.02
0.05
0.05
0.01
0.1
0.03
0.04
0.03
0.04
0.02
0.01
0.04
0.06
0.03
0.15
0.02
0.03
0.07
0.01
0.12
0.02
0.11
0.01
0.06
0.01
0.05
0.04
0.26
0.03
0.003
0.06
0.1
1
0.1
0.39
0.2
0.08
1.06
0.11
0.24
0.03
0.09
0.1
0.16
0.27
0.04
0.38
0.1
0.16
0.06
0.19
0.06
0.07
0.22
0.12
0.14
0.57
0.03
0.09
0.27
0.02
0.28
0.15
0.31
0.08
0.21
0.06
0.18
0.22
0.85
0.08
0.01
0.23
0.1
2
0.18
0.74
0.35
0.13
2.06
0.2
0.47
0.04
0.17
0.19
0.29
0.52
0.07
0.69
0.19
0.29
0.09
0.34
0.09
0.14
0.42
0.19
0.25
1.04
0.05
0.15
0.49
0.03
0.46
0.28
0.54
0.16
0.37
0.11
0.33
0.41
1.51
0.14
0.01
0.42
1
0.1
0.02
0.06
0.07
0.04
0.17
0.02
0.03
0.01
0.02
0.02
0.05
0.05
0.01
0.1
0.03
0.04
0.03
0.04
0.02
0.01
0.04
0.06
0.03
0.15
0.02
0.03
0.07
0.01
0.12
0.03
0.11
0.01
0.06
0.01
0.05
0.04
0.26
0.03
0
0.06
1
1
0.1
0.39
0.2
0.08
1.07
0.11
0.24
0.03
0.09
0.1
0.16
0.27
0.04
0.38
0.1
0.16
0.06
0.19
0.06
0.07
0.22
0.12
0.14
0.57
0.03
0.09
0.27
0.02
0.28
0.15
0.32
0.08
0.21
0.06
0.18
0.22
0.85
0.08
0.01
0.23
1
2
0.18
0.74
0.35
0.13
2.07
0.2
0.47
0.04
0.17
0.19
0.29
0.52
0.07
0.69
0.19
0.29
0.09
0.34
0.09
0.14
0.42
0.19
0.25
1.04
0.05
0.15
0.49
0.03
0.46
0.28
0.54
0.16
0.37
0.11
0.33
0.41
1.51
0.14
0.01
0.42
2
0.1
0.02
0.06
0.07
0.04
0.18
0.02
0.03
0.01
0.02
0.02
0.05
0.05
0.01
0.1
0.03
0.04
0.03
0.05
0.02
0.01
0.04
0.06
0.03
0.15
0.02
0.03
0.07
0.01
0.12
0.03
0.11
0.01
0.06
0.01
0.05
0.04
0.26
0.03
0
0.06
2
1
0.1
0.39
0.2
0.08
1.08
0.11
0.24
0.03
0.09
0.1
0.16
0.27
0.04
0.38
0.1
0.16
0.06
0.19
0.06
0.07
0.22
0.12
0.14
0.58
0.03
0.09
0.27
0.02
0.28
0.15
0.32
0.08
0.21
0.06
0.18
0.22
0.85
0.08
0.01
0.23
2
2
0.18
0.74
0.35
0.13
2.08
0.2
0.47
0.04
0.17
0.19
0.29
0.52
0.07
0.69
0.19
0.29
0.09
0.34
0.09
0.14
0.42
0.19
0.25
1.04
0.05
0.15
0.49
0.03
0.46
0.28
0.54
0.16
0.38
0.11
0.33
0.41
1.51
0.14
0.01
0.42
5
0.1
0.02
0.07
0.07
0.04
0.19
0.02
0.04
0.01
0.02
0.02
0.05
0.05
0.01
0.11
0.03
0.05
0.03
0.05
0.02
0.02
0.04
0.06
0.04
0.16
0.02
0.03
0.07
0.01
0.12
0.03
0.11
0.01
0.06
0.01
0.05
0.04
0.26
0.03
0
0.06
5
1
0.1
0.39
0.2
0.08
1.09
0.11
0.24
0.03
0.09
0.1
0.16
0.27
0.04
0.38
0.1
0.16
0.06
0.19
0.06
0.07
0.22
0.12
0.14
0.58
0.03
0.09
0.27
0.02
0.28
0.15
0.32
0.08
0.21
0.06
0.18
0.22
0.86
0.08
0.01
0.23
5
2
0.18
0.74
0.35
0.13
2.09
0.2
0.47
0.04
0.17
0.19
0.29
0.52
0.07
0.7
0.19
0.29
0.09
0.34
0.09
0.14
0.42
0.19
0.25
1.04
0.05
0.15
0.49
0.03
0.46
0.28
0.54
0.16
0.38
0.11
0.33
0.42
1.51
0.14
0.01
0.42
-------
Evaluating R-MCMfor 91 VT/NH Lakes
Table A-25. Fish Hg Concentrations for range of RlUp and R2Up values.
R1Up
R2Up
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
0.1
0.1
0.11
0.16
0.05
0.16
0.07
0.48
0.26
0.16
0.13
1.46
0.23
0.25
0.11
0.07
0.24
0.08
0.99
0.05
0.09
0.1
0.09
0.03
0.09
0.07
0.07
0.05
0.23
0.17
0.09
0.79
0.07
0.06
0.13
0.07
0.27
0.01
0.74
0.16
0.21
0.15
0.14
0.41
0.4
0.05
0.07
0.24
0.19
0.08
0.18
0.1
1
0.34
0.44
0.21
0.77
0.31
1.07
0.7
0.95
0.56
4.22
0.71
1.04
0.38
0.19
0.85
0.17
1.92
0.17
0.16
0.22
0.25
0.06
0.42
0.41
0.4
0.1
0.53
0.48
0.46
1.61
0.2
0.4
0.37
0.12
1.01
0.03
1.45
0.69
0.46
0.56
0.76
1.9
0.77
0.1
0.14
0.93
0.41
0.19
0.64
0.1
2
0.6
0.76
0.39
1.44
0.57
1.72
1.19
1.82
1.03
7.29
1.24
1.93
0.68
0.33
1.54
0.28
2.96
0.3
0.24
0.36
0.43
0.09
0.79
0.78
0.77
0.16
0.87
0.83
0.88
2.53
0.34
0.77
0.63
0.18
1.84
0.05
2.23
1.28
0.74
1.02
1.46
3.56
1.18
0.15
0.21
1.71
0.66
0.32
1.15
1
0.1
0.12
0.21
0.06
0.22
0.14
0.56
0.32
0.26
0.24
1.54
0.28
0.37
0.16
0.07
0.34
0.12
1.03
0.06
0.15
0.14
0.12
0.05
0.16
0.09
0.17
0.05
0.24
0.22
0.13
0.85
0.08
0.11
0.21
0.09
0.37
0.01
0.78
0.28
0.24
0.2
0.17
0.51
0.41
0.07
0.12
0.34
0.21
0.11
0.26
1
1
0.35
0.49
0.23
0.83
0.38
1.14
0.76
1.05
0.66
4.3
0.76
1.16
0.43
0.2
0.96
0.22
1.97
0.18
0.22
0.26
0.28
0.08
0.5
0.43
0.51
0.11
0.55
0.53
0.5
1.67
0.21
0.45
0.45
0.15
1.11
0.03
1.49
0.81
0.49
0.61
0.79
2.01
0.78
0.11
0.19
1.03
0.44
0.22
0.72
1
2
0.61
0.8
0.4
1.5
0.64
1.79
1.25
1.92
1.13
7.37
1.29
2.05
0.73
0.34
1.64
0.32
3
0.31
0.3
0.4
0.47
0.1
0.87
0.8
0.88
0.17
0.88
0.87
0.92
2.59
0.35
0.82
0.72
0.21
1.94
0.05
2.27
1.4
0.77
1.07
1.49
3.67
1.19
0.17
0.26
1.8
0.68
0.34
1.23
2
0.1
0.13
0.25
0.08
0.29
0.22
0.64
0.38
0.38
0.36
1.63
0.33
0.5
0.22
0.08
0.46
0.18
1.08
0.07
0.22
0.19
0.16
0.07
0.24
0.12
0.29
0.06
0.26
0.27
0.18
0.91
0.09
0.17
0.31
0.12
0.48
0.02
0.83
0.42
0.28
0.26
0.2
0.64
0.42
0.08
0.18
0.44
0.24
0.14
0.35
2
1
0.36
0.54
0.24
0.9
0.45
1.22
0.82
1.16
0.78
4.39
0.81
1.3
0.49
0.21
1.08
0.27
2.02
0.18
0.29
0.31
0.32
0.1
0.58
0.45
0.62
0.11
0.56
0.58
0.55
1.73
0.22
0.5
0.55
0.18
1.22
0.04
1.53
0.95
0.53
0.67
0.83
2.13
0.79
0.13
0.24
1.14
0.46
0.25
0.81
2
2
0.62
0.85
0.42
1.57
0.72
1.87
1.31
2.03
1.25
7.46
1.34
2.18
0.79
0.35
1.76
0.38
3.05
0.32
0.37
0.45
0.51
0.12
0.95
0.83
0.99
0.17
0.9
0.93
0.97
2.65
0.36
0.88
0.82
0.24
2.05
0.06
2.31
1.54
0.81
1.13
1.52
3.79
1.2
0.18
0.31
1.91
0.71
0.37
1.32
5
0.1
0.17
0.4
0.13
0.5
0.45
0.87
0.57
0.71
0.72
1.89
0.5
0.89
0.39
0.11
0.81
0.33
1.23
0.09
0.44
0.34
0.28
0.13
0.49
0.19
0.63
0.07
0.3
0.42
0.32
1.09
0.11
0.33
0.6
0.21
0.81
0.05
0.96
0.83
0.41
0.43
0.3
1
0.45
0.14
0.34
0.76
0.32
0.23
0.62
5
1
0.4
0.68
0.29
1.11
0.69
1.45
1.01
1.5
1.14
4.65
0.97
1.69
0.66
0.23
1.43
0.43
2.16
0.21
0.51
0.46
0.44
0.15
0.82
0.52
0.96
0.13
0.6
0.73
0.7
1.91
0.24
0.67
0.84
0.27
1.55
0.07
1.66
1.36
0.66
0.84
0.93
2.49
0.82
0.18
0.4
1.45
0.54
0.34
1.08
5
2
0.65
0.99
0.47
1.78
0.95
2.1
1.5
2.37
1.61
7.72
1.51
2.57
0.97
0.37
2.11
0.53
3.2
0.34
0.59
0.6
0.62
0.18
1.19
0.9
1.33
0.19
0.94
1.08
1.11
2.83
0.38
1.04
1.11
0.33
2.38
0.09
2.44
1.95
0.94
1.3
1.62
4.15
1.24
0.23
0.47
2.22
0.79
0.46
1.59
-------
Evaluating R-MCMfor 91 VT/NH Lakes
R1Up
R2Up
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
0.1
0.1
0.05
0.17
0.55
0.21
1.07
0.06
0.09
0.07
0.08
0.07
0.11
0.22
0.05
0.2
0.08
0.12
0.17
0.25
0.41
0.01
0.12
0.13
0.08
0.53
0.08
0.08
0.46
0.11
1.21
0.07
0.85
0.03
0.23
0.08
0.35
0.11
2.12
0.46
0.05
0.45
0.1
1
0.16
0.89
1.57
0.42
5.4
0.25
0.51
0.1
0.27
0.21
0.32
0.96
0.15
0.65
0.21
0.37
0.32
0.92
0.93
0.01
0.56
0.23
0.27
1.76
0.12
0.22
1.64
0.17
2.79
0.24
2.32
0.14
0.76
0.31
1.21
0.4
6.64
0.88
0.06
1.55
0.1
2
0.27
1.68
2.71
0.65
10.21
0.45
0.98
0.14
0.48
0.38
0.56
1.78
0.26
1.14
0.36
0.64
0.48
1.66
1.5
0.02
1.05
0.35
0.47
3.13
0.16
0.37
2.95
0.24
4.55
0.43
3.95
0.25
1.35
0.56
2.17
0.72
11.67
1.35
0.07
2.77
1
0.1
0.08
0.28
0.6
0.23
2.34
0.09
0.18
0.09
0.12
0.09
0.16
0.38
0.1
0.28
0.11
0.17
0.19
0.32
0.43
0.04
0.2
0.15
0.13
0.65
0.11
0.1
0.56
0.11
1.26
0.17
0.91
0.05
0.29
0.09
0.45
0.18
2.33
0.69
0.09
0.51
1
1
0.18
1
1.62
0.44
6.67
0.28
0.6
0.12
0.31
0.24
0.37
1.12
0.2
0.72
0.24
0.42
0.34
0.99
0.94
0.05
0.64
0.25
0.31
1.88
0.15
0.24
1.75
0.17
2.84
0.34
2.38
0.15
0.82
0.32
1.31
0.47
6.85
1.1
0.1
1.61
1
2
0.3
1.79
2.76
0.67
11.48
0.49
1.06
0.16
0.52
0.41
0.6
1.95
0.31
1.22
0.39
0.69
0.5
1.74
1.51
0.05
1.13
0.37
0.51
3.25
0.19
0.39
3.06
0.24
4.6
0.53
4.01
0.27
1.41
0.57
2.27
0.8
11.88
1.57
0.11
2.83
2
0.1
0.11
0.41
0.66
0.25
3.76
0.13
0.27
0.11
0.17
0.12
0.2
0.57
0.16
0.36
0.15
0.22
0.21
0.41
0.44
0.08
0.29
0.17
0.18
0.78
0.15
0.12
0.69
0.11
1.31
0.28
0.97
0.07
0.36
0.1
0.56
0.27
2.56
0.93
0.14
0.58
2
1
0.21
1.12
1.68
0.46
8.09
0.31
0.69
0.15
0.36
0.27
0.41
1.31
0.26
0.81
0.28
0.47
0.36
1.08
0.95
0.08
0.73
0.27
0.36
2.01
0.19
0.26
1.87
0.18
2.9
0.45
2.44
0.18
0.89
0.33
1.42
0.56
7.08
1.35
0.15
1.68
2
2
0.33
1.92
2.82
0.69
12.9
0.52
1.16
0.18
0.57
0.44
0.65
2.13
0.36
1.3
0.43
0.74
0.52
1.83
1.52
0.09
1.22
0.39
0.57
3.38
0.23
0.41
3.18
0.25
4.66
0.63
4.07
0.29
1.48
0.59
2.38
0.88
12.11
1.82
0.16
2.89
5
0.1
0.19
0.77
0.85
0.32
8.01
0.23
0.55
0.19
0.31
0.22
0.35
1.12
0.33
0.62
0.26
0.38
0.27
0.66
0.48
0.18
0.56
0.23
0.34
1.18
0.26
0.18
1.05
0.12
1.48
0.61
1.16
0.13
0.57
0.14
0.89
0.53
3.26
1.68
0.29
0.78
5
1
0.29
1.49
1.87
0.53
12.34
0.42
0.97
0.22
0.5
0.37
0.56
1.86
0.43
1.06
0.39
0.63
0.42
1.33
0.99
0.19
1
0.33
0.52
2.42
0.3
0.32
2.23
0.19
3.06
0.78
2.63
0.24
1.1
0.37
1.75
0.82
7.78
2.1
0.3
1.88
5
2
0.41
2.29
3
0.76
17.15
0.63
1.44
0.25
0.71
0.53
0.8
2.68
0.54
1.56
0.54
0.9
0.59
2.08
1.56
0.19
1.49
0.45
0.72
3.78
0.34
0.47
3.55
0.26
4.82
0.96
4.26
0.36
1.69
0.62
2.71
1.14
12.81
2.56
0.31
3.1
-------
Evaluating R-MCMfor 91 VT/NH Lakes
FIGURES
-------
1 .U
g> 0.8
O)
CD 0.6
2
.2 0.4
1
E. 0.2
LU
Q
.
.
.
0
o
8
e S
: ง
: i
I 1
1 .U
0.8
0.6
0.4
0.2
i-i
*
.
*
'
i
ฃ
g "
1 .U
0.8
0.6
0.4
0.2
l-l
.
.
. .
:
5 I
. D
' ' 0
e |
8 '
1 .U
0.8
0.6
0.4
0.2
l-l
.
.
8
S
1
6 8
i ฐ
i
e i
0.01 0.1 1.0 10 100 0 5 10 15 20 5 6 7 8 Well-Mixed Stratified
6r> r> r>
I 5
I- 4
O)
X
c 3
O
.1 2
'o.
LU 1
Q
;
i
\ i :
1 i
i i :
' i
* I
u
5
4
3
2
1
i-i
*
1 o :
8 Q 2
1 i
e 8
U
5
4
3
2
1
l-l
'
ฐ
i
i i
i e
i 8
i ง
1 8
U
5
4
3
2
1
l-l
:
i
0
i i
s !
S S
1 I
0.01 0.1 1.0 10 100 0 5 10 15 20 5 6 7 8 Well-Mixed Stratified
2
IT
c
" 1.5
O)
X
CD
^ 1
O
1 0.5
O
Q.
1 l-l
.
t
ป i
i
8
8
,
2
1.5
1
0.5
i-l
.
: t
| ^
m m
1
> .
2
1.5
1
0.5
i-l
]
I
8 8
t
1
,
1
0.01 0.1 1.0 10 100 5 10 15 20 5678
6
U
"--- c
O)
I 4
X
S 3
o
1 2
O
>, ]
X
Q
8
0
O
.
8
! I
e
" .
:
o
5
4
3
2
1
l-l
0
; e
m
o 5
ป "
*
1
:
o
5
4
3
2
1
n
: :
. 0 . J
8
*
1
0.01 0.1 1.0 10 100 5 10 15 20 5678
Lake Area [km2] Epilimnion DOC [mg/L] Epilimnion pH Stratification
Figure 3.1. Default Modeling Outputs for All Combinations of Drainage Lakes.
Epilimnion and Hypolimnion Mercury Concentrations.
-------
5> 0.015
O)
=3
O)
I
"c
CD
.i o.oos
CD
.
m
0.015
0.01
0.005
.
'
I
I
1
1
0.015
0.01
0.005
.
0.015
0.01
0.005
.
^
0.01 0.1 1.0 10 100 5 10 15 20 56789 Stratified Well-Mixed
^ 0.3
g> 0.25
fe 0.2
X
c 0.15
ฃ
T3 0.1
CD
0.05
.
.
.
. .
I *
i i i
0.3
0.25
0.2
0.15
0.1
0.05
I . . 1
0.3
0.25
0.2
0.15
0.1
0.05
.
.
.
. .
!
\ \
11 !
0.3
0.25
0.2
0.15
0.1
0.05
.
.
.
.
I
1
!
i
0.01 0.1 1.0 10 100 5 10 15 20 56789 Stratified Well-Mixed
.o> 1.2
O)
w
U- 0.8
CD
Q. 0.6
O)
0 0.4
0.2
0
ฐ
i .
Mi
! 1 I
1.2
1
0.8
0.6
0.4
0.2
0
"
'. ' '
1 I ' '
! i i
i i !
1.2
1
0.8
0.6
0.4
0.2
0
*
i
1 1 i
1.2
1
0.8
0.6
0.4
0.2
0
t i
i i
0.01 0.1 1.0 10 100 5 10 15 20 56789 Stratified Well-Mixed
O)
I 2
.c
w
U- 1.5
o
CO
CD 1
CL
O)
X 0.5
CD
*
.
1
I
i 1 i
2
1.5
1
0.5
1 *
iii!
2
1.5
1
0.5
.
i
:
iii
2
1.5
1
0.5
.
i
i i
0.01 0.1 1.0 10 100 5 10 15 20 56789 Stratified Well-Mixed
Lake Area [km2] Epilimnion DOC [mg/L] Epilimnion pH Stratification
Figure 3.2. Default Modeling Outputs for All Combinations of Drainage Lakes.
Sediment and Fish Mercury Concentrations.
-------
Epilimnion MeHg [ng/L]
Epilimnion HgT [ng/L]
1.5
o
-* * 1
o '
CD
ol
0.5
C
4
3.5
3
0
& 2.5
g
cp 2
0.
1.5
1
0.5
C
1
0.8
^
CD
o 0.6
CD
ol
0.4
0.2
C
X
X
X
X
6
X
X
X
X
5
X
X
X
'X "O ,
' CD 4
x 4ii
xx -^
* * x^ Q) Q
... .y ฃ
x
x
' ' "* ' X * 2
QO x*^ c-
X
/
i V/ . 1
J$r.- ,'
/* i i i r\
u
) 0.5 1 1.5 (
Measured
Hypolimnion MeHg [ng/L]
X
x
x
x
x
x
25
X
X
X
X
X
20
xx "O
S
=ง15
. . . / 2
ol
X
10
X
X
. 'X
"***xr. ... 5
x
5**% * *
' n
u
) 1 2 3 4 (
Measured
Fish Cone, [ug/g]
0.6
0.5
' ' 0.4
o
^x CD
xx' O
=5 0.3
x CD
Ql
' ' ' --""'' " 0.2
^*^
o ^^
. : o,r*r .0 0.1
**
*^X"o .
""' n
) 0.2 0.4 0.6 (
Measured
X
X
X
X
X
X
x
X
X
X
X
M X
X
X
X
X
X
x
X
x
x
x
x
x
x
x
o* /
^
x
A ซ
. x
x
. ~' .
.3. ;,ฃ % :
!'
x -
X
) 2 4 6
Measured
Hypolimnion HgT [ng/L]
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 ^5 t
x
ซJ o v
-vป *
/ 1 1 1 1 1
) 5 10 15 20 25
Measured
Sediment Cone, [ug/g]
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 ซ
'
x
X
/ . .1 %ฃ :." o0..
) 0.2 0.4 0.6
Measured
Figure 4.1. Predicted Cone, vs Measured Cone, with Lake Variables for Default Run. Dashed Line is y=x.
-------
Epilimnion MeHg [ng/L]
Epilimnion HgT [ng/L]
1.5
o
CD -i
-* ' 1
O '
CD
Ql
0.5
0
x
X
X
X
6
X
X
X
X
X C
x O
X
X
X
0 xl
_,' CD 4
X g
'' 2 3
X *J
/ Q.
x
x
X
x;' 2
x ซ
- X
X
.xf 1
xV/ซ."ซJ
m^f ฃf * * *^
X '^,ซ^?ft>mtf ^ป * **I ** 1 1 ^
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
^ ป
. X
* " x
:'..yx .-./.
,xv.-.:o7 ..;
xX V.ซ *.*.** 1 . .
(D -D -M y
0 0.5 1 1.5 0246
Measured Measured
Hypolimnion MeHg [ng/L] Hypolimnion HgT [ng/L]
4
3.5
3
o
& 2.5
g
^^
CD 2
CL
1.5
1
0.5
0
X
x
x
x
x
x
25
X
X
X
X
X
20
/' "O
S
.S2 IK
x -Q 13
ฃ
* . '' Ql
x
10
- X
X 9
X
X
X
/.' 5
< * "
o/4> . . "
x _ -
' < . n
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 * * ป
xซ* * %
x*
X *o
u
0 1 2 3 4 051015 20 25
Measured Measured
Fish Cone, [ug/g] Sediment Cone, [ug/g]
1.5
1
-o
-------
Epilimnion MeHg [ng/L]
Epilimnion HgT [ng/L]
1.5
o
-* * 1
o '
CD
Ql
0.5
C
4
3.5
3
0
& 2.5
g
^^
CD 2
CL
1.5
1
0.5
C
1.4
1.2
1
^
CD
^ 0.8
ol 0.6
0.4
0.2
C
X
X
X
X
6
X
X
X
X
X C
x O
X
X
X
'X "O ,
' CD 4
x 4ii
xx -^
xX CD o
x *J
x' Q_
X
x
. ' x>' 2
, X
X
X
X j
,-ป&&ฃ-" "V - . .
* ""* 0
) 0.5 1 1.5 (
Measured
Hypolimnion MeHg [ng/L]
X
x
x
x
x
x
25
X
X
X
X
X
20
xx "O
S
Q 15
x "D IO
s>
Ql
X
. 10
X
X
X
X
x*
' I-
5
*ป.*
t./
- < ". n
u
) 1 2 3 4 (
Measured
Fish Cone, [ug/g]
0.6
0.5
0.4
-o
CD
O
." **" Ql
^-'' 0.2
. . ^-'
.-.-;'''' ฐ-1
* jป ** * *
% *** * * *
-'"" "'* v .'. n
1 V I (J
) 0.2 0.4 0.6 (
Measured
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
^
xป .
X.
vi X'
"># *!:"
X-T/^i'/V "."."":
) 2 4 6
Measured
Hypolimnion HgT [ng/L]
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 .*
.0 X
x' *...- . %
x' *....-.
x o *
) 5 10 15 20 25
Measured
Sediment Cone, [ug/g]
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
xซ
.
. . ...
x
j"a,i./*r*r.
) 0.2 0.4 0.6
Measured
Figure 4.3. Predicted Cone, vs Measured Cone, with Lake Variables for Tier 2 Run. Dashed Line is y = x.
-------
Epilimnion MeHg [ng/L]
Epilimnion HgT [ng/L]
1.5
o
CD j
-* ' 1
O '
CD
Ql
0.5
0
x
X
X
X
6
X
X
X
X
X C
x O
X
X
X
0 xl
_,' CD 4
x' .^
>' CD o
X s- O
^
X
X
. . / 2
X *
x
qX
.*' 1
/{/*.ฃ
5^ ,งซ, * ' .
, v.wsk.f ป ..*. . . n
X
X
X
X
X
X
x
X
X
X
X
X
X
X
X
X
X
X
X
x
x
x
x'
x'
ซ xx
X
X
X
X
X
1 x
.-./
>iv~;.yr...' ;
n . y
0 0.5 1 1.5 0246
Measured Measured
Hypolimnion MeHg [ng/L] Hypolimnion HgT [ng/L]
4
3.5
3
o
& 2.5
g
^^
CD 2
CL
1.5
1
0.5
0
X
x
x
x
x
x
25
X
X
X
X
X
20
xx "O
S
Q 15
x "D lo
s>
Ql
X
. 10
X
X
x
X
xป
Xป I-
'*. X
t> * *
^t % *
' * n
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 *ฃ. *
x *o *
u
0 1 2 3 4 051015 20 25
Measured Measured
Fish Cone, [ug/g] Sediment Cone, [ug/g]
2
1.5
^
o>
o
ID 1
ol
0.5
0
0.6
0.5
0.4
-o
. ^
o
1 0-3
ol
^^- 0.2
t ^-'''
: " ^^'^. * ฐ-1
* I-*-**'.'** .
-***" *vซ*ซi* i n
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 own OซOM
/ .!?*** ^foซซ , ,
0 0.2 0.4 0.6 0 0.2 0.4 0.6
Measured Measured
Figure 4.4. Predicted Cone, vs Measured Cone, with Lake Variables for Tier 3 Run. Dashed Line is y = x.
-------
Epilimnion MeHg [ng/L]
Epilimnion HgT [ng/L]
1.5
o
-t1 ~\
0 '
CD
Ql
0.5
0
x
X
X
X
6
X
X
X
X
5
X
X
X
0 xl
x CD 4
x -iii ^
x o
X -^
x' CD o
Ql J
" ' x
X
2
.* x'' "
M 1
x 1 '
"ปf V .
V^^Sfe"**" "'''. . , n
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<
/ ?*''
?rCt5D* ปvป *
x? V .**""
/ .
. i i>i. y
0 0.5 1 1.5 0246
Measured Measured
Hypolimnion MeHg [ng/L] Hypolimnion HgT [ng/L]
4
3.5
3
o
& 2.5
g
"CD 2
0.
1.5
1
0.5
0
x
X
X
X
X
X
25
X
X
X
X
X
20
-o
52
x' ฐ IK
x' . ^15
. g>
x
x n
X
10
. X
x
. x
. x
''" ' '5
. . .X . ฐ
/"". '
X
- - . n
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'
.*xx:* ......
-'*ซ.
x V *
x
/ 1 1 1 1 1
u
0 1 2 3 4 051015 20 25
Measured Measured
Fish Cone, [ug/g] Sediment Cone, [ug/g]
1.5
CD .
o 1
ol
0.5
0
0.6
0.5
0.4
"8
o
ol
..."
^'-' 0.2
^ .* -* ""
**"** *
**
.'"'' 0.1
Jป"S *
p-*""* * * * 1 * *
->""*^ ****** r\
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
'' :***
xx ฐ* 2*. *'.".. I .
/ * *
X o o *ปo
0 0.2 0.4 0.6 0 0.2 0.4 0.6
Measured Measured
Figure 4.5. Predicted Cone, vs Measured Cone, with Lake Variables for Tier 4 Run. Dashed Line is y = x.
-------
Epilimnion MeHg [ng/L]
Epilimnion HgT [ng/L]
1.5
o
CD -i
-* ' 1
O '
CD
ol
0.5
0
x
X
X
X
6
X
X
X
X
x c
x O
X
X
X
0 xl
_,' CD 4
X g
'' 2 3
X *J
/ Q_
X
x
2
X
X
X 1
*
"al^Sy?**?*** ... -.
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 *
J *xป *
.j.f . /..ซ.
."JS*-o***0ซ .;** .
x * ^ *
. i i i Q
0 0.5 1 1.5 0246
Measured Measured
Hypolimnion MeHg [ng/L] Hypolimnion HgT [ng/L]
4
3.5
3
o
& 2.5
g
^^
CD 2
CL
1.5
1
0.5
0
X
x
x
x
x
x
25
X
X
X
X
X
20
/' "O
S
y 15
X "O IO
2
Ql
X
10
X
X
X
. X
* '* t i-
.x' ' 5
-'
^.'fc
.**? ' v . 0
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
^ซV i-' '."
u
0 1 2 3 4 051015 20 25
Measured Measured
Fish Cone, [ug/g] Sediment Cone, [ug/g]
1.4
1.2
-i
1
^
1 ฐ-8
^
CD
ol 0.6
0.4
0.2
0
0.6
0.5
0.4
-o
CD
O
TS 0.3
^-""' ol
.--'''' 0.2
***
' ^r::' ' ฐ-1
* "** *
J^""""ป** o
i -4* ^
***"* i n
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
XX
X
X
X
x'
/ ' "' ..
'' . "" o"~ ""..". . .
X fM
X*o*oซoซo
y
0 0.2 0.4 0.6 0 0.2 0.4 0.6
Measured Measured
Figure 4.6. Predicted Cone, vs Measured Cone, of Lake Variables for Tier 5 Run. Dashed Line is y = x.
-------
Error Sum of Squares for Default and Different Tiers
Default
Tierl
Tier 2
Tier3
Tier 4
TierS
EPI MeHg
7.9
10.6
10.5
10.9
10.1
10.2
EPI HgT
291.3
289.0
292.5
289.4
234.9
228.4
HYP MeHg
43.4
52.5
54.0
54.0
50.0
52,4
HYP HgT
4005.5
4949.6
4675.7
4680.9
4540.3
4958.9
Fish
4.6
4.0
3.7
6.1
4.8
3.6
Sediment
3.5
4.2
4.3
4.0
3.5
4.1
Percent Change
28.4%
-21.6%
20.7%
23.8%
-22.3%
17.6%
Error Sum of Squares (Residual Sum of Squares) for
Default and Tiers 1 through 5
10000.0
1000.0
100.0
1.0
10.0 rrr
EPI MeHg EPI HgT HYP MeHg HYP HgT Fish Sediment
Figure 4.7. Error Sum of Squares for runs and variables.
-------
Epilimnion MeHg [ng/L]
Epilimnion HgT
1.5
o
+ ' 1
0 '
^
CD
ol
0.5
0
x
X
X
X
6
X
X
X
5
X
X
X
S 4
ฃ
/ -^
X >KX X-)XXK '' 23
x Q_
xx xxX x x.' x x
xOQc 0 ^
X
0 CM-'
X *x >3Kx,/'''^ O 1
Wx< XO x
XXX :. X
x
x
x
x
X
X
X
. X
O /
x
. X X< XX XX X xx
x
X
X
ox a^o xxx x
x
x
>ฃxx 9<
x Ox .. ,' O
/w x^ V!? yy v c
(^/N /N /XWS/N v> ^ ^^ y
J^^ X " ^N "
XXX x
i p\
v#^4- -1-
xl i
U
0 0.5 1 1.5 0246
Measured Measured
Hypolimnion MeHg [ng/L] Hypolimnion HgT [ng/L]
4
3.5
3
o
& 2.5
o
T3
CD 2
ol
1.5
1
0.5
0
X
x
x
x
x
x
25
X
X
X
X
X
20
-o
S
x' ฐ IR
xx T3 I0
<$)x x X a!
x 0 X x 10
X
D x H X
rljij^r "T" ^ j
^7T=f- T X ^ ^ ^ ^
X
x
x
X
X
x
X
X
x
/
X
X
X
s'
s
s
/
s'
s
/
s
+xQx" Ofr O x O x ฐ
.J-^ l^ป Tpl y Ay V ^\
-**' x\ 1
-------
Epilimnion MeHg [ng/L]
Epilimnion HgT [ng/L]
1.5
o
+ ' 1
0 '
CD
ol
0.5
0
x
X
X
X
6
X
X
5
X
X
X
a /,
x" B 4
.y
X QS _
^^x ฃ
X
ooocPo ox"ฐ ฐ o
XXK )(. ^
X
0 Ort
"Sa? 1
X$KJ>ฃJJJI-' O X
. O ฉ OO OD O /'
/
/
xx **xx /' x
/
/
/'
Cฎ3D w xx
^xJxJ^O
QDO ^
.- vx
^//X\
U
0 0.5 1 1.5 0 2 4
Measured Measured
Hypolimnion MeHg [ng/L] Hypolimnion HgT [ng/L]
4
3.5
3
o
& 2.5
o
CD 2
ol
1.5
1
0.5
0
X
x
x
x
x
x
25
X
X
X
X
X
20
-o
S
15 15
xx x / 2
xx xx x Q_
. x x xxx x x IQ
X
X ^ X X
KXXX XX'
xฃQ$^A xx 5
x
x^Sj- x X
' i n
X
x''
s'
s
s
/'
s
/
s
,'
X X^^- "^K X X ^ x
^\x x^K xxx x
/X ^^K^ X
u
0 1 2 3 4 051015 20
Measured Measured
Fish Hg [ug/g] Sediment Cone.
1
0.8
^
CD
o 0.6
CD
ol
0.4
0.2
0
0.6
GSD 0.5
x ฐ 0.4
0
O '' <*>
s' O
1 ฐ-3
x x ฃ
o o o P-" ฎ n9
_,..- x ^^'^
x xx,.-''' x
x x^
X X XvXxX ซ< 0.1
V rf""* "
T^'x x x
' i n
/
/
/
/
s
/
/
/
s
/
0 / 0
/
o /'
*-* /
x /
/ x
,* \f
Ci
X^^Sp^xO x ฐx
/ O ^*Gฃ (^*O
/ f-. /^t^S^BX/cX^, X
u
0 0.2 0.4 0.6 0 0.2 0.4
Measured Measured
/
,'
/
x 0 o
6
X
X
X
X
X
x
X
X
x
X
X
25
X
X
X
X
X
X
X
X
X
O xx
0.6
Figure 5.2. Default Results. Lakes Separated by Stratification: o: Well Mixed, x: Stratification.
-------
Epilimnion MeHg [ng/L]
Epilimnion HgT [ng/L]
1.5
o
+ ' 1
0 '
CD
ol
0.5
0
x
X
X
X
6
X
X
5
X
X
X
a /,
X" B 4
.y
X QS _
Ui>x ฃ 3
X
ooocPo ox"ฐ ฐ 0
ooe> 0 ^
x
O Orf
fe^xO x 1
/^^BOO X x'
XX XX X X (
C&00xx 5
,*m7x 0 0
X^ O X} Cx O
y . . . | rป
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
^('^ OQX X Q
*3ฃฐ1& S&31 ^ x
x ^^ /\y X t_/
u
0 1 2 3 4 051015 20
Measured Measured
Fish Hg [ug/g] Sediment Cone.
1
0.8
^^
CD
o 0.6
CD
ol
0.4
0.2
0
0.6
C5D 0.5
0.4
^' T3
O ^x CD
^x O
o ,''*" 1 ฐ-3
-''K ฐ-
X X X s r\ o
x^x x *^
x xx,.-''' x
O O O^kO ซ< 0.1
^ปX*JC X *
X?
'' i n
X
X
X
X
x
x
X
x
x
x
o xx o
x
o ^'
v-/ X
0 x'
X x o
y^^tegftgb o ฐo
xx /ฉt^gj^fcx/^b^ X
yX O ^^rrS^)^eCxJ3^"s O
U
0 0.2 0.4 0.6 0 0.2 0.4
Measured Measured
X
X
X
X
X
X
X
X
x 0 0
6
X
X
X
X
X
x
X
X
x
x O
o
o
25
X
X
X
X
X
X
X
X
X
0 GD
0.6
Figure 5.3. Default Results. Lakes Separated by Lake Size: o: Small, x: Medium.
-------
Epilimnion MeHg [ng/L]
Epilimnion HgT [ng/L]
1.5
o
+ ' 1
0 '
CD
n
0.5
0
X
X
X
X
6
X
X
5
X
X
X
a /,
$> 4
x O
X -^
, .ui. /^ CD o
^/ ol
XX XX X ^x p
x
fit* * * 1
x X^K o q n
x
x
x
x
X
X
X
*
*vU> \lxxL- \ly\L- vL- *
/
X
^. ^ ^ xx
^ ^^ xx 5K1
x
X
x
XK XK /*ป x
^*|*x>|i* * +
/(Jy O^ ^
x
iii y
0 0.5 1 1.5 0246
Measured Measured
Hypolimnion MeHg [ng/L] Hypolimnion HgT [ng/L]
4
3.5
3
o
B 2.5
o
CD 2
D-
1.5
1
0.5
0
x
x
x
x
x
x
25
X
X
X
20
-o
52
x' - 1 R
x "0 I 3
ax^ *D
x >-
x n
ฎ * xx 10
X
-------
Epilimnion MeHg [ng/L]
Epilimnion HgT [ng/L]
1.5
o
+ ' 1
0 '
CD
ol
0.5
0
x
X
X
X
6
X
X
X
5
X
X
X
a /,
S3 4
^
x "^
'' 23
X *J
/ Q.
x
x
Vo' 2
X^hifc ฃ x
*?>ป$&? *& *Wx v x . . x n
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^
//"^ "
1 >^^^F c^ฐ^ฐ^x ฐ
xx"r + n .
VJ
0 0.5 1 1.5 0246
Measured Measured
Hypolimnion MeHg [ng/L] Hypolimnion HgT [ng/L]
4
3.5
3
o
& 2.5
o
CD 2
D-
1.5
1
0.5
0
X
x
x
x
x
x
25
X
X
X
20
-o
S
- 15
1
*
10
X
X
X
' Xx,X*0 @ 5
*35xO ,
Sj^ xxx x x x
/*' / \ ~ . . . . r>
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
o
X
X
X
X
X
X
X
XX!U LJ~Ckx ^ x O
-I^Fh ?p^.9Sr jjjX x p
A U
0 1 2 3 4 051015 20 25
Measured Measured
Fish Hg [ug/g] Sediment HgT [ug/g]
1.4
1.2
-i
i
^
CD 0
o 0-8
^
CD
ol 0.6
0.4
0.2
0
0.6
0.5
0.4
v -o
X CD
x -5
13 ฐ-3
-'" ^
x ,-'*' ฃ
\ -'"'**' ฐ-2
+ X ฐ-*""'' X ฐ'1
-** i v Oi i r\
X
X
X
X
X
X
X
X
X
X
X
X
X
x
x
X
X
x
x
x
x
x
x
o
X
ox /
x*4-
/ ~h
'0 jfejc x x
x/x - x -xoo ^
.^ ^ XXX X y X X X O
X \y V" V ^A^v'" id
X^ X X*' ^ ^^ /O . . X
A ^w \J
0 0.2 0.4 0.6 0 0.2 0.4 0.6
Measured Measured
Figure 5.5. Tier 5 Results. Lakes Separated by Acidity: o: Acidic, x: Circumneutral, +: Alkaline.
-------
Epilimnion MeHg [ng/L]
Epilimnion HgT [ng/L]
1.5
o
+ ' 1
0 '
CD
ol
0.5
0
x
X
X
X
6
X
X
X
x c
x O
X
X
X
a /,
S3 4
^
x "^
'' CD Q
X *J
/ Q.
X
x
0 0 / 2
x'O o
CL/O 1
ฃi5lM&?hฎ ^
/iaMJirolx^ x x %
i"> A U
0 1 2 3 4 051015 20 25
Measured Measured
Fish Hg [ug/g] Sediment Cone.
1.4
1.2
-i
1
7^3
1 ฐ-8
^
CD
ol 0.6
0.4
0.2
0
o
0.6
0.5
0.4
o
CD
o ts
T3 0 3
^^-" CD
0 .''*' ฃ
^-'" 0.2
Xx-"**
vv -'** ^ n i
x.-^o x
NX O'X^^V'^ yy C^d NXX 3C
tyfy\ "*7\ " " 'y ^^/ /JK "
X
X
X
X
X
X
X
X
X
X
X
X
X
x
x
X
X
x
x
x
x
x
x
x
o
X
GO /
xt^)
/xxfc9ฐ,o
x ,r jv jiffpy^c . - o x o ox
X Q^XV^^K^^^KX vX . x
*" i X X ' i Q /N /s
0 0.2 0.4 0.6 0 0.2 0.4 0.6
Measured Measured
Figure 5.6. Tier 5 Results. Lakes Separated by Stratification: o: Well Mixed, x: Stratification.
-------
Epilimnion MeHg [ng/L]
Epilimnion HgT [ng/L]
1.5
o
+ ' 1
0 '
CD
ol
0.5
0
x
X
X
X
6
X
X
X
X
x c
x O
X
X
X
a /,
S3 4
^
x "^
'' 23
X *J
/ Q.
X
x
0 0 / 2
xฐ ฐ
a1
_ Q5> Q
^WOn n . x . v n
X
x
x
x
x
x
x
x
x
x
x
x
O X
X
X
X
X
x
x
x
x
x
x
x
x
x
(J XQ O
' oJllSk P^>X^ฐ 00ฐ
oiBffi^!)jDgj"^D O O x Q
x/O^-^; v .
0 0.5 1 1.5 0246
Measured Measured
Hypolimnion MeHg [ng/L] Hypolimnion HgT [ng/L]
4
3.5
3
o
& 2.5
o
^^
CD 2
D-
1.5
1
0.5
0
x
x
x
x
x
x
25
X
X
X
20
/' "O
S
.S2 IK
" "O IO
ฃ
Ql
X
10
X
X
.0 x'
xrv. x'o 5
f XO O
xX*O x x Q
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
ฐ^fe)x^i^'lc^x ฐ ฐ ฐ
^^ XX;K -X ฃ* x x ฉ
0 1 2 3 4 051015 20 25
Measured Measured
Fish Hg [ug/g] Sediment Cone.
1.4
1.2
-i
1
ฃ3
03
f> 0.8
^
CD
ol 0.6
0.4
0.2
0
n
0.6
0.5
0.4
v -o
X CD
o -ง
TS 0.3
0 .''*'' ฃ
^-'" 0.2
O -^'*''' x
nl-rf^ x ^
^W^ ^^Qj) ^ Y Y rป
X
X
X
X
X
X
X
X
X
X
X
X
X
x
x
X
X
x
x
x
x
x
x
x
o
X
ox /
xt^)
xb xv* ฐ ฐ
/XjปSxXx * 0 0ฐ 0 0
/x XQXV? x x >^wraT^ ^D . . O
*" ' ' ' ' ' U "* -"
0 0.2 0.4 0.6 0 0.2 0.4 0.6
Measured Measured
Figure 5.7. Tier 5 Results. Lakes Separated by Lake Size: o: Small, x: Medium.
-------
Epilimnion MeHg
Epilimnion HgT
1.5
o
+ ' 1
0 '
CD
Ql
0.5
0
X
X
X
X
6
X
X
X
X
X C
x O
X
X
X
a /,
' CD 4
X Jฑi ^
.9
X "Q
x' CD o
X *J
xx Q_
X
x
x
x x/ 2
xX+ +
^\ ^
x^ 1
fSLiaJk *
X^Mnftq wPp1?^. V^fc 4fc . -5fe- | ^f. r\
X
X
X
x
x
x
x
x
x
X
X
X
O x"
X
X
X
X
X
X
X
x
x
x
x
x
x
fL v * ^ /
$r
^fe;
^9^ 'o ^Tfefex^x^-
lljii^j^ - *+ * ซ
xx ^ Xj ^t .
\J
0 0.5 1 1.5 0246
Measured Measured
Hypolimnion MeHg Hypolimnion HgT
4
3.5
3
o
& 2.5
o
^^
CD 2
D-
1.5
1
0.5
0
x
x
x
x
x
x
25
X
X
X
X
X
20
-o
S
y 15
x "D lo
2
ฃ
10
X
X
. x
X XS *: r-
<^xx/^ - 5
^K^m^ Q?^ X -^- "^ i^
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
v^y* "t^-^X "I x *
^^ ^X *^ (^ * i ฉ
*r ^ U
0 1 2 3 4 051015 20 25
Measured Measured
Fish Cone. Sediment Cone.
1.4
1.2
-i
1
ฃ3
3 0.8
ID
ol 0.6
0.4
0.2
0
0.6
0.5
0.4
o
X CD
U/ -1 I
0
1 ฐ-3
x ,-'*' ฃ
. \ ^"'" ฐ-2
+ x ฐ-ป'"'"' * ฐ-1
** *Iu VM ^f-
^ *^*k y^ ^K-X> ^ %3f (^
X
X
X
X
X
X
X
X
X
X
X
X
X
X
x
x
x
x
x
x
x
x
x
o
GO /
xx// X X
x 7Jg> j^Mfe-XT ^ O ^K X $|fc
^ X^^^j[^^^^^'\j|feX -5|6
/x X ^ ^5c ^>?K . . ^
0 0.2 0.4 0.6 0 0.2 0.4 0.6
Measured Measured
Figure 5.8. Tier 5 Results. Lakes Separated by Trophic Status:
o: Oligotrophic, x: Mesotrophic, +: Eutrophic, *: Dystrophic.
-------
, ,
ฃ0.4
O)
X
CD
I 0.3
g
c
E
1.0-2
LU
CD
1 ฐ-1
CD
ol
0
C
x O v=0.5 m/s
x v=1 .0 m/s
0 + v=10m/s
x.
ฎ x
X X
+ฎ ^
ฐ+ +
-jฃ* i
'xj&x* +
J^p^*ฎ*
) 0.5 1 1.5
Observed Epilimnion MeHg [ng/L]
2.5
IT
c
' ' 2
x
CD
1 1'5
E
8. 1
X
1 0.5
CD
Ql
n
+
O v=0.5 m/s
x v=1 .0 m/s
+ v=10m/s
h +
X
x x +
JฃQ* ~&+ + ^
Sฐ/S0 x +
aง3EPง> tf\ x jo O X
W-L. + .
012345
Observed Hypolimnion MeHg [ng/L]
j
O) ji
f A
\-
O)
X
c 3
0
'c
E
1. 2
LU
T3
CD
t5 H
=5 1
CD
ol
0
^ O v=0.5 m/s
x v=1 .0 m/s
+ v=10m/s
6^ ^ ~1~ S)
Tii_ i^p. Q ^r
-p/Mm^-i.yC
^?r^ ^v*t
j|p X '^TJ|Q|Jf& O
M/^Jง^^^ \5^ ฎ ฎ ^
^jS^pZjJfe^ ^" ฎ + ฎ + ffi
^BJBLx^roSc +
'^"Frfr Ijrtf I , rT~ i i
0246!
Observed Epilimnion HgT [ng/L]
30
mr
ฃ 25
t
O)
1 20
c
o
'c
.i 15
o
Q.
>,
^ 10
T3
&
O
=6 5
CD
5^_
CL
Q
+ O v=0.5 m/s
x v=1.0m/s
+ v=10m/s
~r
I
~r
i
/5?\
' + + + + 4 + +
~h ~h
++++ ^ + +
H~ ~ts sK-
s^ "^"Y"^ ฉ 0*1^1 V r^ 6& V ^ H*
^ J?$tf& (# 6 rt ^
^^ etoฎ^ , *-^ _i_ , ~^
0 10 20 3
Observed Hypolimnion HgT [ng/L]
0.35
!3 0.3
O)
I- 0.25
O)
X
1 0-2
E
IS 0.15
! _l_ V V V
,f Be/ % ^+ ^ (
IjyiiSK &. ?
0.2 0.4 0.6
Observed Sediment HgT [ug/g]
0.8
Figure 6.1. Effects of Settling Velocity on Predicted Mercury Concentrations.
-------
15
0.5 1 1.5
Observed Epilimnion MeHg [ng/L]
2.5
x
CD
"5
Q.
T3
t> 0.5
s>
ol
0
+ +
-$-
012345
Observed Hypolimnion MeHg [ng/L]
O)
10
o
'c
E
246
Observed Epilimnion HgT [ng/L]
O)
c
o
o
Q.
CD
O
CD
Ql
10
8
6
4
2
0
o
0 10 20 30
Observed Hypolimnion HgT [ng/L]
0.2 0.4 0.6 0.8
Observed Sediment HgT [ug/g]
Figure 6.2. Effects of Photoreduction (default rate and none) on Predicted Mercury Concentrations.
-------
_Cl O ^" ^^
pH i f^ "ฐ
Srง + V+ ง
&)*0 ฐOQL + 1 2
ฎcP * D * ^
T + 1 i i i i n
r *
0 Default
+ None&v=10m/s
+
~r
-O '
Dฎ^ ^ 9 o o
01234
Observed Hypolimnion MeHg [ng/L]
0 10 20 30
Observed Hypolimnion HgT [ng/L]
_0.7
-CJ
i'o.e
CD
E 0.4
CD
-------
Epilimnion MeHg
Epilimnion HgT
0.2 0.3
Observed
2.5
=0 1-5
CD
Ol
0.5
1 2
Observed
Hypolimnion MeHg
1.4
1.2
CD
Ql
0.6
0.4
0.2
0
0.5
1
Observed
1.5
15
Hypolimnion HgT
10
CD
CD
Ql
5 10
Observed
15
Figure 6.4. Predicted vs. Observed Epilimnion and Hypolimnion Mercury Concentrations
for the Hypolimnion Area Sensitivity Runs.
-------
Sediment MeHg
0.03 -
0.005
0
0 0.005 0.01 0.015 0.02 0.025
Observed
0.35
Sediment HgT
0.2
Observed
Fish Tissue Hg
0.35
0.3
0.25
-------
2 i-
1.8
1.6
1.4
O)
O)
X
Q_
LU
1.2
O
T3
CD
0.8
0.6
0.4
0.2
I
I
I
I
I
0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
R, Hypolimnion/Epilimnion Surface Area Ratio
0.5
0.55
0.6
Figure 6.6. Hypolimnion Area Sensitivity Analysis. Predicted Concentration vs. R, Hypolimnion/Epilimnion
Surface Area Ratio. Each lake is connected by a line.
-------
ฃ
g
1
c
IP
o
o
O
IP
ts
0)
o
0)
U)
o
Q.
0) 05
ซ 0)
IZ O
to >,
ฃ -c
"1.1
i ฃ"
(O
O)
-------
[-|/6u]
-------
re o>
m
0
01
^
0
1
E
0)
0
0
0
a
01
"5
T3
0)
=1,
ง
'E
M
^
^
X
ai
(0
o)
^
D)
L_
in
csi
CN
/
ID
/
If)
O
/I
O LD
[l/6u]
-------
c
01
o
o
O
-a
a)
"a-
Q.
d
re
0)
o
0)
a)
o
o
.0
re
I
g
'c
"5
o
(6
D)
CM
/
O
CO
CM
[l/6u] uojiEJjuaouoo
-------
a
/
3
CD
/
m
CO
0
CD
/
00
0
CD
/
in
CM
0
CD
/
CM
0
CD
/
m
0
CD
7
0
CD
/
0
0
CD
/\
0 in
CM
0
CD
a
I
c
0)
o
o
O
1
c
(U
"5
-------
in
o
E
c
01
o
o
O
-d
HI
O
T3
S
CL
0)
O
0)
U)
O
Q,
U)
0)
a:
a)
0) _
I!
fl
CN
(6
I
D)
[6/6n]
-------
(/>
g
'-*-*
I
o
o
O
TJ
OP
CL
T3
re
0)
O
W
o
a.
U)
0)
a:
5 ซ
< o
to
O
1 E-
C Z3
1U O
|P
I-s
I-
-^.o
(O
D)
UOIJEJJU30UOO
-------
o
-------
-------
o
o
O
ir
K
DC
;o
O
T3
CO
Q
CD
s
CD
0_
g
*^
'o
CD
Q_
CO
CO
o
'cป
CD
DC
CC =3
CD O
w CD
'
ffl.Q
CD
CD >
c c
O CD
cW
Eo3
ฐ
CD
CD
CD
13
O)
l.=a 'h/6u]
-------
o
o
O
DC
K
DC
T3
O
T3
CO
Q
(M
(M
-------
in
o
in
CM
o
r>
o
o
o
O
ir
K
DC
;o
O
T3
CO
Q
-es-
un
oo
o
oo
O
un
CM
O
CD
CD
D_
g
^
'o
CD
Q.
CO
"5
w
CD
DC
m
CD
W CD
'w E
>^>,
m .c
E
CD >^
o -2 g
o "CD
-
g ป
O CD
CD
O)
un
o
un
(M
o
(M
o
CD
un
o
CD
O
o'
un
o
o
o
l.=a '[6/6n]
-------
l.=a '[6/6n]
-------
l.=a '[6/6n]
-------
1.5
o
o
O
O)
X
c
o
I 0.5
Q_
LU
0
0 0.2 0.4 0.6 0.8 1
R: Hypolimnion Area to Lake Area
2.5
o
o
O
1.5
o
Q.
0.5
0
MeHg
Hgll
HgT
0 0.2 0.4 0.6 0.8 1
R: Hypolimnion Area to Lake Area
0.18
0.16
0.14
io.12
^
d
ง 0.1
O
O)
;E 0.08
c
CD
I 0.06
CS)
0.04
0.02
0
0 0.2 0.4 0.6 0.8 1
R: Hypolimnion Area to Lake Area
Figure 6.21. Hypolimnion Area Sensitivity Evaluation. Hypothetical Default Model Runs with Changes in R.
-------
1.4
1.2'
ง0.8
O
O)
ง0.6
1&0.4
0.2
MeHg
Hgll
HgT
200 400 600 800
Mean Hypolimnion Depth [m]
1000
200 400 600 800
Mean Hypolimnion Depth [m]
1000
0 200 400 600 800 1000
Mean Hypolimnion Depth [m]
Figure 6.22. Hypolimnion Area Sensitivity Evaluation. Default Model Runs with Changes in Mean Hypolimnion Depth.
-------
3.5 r
2.5
O
O)
X
'5.
LU
0.5
3.5 r
Well-Mixed,
Top Layer Only
Well-Mixed,
Both Layers
Modeled as One
0.2 0.4 0.6 0.8 1
Mean Hypolimnion Depth [m]
Well-Mixed,
Top Layer Only
Well Mixed,
Both Layers
Modeled as One
0.2 0.4 0.6 0.8 1
Mean Hypolimnion Depth [m]
0.2
0.18
0.16
-50.14
0.12
0.1
O)
i
o
o
O
O)
X
ง 0.08
$ 0.06
0.04
0.02
0.2 0.4 0.6 0.8 1
Mean Hypolimnion Depth [m]
Figure 6.23. Hypolimnion Area Sensitivity Evaluation. Default Model Runs with Wider Range Variation in R, and
Well Mixed Models with Dimensions Similar to R = 0.95 and R = 0.
-------
Epilimnion CT
Hypolimnion CT
0.16
0 0.2 0.4 0.6 0.8 1
R: Hypolimnion Area to Epilimnion Area
0.34
0.32
0.3
0.28
O)
0.26
0.16
(b)
Epilimnion CT
Hypolimnion CL
0 0.2 0.4 0.6 0.8 1
R: Hypolimnion Area to Epilimnion Area
O)
6.5
6'
5.5
5
4.5
CD
O
o
3.5
3
2.5
(c)
Epilimnion CT
Hypolimnion CT
Input Parameters:
Cjn = 15 ng/L
Epi Area = 1,000,000 m2
(a) and (b)
Q. = 25,000 m3/d
in o
Qout = 25,000 m3/d
Depth . = 5 m
= 5 m
Depth
Hyp'
(a) v1 = 1.0 m/d
v2 = 0.5 m/d
(b) v1 = 0.5 m/d
v = 1.0 m/d
(c) Qjn = 35,000 rrT/d
cTt = 35,000 m3/d
DepthEpj = 8 m
Depth,, = 8 m
Hyp
v1 = 0.018 m/d
v = 0.040 m/d
0 0.2 0.4 0.6 0.8 1
R: Hypolimnion Area to Epilimnion Area
Figure 6.24. Hypolimnion Area Sensitivity Evaluation. Output from Simple Mathematical Formulation
of Arbitrary Lake System with structure of Default Model.
-------
D)
"OS 1
C
O
c
.ฃ.5
'5.
LJJ
1(
6
51
^5
oil
^
.23
E
12
LJJ
1
1(
O)
-3
CD
.22
E
g.1
l"
1(
30
U
SO
D)
^15
0
ElO
"o
Q.
1ฐ
1(
1.
! 5
^is;v.-:
Dฐ 102 1
%
.
* * "
V;...; .
"X *f|6*ซ* * *
5r *
Dฐ 102 1
.
.
m
*.. "
*ฐ ซ*.ฃ .*
Dฐ 102 1
e
^ป *
0 -,
""* " .*
Dฐ 102 1
Lake Area [ha]
1.5
1
0.5
D4 <
6
5
4
3
2
1
D4 <
4
3
2
1
D4 <
30
25
20
15
10
D4 <
ป
. 8 . . ฐ
.'jS(ง?.'V
) 5 10 1
.
"* *
' o*^*"'
ป..
) 5 10 1
" .
f.
tj$Sr *."
) 5 10 1
0
* ซ?
ซ."* *
8* *
) 5 10 1
Epilimnion DOC [mg/L]
1.5
1
0.5
5 i
6
5
4
3
2
1
5 i
4
3
2
1
5 <
30
25
20
15
10
5 i
ป
" s * *
;: .ซป. ." *ซ
. :*/c*..
* '"' tcp?!f*
I 6 ซ
" . '
." .
"* ซ" * '^jl
\ 6 i
m
.
ฐ *
** .'
I 6 ซ
Epilimnion pH
1.5
1
0.5
3
6
4
2
3
3
3
8
3 8
8 I
1 1.5 2
8
.
5
1 8
I
1 i
1 1.5 2
Figure 6.25. Observed Mercury Concentrations versus the Default Level
Classifications/Characteristics for the VT and NH Lakes Dataset.
-------
0.02
[3
!>0.015
D)
n:
CD
-ฃ 0.01
CD
E
'-o
CD
OT 0.005
Q
"*
.
ป
** *
**
V. ป o. " ..
ฃ no o ซป
MOD *
0.02
0.015
0.01
0.005
n
* *
..*
" .
.
-*ป* : " . *
goto M
ฃ>
0.02
0.015
0.01
0.005
n
* *
t
* * *
* *
9 *ป*** tป*
0.02
0.015
0.01
0.005
n
0
'
0
5
a 2
i \
8 o
t
10o 1Q2 1Q4 0 5 10 15 4 6 81 1.5 2
07 n -7 n -7 no
. /
0.6
roฐ-5
I5
t,0-4
i
| 0.3
T3
CD
OT 0.2
0.1
Q
*
m
1 . m
V^ :
ป..
^*''< '- '
." ***
a
v. /
0.6
0.5
0.4
0.3
0.2
0.1
n
.
m
9
.
"lซ-.
*L. *
$ฃ':"
:* . .
0
u. /
0.6
0.5
0.4
0.3
0.2
0.1
n
'
.
*
.
.'... -._ ;
. * *
;&->''
> * * * *."ป
u.o
0.7
0.6
0.5
0.4
0.3
0.2
0.1
n
.
.
'
. ^
8 *
J ฐ -
. ง
e .
9 ง
0
*
10ฐ 1Q2 1Q4 0 5 10 15 4 6 8 1 1.5 2
.O
0.7
"ro
lo.6
CD
1 0.5
i
"5 0.4
CD
Qz-
W ฐ'3
iZ
2 0.2
-g
CD
ฃ 0.1
n
. .
.
.' "
*
.. '.
, *
"."
* *
U.(J
0.7
0.6
0.5
0.4
0.3
0.2
0.1
n
.
.
.
*
' *
f
?rฐ
0
.
~: '
.
*
U.(J
0.7
0.6
0.5
0.4
0.3
0.2
0.1
n
.
m
_
.
.. .
9 ป
* ** ซ . .'
*
U.(J
0.7
0.6
0.5
0.4
0.3
0.2
0.1
n
0
0
8
s
1
o " -
1
ฃ
e ฐ
Q
10ฐ 1Q2 104 ฐ 5 10 5 6 7 8 1 1.5 2
.. ... . ^^r , ...,..., 1 -Stratified, 2-Well Mixed
Lake Area [ha] Epihmmon DOC [mg/L] Epihmmon pH
Figure 6.26. Observed Mercury Concentrations versus the Default Level
Classifications/Characteristics for the VT and NH Lakes Dataset.
-------
Predicted Results
Observed Results
yu
40
E
^ 30
O)
X
CD
2 20
10
0
.
.
ป
.
. * . ฐ .
' *
* .. . " '. ..* .*
ป ป. .' * :
ou
70
60
50
40
30
20
10
0
.
*
ป.
. *'., *'
'.. ซ
.' . .'t
' ซ*..*. '
" '* - "*. . ' .f
*
ou
ง70
'c
|BO
ง>o
O)
X40
CD
^OOA
CD
c^ U
^3
CD
ฃ10
0
.
\'
* '
Trf ซ^s.* \ ซ
_ *
0 50 100
Observed %MeHg in Epilimnion
40
35
30
0.
|25
X
CD
215
5?
10
5
0
.
*
mป
^
.
.*'*
.
i A i i i i
40
35
30
25
20
15
10
5
0
*
^
.
ฐ *.****
""*.* *"
'. * .' . '
40
|35
I
130
Q.
^25
c
^20
CD
p!5
_
B 10
o
CD K
s_ y
CL
0
.ซ
*
*
."
'' '
*
.
^ t
0 20 40
Observed %MeHg in Hypolimnion
40
35
30
o
$25
c
cป20
X
CD
^0
n r\
1 U
5
n
. * . . .
.
. * ฐ ...
%ป ^ ซ
"** *
30
25
20
15
10
5
n
9
M **
./ซป* ,*."/ ,ฐ.ซฐซ-"-
40
c 35
CD
E
=5 30
CD
.E 25
O)
o5 20
^
5s 15
~o
CD
T^ in
.Q ID
CD
Ql 5
n
.
.
.
" ป " *
is%8!
0 20 40
Observed %MeHg in Sediment
Figure 6.27. Percent Methylmercury in total mercury in the hypolimnion, epilimion and sediments. For the first two
columns of plots, the data are plotted in order of lakes in database (x-axis is arbitrary). The third column
is the percent methylmercury of the predicted Tier 5 data versus the observed percent methylmercury.
-------
1.5
o
n 1
Q.
;ฃ
DC
0.5
0
X
X
X
x
1.5
X
X
x
X
X
1
X
x
X
/
/
' '/ . ฐ-5
.xซ
ywu* v
^AtiHv***** . rt
2.5
X
x
x
X
2
X
X
1-5
x'
X
1
. x'
" , x'
AY.' ' 0.5
2^\ ป
45$ A *
a^nR^^ .** * . . . rป
"
X
X
^r
X
x
xx
X
-.' ' x-x
X
.V X
1 **** *
^ffj* "* *
^JTRTy . .
^ U U '
0 0.5 1 1.5 0 0.5 1 1.5 0 0.5 1 1.5
1.5
n 1
Q_
z>
E
0.5
0
x'
x
1.5
X
X
X
x
X
1
x '
X
X
X
X
X
' V.'. 0.5
X
X
fjSftf?*^' . * n
3
X
X
2.5
2
X
X
. x
x' 1.5
X
x
1
X I
X
B xซ
:%'... 0.5
^Wl"?* * " " * * n
,
x
. ' -x'
. x'
X**"
^^
"JL** "*''
^Sjr?j/* %
ซiI^R\ . . .
* ' ' ' 0 u
0 0.5 1 1.5 0 0.5 1 1.5 0 0.5 1 1.5
1.5
CM
U 1
Q_
^
E
0.5
0
X
X
X
x
2
X
x
X
X
1.5
X
x
X
x
1
. X
0.5
/^*<^% 1^ rt
.
3
X
X
^^
2
X
X
* X
X
.':": ซ--''' 1
r^^t i 1 1 n
1
ป "*"
'&:..: ^"'
^StfLA"**^ * *
0 0.5 1 1.5
R2Up = 0.1
0 0.5 1 1.5
R2Up=1
0 0.5 1 1.5
R2Up = 2
Figure 6.28. Predicted (y-axis) versus Observed (x-axis) Epilimnion Methylmercury Concentrations for different
combinations of R1 Up and R2Up. Default/Baseline case is R1 Up = 0.1, R2Up = 0.1.
-------
6
_ 5
4
Q.
zi 3
~
2
1
0
25
/' 20
/
15
/
. ../ 10
f}& * <. K
._ปC." . . 5
/*.." . . ri
50
40
30
20
* B
0&^.;^r-~'rr~. 10
^t^-i'fT' * . . " r,
;&.,V:v; . -.
^feV-H*^""" ^~
U u
0246 0246 0246
6
5
'ii 4
Q.
z> 3
01 2
1
0
30
25
20
xx
15
../ 10
." ^
jfe/ .<[' " . 5
''* * * n
.
50
40
30
20
.'. .
.'ซ:'."-'-"".".' 10
_^t\-ป*fT' * . . " n
.
B.
''.^i> ;""' .* _ .
4fc\^s*-7ป~~ *
U u
0246 0246 0246
6
5
CM
ii 4
Q.
=3 3
ฃ 2
1
0
30
25
20
/'
15
.'"S 10
* *" ** ซ.
ss^r *ซ* * ฐ *
' * * i n
50
40
30
20
.*".
-------
4
- 3
o
ii
t2
E
1
0
x'
4
X
X
X
X
3
X
X
X
2
X
x
X
'. ''' 1
X *
' ** * t i n
/
6
xx
4
;. . ' / 3
''. ' /' ' 2
/ '
** */ ซ 1
k'
x
X
X
r ป . xxx
V. ' X'
4 xx
ซ* . xx
V
,:*T'
U u
024024024
4
3
^2
DC
1
0
x'
4
X
x
X
X
3
X
x
X
2
X
X
4 x
.'...X'-'' ' 1
*ซ.' *ซ
*'- - " - o
,
6
X
5
4
3
""ป''. 2
>* . x' '
* / * * 1
V*
x^= . . .
.
x
x
* x
ซ .
" ' x''"
>' . ,'-x'
'* xx
Vr-
U u
024024024
4
3
(M
II
^2
DC
1
0
4
X
X
X
3
X
X
X
x
2
X
X
x
. ' .
.... V 1
jซ. V. .-
V--
' e ' n
6
5
4
* ' xX/ ' 3
x
'. 'x ' ' 2
* / '
ฐVX 1
xx
X
o xX
x
4 4 x
^x
1* "* ^x
e x
* * X
U u
024024024
4
3
10 ฐ
II
^2
E
1
0
x
X
'' 4
X ^
x
X
x
X
3
X
x
, x
' . / 2
.. x
{" " X .
-S / 1
*v -
*
/I .
' n
**
6
X
X
1 X
x
X
* ^
X vl
: . /
x
ป x
; . X . ' 2
V
t
* * x
x
X
x^
X
;-: /"..
'' x'''
"* *
U u
024024024
R2Up = 0.1 R2Up=1 R2Up = 2
Figure 6.30. Predicted (y-axis) versus Observed (x-axis) Hypolimnion Methylmercury Concentrations for different
combinations of R1 Up and R2Up. Default/Baseline case is R1 Up = 0.1, R2Up = 0.1.
-------
25
^. 20
o
A. 15
'
E 10
5
0
/
25
X
X
20
X
X
15
X
/
10
X
X R
xX* -. .
'ป n
/
25
20
15
,ฃ
/: 10
1 X
/':" . . 5
''/'.' . .
X . *
X
X
X
X
x
xX
X
x'
* x'
x
X
X
x
. t x
/
X
* X
" x . .
x'
X
U \J
0 10 20 0 10 20 0 10 20
25
20
II
o. 15
E 10
5
0
,
25
X
x
20
X
15
X
ป X
10
'''... 5
H^'io at"*"'* * '
' '" n
,
25
X
20
X
' .- x/ ^
* ^
"X '. ' 10
ซ x"
*..-. 5
,x-'. O
V'.t . .
x .
/ . rป
X
X
X
x
X
x
_ x
x
X
" X
. . . x
'x
. X
.x .
. : x *
x
xiV .
'. /.
x
x
^
x
U u
0 10 20 0 10 20 0 10 20
25
CM 2ฐ
Q. 15
1_
DC 10
5
0
,
25
X
20
/
15
X
, X
10
X
/ .... . 5
/^* ซ* * * * * r>
,
25
20
X
X
15
X
* * X *
./ . 10
ซ *
'.X
'/'xx .'''" * 5
X *
X
X
x
X
x
/
x
X
X
X
, x
..." .'X
V . .
X
B x
"/ .
xX
0 10 20 0 10 20 0 10 20
25
20
||
o. 15
'
E 10
5
0
,
25
X
20
X
15
X
10
*x
'*** 5
."/"&0 ot* * " . "
' n
X
X
X
25
X
X
20
X
X
./ 15
X * *
' . .X . 10
X
**"*"x/t*ซ"*
x **
X
X
x
X
x
X
X
x
X
"X
x
X
X
x
. X
x t
.""' .'/
' '
' ''.
x .
xx ^ .
U u
0 10 20 0 10 20 0 10 20
R2Up = 0.1 R2Up=1 R2Up = 2
Figure 6.31. Predicted (y-axis) versus Observed (x-axis) Hypolimnion Total Mercury Concentrations for different
combinations of R1 Up and R2Up. Default/Baseline case is R1 Up = 0.1, R2Up = 0.1.
-------
- 1
o
II
Q.
E 0.5
C
1.5
_ 1
n
Q.
DC 0.5
C
1.5
CM -]
n
Q.
01 0.5
C
1.5
LO
E
0.5
n
4
3
2
jf
^ 1
* ^'* *
* * *
ปi*i^ซ* v o
) 0.2 0.4 0.6 C
4
3
2
** *-'"'*'*
*fr*f
-** *ฐ " n
) 0.2 0.4 0.6 C
4
3
2
^"*
ff
m *^"" -i
* ** 1
.* ' V-f'if
--*"' ' o
) 0.2 0.4 0.6 C
4
3
'. . ' 2
* * -''*"
,*-"'"'* , n
6
4
t 2
i . _____
) 0.2 0.4 0.6 (
6
4
. 2
1 *
j*-^*V**""* . .. n
) 0.2 0.4 0.6 (
6
4
'. . " 2
*. " _---
j*-*^"-fr . * r>
) 0.2 0.4 0.6 (
6
4
mm
'. ' 2
% t < --
.-. .',.'. , '
) 0.2 0.4 0.6
-. v;; , ' -
) 0.2 0.4 0.6
9
9 m
ซ. " 7" . *
) 0.2 0.4 0.6
'
* *
0 0.2 0.4 0.6
R2Up = 0.1
0 0.2 0.4 0.6
R2Up = 1
0 0.2 0.4 0.6
R2Up = 2
Figure 6.32. Predicted (y-axis) versus Observed (x-axis) Fish Tissue Mercury Concentrations for different
combinations of R1 Up and R2Up. Default/Baseline case is R1 Up = 0.1, R2Up = 0.1.
-------
0.6
0.5
5 0.4
ii
a. 0.3
2
DC 0.2
0.1
0
1
X
X
X
0.8
/
0.6
X
/ ' 0.4
/,
/ .ป- 0.2
S /
2
1.5
^^
' ' /'' 1
* 1 s''
'" H ,''
~':-^f* . ฐ-5
^''itt, ;!.ป .
'
.
. "
.
.'*'. *
/..;. . --''
**'\* '!ป'*-'''''' " '
*'-*/r*'**~'. *'
* ' ' ' 0 \j
0 0.2 0.4 0.6 0 0.2 0.4 0.6 0 0.2 0.4 0.6
0.6
0.5
^ 0.4
ii
3- 0.3
Z)
>
^ 0.2
0.1
0
1
/
/ 0.8
/
0.6
XX
/ 4
X 0.4
/ฃ-'/ ...... ฐ-2
/^ * ^ ฃฃ*vซซk ป * rป
2
.
1.5
.
1
X
* ^
*m -m ''
~:-^S*' '. O-5
V'O ".'"' *'
j. ** * * * ^% ' **^ * . r\
,
.
. *
s
* A:-' -:-"::'r -
_ ***'.* **t Vป*%*ป %
' ' ' U u
0 0.2 0.4 0.6 0 0.2 0.4 0.6 0 0.2 0.4 0.6
0.6
0.5
CM 0.4
ii
3- 0.3
Z)
ฐ= 0.2
0.1
0
1
X
X
X
0.8
/
/
/
0.6
X
X
>
. . ,' 0.4
/.
X' ซ * / . O-2
/x ."."** * '* ' *
X !ซ *Ui*'?ii*". j n
2
e
1.5
^
' ' x-'" 1
X
"'rl^-/?' '. . O-5
*** A! *"* "* ฐ
,xX**.* t **** - . n
I
.
.* .
. t*'
'!.:;.;. ^-;'
*"**-^'*ป'"" *' *
^^*v^.**ซ, %.^*. ป
' ' ' U u
0 0.2 0.4 0.6 0 0.2 0.4 0.6 0 0.2 0.4 0.6
0.6
0.5
10 0.4
ii
3r 0.3
Z)
E 0.2
0.1
0
/
1
X
X
X
0.8
X
X
X
0.6
X
X
.
..x ฐ-4
>
/:?;/ ฐ-2
x' .'^ajetjfc" "* " " *
/^ *S. * ra* ^TMA . ^ * rt
2
m
1.5
^ xx
' /'' 1
. . t . /''
"*ฃ**' '. O-5
x't"' **3 ***". * *
,s * . *% ^ % * m . r\
I
.
.
tf ' *
'! .*. *ฐ *. i.-''"'*"*"
.:*"vlป"ป"'"7 * '
l."-**-jSj"ป'""
~'~" *.** t ^.^** ซ
-T r (J y
0 0.2 0.4 0.6 0 0.2 0.4 0.6 0 0.2 0.4 0.6
R2Up = 0.1 R2Up=1 R2Up = 2
Figure 6.33. Predicted (y-axis) versus Observed (x-axis) Sediment Total Mercury Concentrations for different
combinations of R1 Up and R2Up. Default/Baseline case is R1 Up = 0.1, R2Up = 0.1.
-------
Epilimnion MeHg
Epilimnion HgT
60
0 0
ฃ 8000
R1Up
0 0
R1Up
Hypolimnion MeHg
Hypolimnion HgT
300
R2Up
0 0
5000
ฃ4500
re
FMUp
R2Up
0 0
R1Up
Fish Tissue Hg
Sediment HgT
150
0 0
15
R1Up
R2Up
0 0
R1Up
Figure 6.34. Surface interpolation plots of the sum of squares values for all combinations of the
R1Up and R2Up values explored in Figures 6.24 to 6.29.
-------
Epilimnion MeHg
Epilimnion HgT
0 0
R2L)P R1Up
Hypolimnion MeHg
R2Up
0 0
R1Up
Hypolimnion HgT
2.5
R2Up
0 0
R1Up
R2Up
0 0
R1Up
Fish Tissue Hg
R2Up
0 0
R1Up
Sediment HgT
0 0
R2Up R1Up
Figure 6.35. Surface interpolation plots of the estimated standad deviations for all combinations of the
R1Up and R2Up values explored in Figures 6.24 to 6.29.
------- |