United States
Environmental Protection
Agency
Results of the Lake Michigan
Mass Balance Project:
Polychlorinated Biphenyls
Modeling Report
-------
EPA-600/R-04/167
December 2006
Results of the Lake Michigan Mass
Balance Project:
Polychlorinated Biphenyls
Modeling Report
Prepared for
U.S. Environmental Protection Agency
Great Lakes National Program Office
77 West Jackson Boulevard
Chicago, Illinois 60604
Prepared by
U.S. Environmental Protection Agency
Office of Research and Development
National Health and Environmental Effects Research Laboratory
Mid-Continent Ecology Division
Large Lakes and Rivers Forecasting Research Branch
Large Lakes Research Station
9311 Groh Road
Grosse lie, Michigan 48138
Ronald Rossmann, Editor
/T~y Recycled/Recyclable
Printed with vegetable-based ink on
paper that contains a minimum ot
50% post-consumer fiber content
processed chlorine Iree
-------
Notice
The information in this document has been obtained primarily through funding by the United States
Environmental Protection Agency (USEPA) under the auspices of the Office of Research and Development
(ORD) and by the Great Lakes National Program Office (GLNPO). The report has been subjected to the
Agency's peer and administrative review and it has been approved for publication as a USEPA document.
Mention of trade names or commercial products does not constitute endorsement or recommendation for use.
-------
Foreword
The Lake Michigan Mass Balance Project (LMMBP) was initiated by the United States Environmental
Protection Agency (USEPA), Great Lakes National Program Office (GLNPO) to determine strategies for
managing and remediating toxic chemicals in the lake basin. Within the ecosystem approach, the mass
balance framework is considered the best means of accomplishing this objective, and GLNPO requested the
assistance of the USEPA Office of Research and Development (ORD) to facilitate and produce mathematical
models that account for the sources, sinks, transport, fate, and food chain bioaccumulation of certain
chemicals. This approach has been used in the past and builds upon the modeling efforts that have occurred
in the Assessment and Remediation of Contaminated Sediments (ARCS) Program and the lower Fox
River/Green Bay Mass Balance Project. The feasibility of such studies and resultant alternative management
options for contaminants in large rivers and a large embayment were demonstrated, and a logical extension
to the entire Lake Michigan receiving water body and major tributaries was warranted. There were a large
number of cooperators in this project, and by focusing Federal, State, local, private, and academic efforts and
resources on a common goal, much more was accomplished than if these entities acted independently.
The project was conducted in conjunction with the Enhanced Monitoring Program and the approach required
that all monitoring and field research be coordinated and common methodologies used. Mathematical
modelers were consulted during planning for sample design, parameters, and temporal and spatial sampling
considerations. The product was then a consistent and reliable database of information that was accessible
by project participants and the public. Data for the LMMBP were collected primarily during 1994 and 1995 and
have been compiled according to specified quality assurance/quality control (QA/QC) requirements, and other
data assessments have been made for modeling purposes.
The need to consider the environmental benefits and consequences of alternative remediation choices to
protect and improve our environment continues to intensify as: 1) environmental problems become more
complex; 2) the means to address and investigate problems become more technical, time-consuming, and
expensive; and 3) the actual costs to implement action strategies has escalated. The integrated PCBs mass
balance modeling results are presented in this document and can aid managers in establishing priorities for
both lake-wide and local improvements. The forecasting of PCB concentrations in top predator fish is one of
the primary endpoints of this investigation as it relates to both ecosystem and human health. The capability
of forecast modeling presented here is a salient feature of this approach directed toward providing multiple
alternatives, which then can be examined through benefit-cost analyses.
This report presents the current status and results of the PCB modeling effort through the summer of 2006.
Within this document some recommendations have been provided for potential future work with the models.
Of course, a model and modeling applications are never complete, and it is expected that further efforts will
change some results, insights, and our understanding of Lake Michigan. These efforts require an investment
of resources and time, and improvements with additional model run executions are measured in years. In the
larger picture, the need for Agency modeling technologies continues to intensify and the requirement for
-------
reduced uncertainty will lead to future improved generations of models. We have put great ernp' .
following guidance provided by the USEPA and other agencies in assuring that the scientlTIC *
implemented accurately and completely by model computer code and that best modeling practices nave ueen
instituted. We also submitted this to scientific peer review using an interdisciplinary panel of scientists and
experts that reviewed model theory and application which evolve on a continuing basis. The purpose is to
ensure that decisions based on the modeling efforts are reliable and scientifically credible.
This document is not intended to include all of the details and background required to understand the entire
LMMBP. Rather the reader should refer to the LMMBP Work Plan and other materials on the GLNPO web
site and the Lake Michigan Mass Balance Modeling Quality Assurance Plan on the ORD-Grosse lie web site
for further information.
This document includes replies to peer reviewer comments made during a peer review conducted 27-28 July
2004 in Romulus, Michigan. These replies and the original peer review comments are found in Part 7.
IV
-------
Abstract
The Lake Michigan Mass Balance Project (LMMBP) was conducted to measure and model nutrients, atrazine,
polychlorinated biphenyls (PCBs), frans-nonachlor, and mercury to gain a better understanding of the sources,
sinks, transport, fate, and effects of these substances within the system and to aid managers in the
environmental decision-making process for the Lake Michigan Basin. The United States Environmental
Protection Agency (USEPA) Office of Research and Development (ORD) was requested to conduct and
facilitate modeling in cooperation with the USEPA Region V; the USEPA Great Lakes National Program Office
(GLNPO); other Federal agencies; the States of Michigan, Wisconsin, Illinois, and Indiana; the Tribes; and the
public and private sectors. The effort was supported by intensive sampling of the atmosphere, major
tributaries, sediments, water column, and biota during the 1994-1995 field years as well as by extensive quality
assurance and database development. Multimedia, mass balance modeling frameworks were applied to
examine primary source and loss categories and make various model forecasts for a variety of loading
scenarios. This report focuses on the modeling practices applied and results for PCBs from the MICHTOX
screening-level model and the higher-resolution LM2-Toxic and LM Food Chain models. A unique aspect of
this work is the modeling of PCBs on a congener-level basis to make predictions of total PCBs in the system.
Results of the system mass balance show that the greatest, external gross input of PCBs to the system is
atmospheric vapor phase absorption followed by tributary inputs and atmospheric deposition, respectively.
The greatest gross losses from the system are volatilization and deep burial in sediments. Internal PCBs
loading from sediment resuspension is substantial. Gross PCBs inputs to, losses from, and cycling processes
within the system each typically exceed 1000 kg/year. Tributary inputs and atmospheric deposition are
approximately 381 and 980 kg/year, respectively. Results indicate that during the mass balance field collection
years of 1994-1995, the Fox, Grand, Calumet, and Kalamazoo Rivers had the largest tributary loads of PCBs
to Lake Michigan. When all gross input and output fluxes are summed, the system exhibits a net loss of
approximately 3,229 kg/year of PCBs. The mass balance results demonstrate the importance of contaminant
cycling and the dynamic interactions among air, water, and sediments. These interactions, with present PCS
inventories already in the lake, will continue to control PCS concentrations in the system.
LM Food Chain, linked to LM2-Toxic, and MICHTOX were used to forecast future concentrations of PCBs in
lake trout at two sites for various loading scenarios. Scenarios included constant 1994-1995 conditions, fast
continued recovery with an atmospheric load half-life of 6.0 years, slow continued recovery with an
atmospheric load half-life of 20.0 years, and various combinations of reduced atmospheric and tributary
loadings. Forecasts indicate that PCBs concentrations in lake trout will continue to decrease. For the fast
continued recovery scenario, the target level for the unrestricted consumption of fish (0.075 ppm) was
forecasted to be achieved for five to six year-old lake trout between the years 2030 and 2036. The narrow
forecast range for scenarios, past actions, the long-term decrease in loads, and decreasing PCB
concentrations in the system indicate that PCBs are presently controlled by dynamic interactions among
media, as well as air and sediment cycling. Model results from the present two models are compared. In the
future, these results will be compared to those from a greater-resolution model under development (LM3-
-------
Toxic). It is anticipated that the higher-resolution model will better delineate the nearshore and sediment
zones, define lake interactions with tributary inputs, and describe PCBs in lake trout populations.
This synthetic lake-wide perspective is anticipated to aid managers in moving forward on pollution prevention,
remedial actions, and legislative priorities associated with the Lake Michigan Lake-wide Management Plans.
It will also help describe expected local improvements associated with Remedial Action Plans in Areas of
Concern. This abstract does not necessarily reflect USEPA policy.
VI
-------
Table of Contents
Notice ii
Foreword jjj
Abstract v
Table of Contents vii
List of Figures xviii
List of Tables xxx
Abbreviations xxxvi
Acknowledgments xxxviii
Executive Summary xxxix
Part 1 Introduction 1
Chapter 1 Project Overview 1
1.1.1 Background 1
1.1.2 Description 2
1.1.3 Scope 3
1.1.3.1 Modeled Pollutants 3
1.1.3.1.1 PCBs 3
1.1.3.1.2 Isomer frans-Nonachlor 5
1.1.3.1.3 Atrazine 5
1.1.3.1.4 Mercury 5
1.1.3.2 Other Measured Parameters 6
1.1.3.3 Measured Compartments 6
1.1.4 Objectives 6
1.1.5 Design 7
1.1.5.1 Organization 7
1.1.5.2 Study Participants 9
1.1.5.3 Workgroups 9
1.1.5.4 Information Management 9
1.1.5.4.1 Data Reporting 10
1.1.5.4.2 Great Lakes Environmental Monitoring Database 11
1.1.5.4.3 Public Access to LMMBP Data 11
1.1.5.5 Quality Assurance Program 11
1.1.6 Project Documents and Products 13
VII
-------
Chapter 2 PCBs Modeling Overview
1 .2. 1 Background [[[
1 .2.2 Modeling Objectives .................................................. '«
1 .2.3 Historical Modeling [[[ 1?
1.2.3.1 Lake-1 [[[ 17
1 .2.3.2 Completely-Mixed Model ....................................... 18
1 .2.3.3 General Mass Balance Framework for Toxic Chemicals in the
Great Lakes ................................................. 18
1 .2.3.4 Food Web Bioaccumulation Model ............................... 18
1 .2.3.5 MICHTOX .................................................. 18
1 .2.3.6 Green Bay Mass Balance Project ................................ 19
1 .2.3.7 SEDZL [[[ 19
1 .2.4 Model Resolution [[[ 19
1 .2.5 Models Developed and Applied ......................................... 21
1 .2.5.1 Lake Process Models ......................................... 21
1 .2.5.2 Hydrodynamics (POM) ........................................ 21
1 .2.5.3 Eutrophication/Sorbent Dynamics (LM3-Eutro) ...................... 22
1 .2.5.4 Contaminant Transport and Fate (LM2-Toxic) ....................... 22
1 .2.5.5 Food Web Bioaccumulation (LM Food Chain) ....................... 23
1 .2.6 Model Quality Assurance .............................................. 23
1 .2.7 Model Application and Computational Aspects ............................. 23
1 .2.7. 1 Annual Simulations ........................................... 23
1 .2.7.2 Long-Term Simulations ........................................ 23
Chapter 3 Information Management ........................................... 26
1 .3. 1 Overview of Information Management at the LLRS .......................... 26
1 .3.2 Calculation of Total PCBs ............................................. 28
1 .3.3 Regression Analysis of Measured Congener, Total PCB Data .................. 28
1 .3.4 Summary [[[ 30
Appendix 1 .3.1 List of Parameters Analyzed and Principal Investigators for
the LMMBP ................................................. 31
Appendix 1 .3.2 Example of Data Verification Checklist Used for the LMMBP ............ 35
Appendix 1 .3.3 Printout of Information Stored in the LMMBP Tracking Database
(R:/Access2000/lmmb/lmtrack.mdb) .............................. 39
Appendix 1 .3.4 Generalized Format for the LMMBP Water Data to be Analyzed
With IDL Programs ........................................... 44
Appendix 1 .3.5 Generalized Format for the LMMBP Sediment Data to be Analyzed
With IDL Programs ........................................... 46
Appendix 1 .3.6 Generalized Format for the LMMBP Fish Data to be Analyzed With
IDL Programs ............................................... 48
-------
Chapter 4 Representativeness of the LMMBP Years Relative to Lake Michigan's
Historic Record 55
1.4.1 Introduction 55
1.4.2 Ice Cover 55
1.4.3 Water and Air Temperatures 56
1.4.4 Lake Water Levels 59
1.4.5 Precipitation 59
1.4.5.1 Annual Comparisons 59
1.4.5.2 Monthly Comparisons 59
1.4.6 Tributary Flows 59
1.4.7 Wave Heights 59
1.4.8 Summary 63
Chapter 5 PCBs in the Lake Michigan Ecosystem 66
1.5.1 Introduction 66
1.5.2 Atmospheric 66
1.5.2.1 Vapor Phase 66
1.5.2.2 Precipitation 69
1.5.2.3 Paniculate 71
1.5.2.4 Dry Deposition 71
1.5.3 Lake Water 71
1.5.3.1 Total PCBs 71
1.5.3.2 Dissolved PCBs 71
1.5.3.3 Paniculate PCBs 78
1.5.4 Tributaries 78
1.5.5 Sediment 83
1.5.6 Biota 86
1.5.7 Summary 86
Chapter 6 Congener Pattern Matching of Data Collected for the Lake Michigan
Mass Balance Project (LMMBP) 93
1.6.1 Introduction 93
1.6.2 Analytical Approach 94
1.6.3 Methodology 94
1.6.4 Results 94
1.6.4.1 Comparison of Modeled Congener Patterns to All Analyzed Congener
Patterns 94
1.6.4.2 Comparison of Median to Mean Data 97
1.6.4.3 Comparison of Congener Patterns in Different Media in Segment 21
Saugatuck Biota Box 97
1.6.4.4 Comparison of Atmospheric Congener Data 101
1.6.4.5 Comparison of Tributary Congener Patterns 103
1.6.4.6 Comparison of Ages 5 and 6 Lake Trout Congener Patterns in All
Biota Boxes 105
1.6.4.7 Comparison of Different Lake Trout Age Class Congener Patterns
in Saugatuck Biota Box 105
1.6.5 Conclusions 105
IX
-------
Chapter 7 Hindcasting and Forecasting Functions for PCBs in the Lake Michigan
Ecosystem "
1.7.1 Introduction 11°
1.7.2 Forecast Functions 110
1.7.2.1 Tributary Loads 111
1.7.2.2 Atmospheric Loads 112
1.7.3 Hindcast Functions 112
1.7.4 Estimated PCB Storage 117
Part 2 LM2-Eutro 120
Chapter 1 Conclusions (Executive Summary) 120
Chapter 2 Recommendations 123
Chapter 3 Model Description 125
2.3.1 Transport Scheme for Lake Michigan 125
2.3.2 Sediments 126
2.3.3 Formulation of Eutrophication Equations 126
Appendix 2.3.1 Development of LMS-Eutro Equations 130
A2.3.1.1 Phytoplankton Growth 130
A2.3.1.2 Zooplankton Kinetics 133
A2.3.1.3 Carbon Interactions 133
A2.3.1.4 Phosphorus 135
A2.3.1.5 Nitrogen 137
A2.3.1.6 Silica 138
Chapter 4 Model Input and Field Data 140
2.4.1 Loading and Sediment-Water Interactions 140
2.4.1.1 Atmospheric Loads 140
2.4.1.2 Tributary Loads 140
2.4.1.3 Shoreline Erosion 141
2.4.1.4 Sediment 142
2.4.2 Field Data 142
2.4.2.1 Open Lake Nutrient and Carbon Data 142
2.4.2.1.1 Total Phosphorus 142
2.4.2.1.2 Dissolved Phosphorus 144
2.4.2.1.3 Soluble Reactive Phosphorus 144
2.4.2.1.4 Nitrate 145
2.4.2.1.5 Ammonia 145
2.4.2.1.6 Total Kjeldahl Nitrogen 145
2.4.2.1.7 Dissolved Silica 145
2.4.2.1.8 Dissolved Organic Carbon 145
-------
2.4.2.1.9 Particulate Organic Carbon 145
2.4.2.1.10 Green Bay Nutrient Data 145
2.4.2.2 Plankton 146
2.4.2.2.1 Phytoplankton 146
2.4.2.2.2 Chlorophyll a 146
2.4.2.2.3 Phytoplankton Carbon 147
2.4.2.2.4 Zooplankton 149
2.4.2.2.5 Zooplankton Carbon 150
2.4.3 Initial Conditions 150
2.4.4 Parameter Estimation 152
2.4.4.1 Physical Measurements 152
2.4.4.1.1 Secchi Disk 152
2.4.4.1.2 Solar Radiation and Temperature 152
2.4.4.2 Primary Production Estimates 152
Appendix 2.4.1 Modeled Versus Measured Variables 157
Chapter 5 Calibration 159
2.5.1 Description of Process 159
2.5.2 Selection of Best Calibration 160
2.5.2.1 Phytoplankton 160
2.5.2.2 Particulate Organic Carbon 167
2.5.2.3 Total Phosphorus 167
2.5.2.4 Dissolved Silica 167
Chapter 6 Model Confirmation 168
2.6.1 Additional Field Data 168
2.6.2 MICH1 Model 168
2.6.3 Comparison of LM3-Eutro to the MICH1 Model and Field Data 168
Chapter 7 Results - Application of Model 171
2.7.1 Scenario 1 - Constant Conditions 171
2.7.1.1 Description of Assumptions 171
2.7.1.2 Results and Discussion 171
2.7.2 Scenario 2 - Virtual Elimination (Lower Bound) 171
2.7.2.1 Description of Assumptions 171
2.7.2.2 Results and Discussion 171
2.73 Scenario 3 - Best Estimate of Current Trends Resulting From Previous
Actions 173
2.7.3.1 Description of Assumptions 173
2.7.3.2 Results and Discussion 173
2.7.4 Scenario 4 - Scenario 1 With Instantaneous Reduction of Tributary Loads to
Zero 173
2.7.4.1 Description of Assumptions 173
2.7.4.2 Results and Discussion 174
XI
-------
275 Scenario 5 - Scenario 1 With Instantaneous Reduction of Atmospheric Loads
to Zero }74
2.7.5.1 Description of Assumptions ' '4
2.7.5.2 Results and Discussion 174
2.7.6 Scenario 6 - Scenario 1 With Tributary and Atmospheric Loads Increased 20% .... 174
2.7.6.1 Description of Assumptions 174
2.7.6.2 Results and Discussion 176
2.7.7 Scenario 7 - Application of Great Lakes Water Quality Agreement Loads to
Model 176
2.7.7.1 Description of Assumptions 176
2.7.7.2 Results and Discussion 177
2.7.8 Scenario 8 - Estimate of Total Maximum Daily Loads to Reach International
Joint Commission's Target Total Phosphorus Concentration 177
2.7.8.1 Description of Assumptions 177
2.7.8.2 Results and Discussion 177
2.7.9 Scenario Comparison and Discussion 177
2.7.10 Mass Budget 177
Chapter 8 Results Provided for LM2-Toxic 182
2.8.1 Description 182
2.8.2 Manipulation of Results 182
Part 3 Level 1 Models 183
Chapter 1 MICHTOX PCB Model Executive Summary 183
Chapter 2 MICHTOX Recommendations 185
Chapter 3 MICHTOX PCB Fate and Transport Modeling 186
3.3.1 Description 186
3.3.2 Description of Data Used in MICHTOX 189
3.3.2.1 Water Column PCB Concentrations 189
3.3.2.2 Surficial Sediment PCB Concentrations 189
3.3.2.3 Atmospheric and Tributary Loads 189
3.3.3 Model Confirmation 191
3.3.3.1 Description of Hindcast Process 191
3.3.3.2 Hindcast Results 192
3.3.3.3 Comparison to the LMMBP Data 196
3.3.4 Model Uncertainty 198
Chapter 4 MICHTOX Food Chain Modeling 201
3.4.1 Model Development 201
3.4.2 Description of the Data Used in MICHTOX Food Chain 202
3.4.2.1 Description of Data 202
3.4.2.2 Sources and Choice of Constants '.'.'.'.'. 203
XII
-------
3.4.3 Model Confirmation 203
3.4.4 Results - Forecast Scenarios 207
3.4.4.1 Conditions Remain the Same as 1994-1995 (Constant Conditions) .. .. 207
3.4.4.2 Continued Recovery - Fast 208
3.4.4.3 Continued Recovery - Slow 209
3.4.5 Model Sensitivity 209
3.4.5.1 No Atmospheric Wet and Dry Deposition Loadings 209
3.4.5.2 No Tributary Loadings 210
3.4.5.3 No Atmospheric Deposition and No Tributary Loadings 210
3.4.5.4 Sediment Total PCB Concentration Initial Conditions Set to Zero 210
Appendix 3.4.1 Derivation of a Hypothetical Lake Michigan Lake Trout Fish
Consumption Criteria for PCBs 212
Part 4 LM2-Toxic 216
Chapter 1 Executive Summary 216
Chapter 2 Recommendations 221
Chapter 3 Model Description 223
4.3.1 Model Framework 223
4.3.2 Model Configuration 224
4.3.2.1 Spatial Resolution - Segmentation 224
4.3.2.2 Temporal Resolution 228
4.3.3 Water Balance 228
4.3.4 Solid Balance 231
4.3.4.1 Solid Kinetics 232
4.3.4.2 Sediment Transport 233
4.3.4.2.1 Steady-State Resuspension Calibration 234
4.3.4.2.2 Empirical Wave-Induced Resuspension Calculation 235
4.3.4.2.3 The Sediment Bed - Semi-Lagrangian Option 237
4.3.5 Chemical Balance 238
4.3.5.1 PCB Partitioning 239
4.3.5.2 PCB Air-Water Exchange 240
4.3.5.3 PCB-Specific Parameterization 242
4.3.6 Modification 242
Appendix 4.3.1 Lake Michigan Resuspension Field Data Set 246
Appendix 4.3.2 Notes From Nathan Hawley on the Data Set in Appendix 4.3.1 247
Chapter 4 Model and Field Data 248
4.4.1 Water Transport 248
4.4.1.1 Circulation 248
4.4.1.2 Vertical Dispersion 250
4.4.1.3 Verification of Water Transport Fields 250
XIII
-------
4.4.2 Organic Carbon 257
4.4.2.1 Loads 257
4.4.2.2 Field Data and Initial Conditions 258
4.4.2.3 Parameterization 266
4.4.3 PCBs 271
4.4.3.1 Loading 272
4.4.3.2 Field Data, Initial Conditions, and Boundary Conditions 273
4.4.3.3 Parameterization 273
4.4.3.4 Kinetic Time Functions 281
Appendix 4.4.1 Sample Data Interpolation for the LMMBP 286
A.4.4.1.1 The Distance Square Inverse Method 287
A.4.4.1.2 The Natural-Neighborhood Method 287
A.4.4.1.3 Application 288
A4.4.1.3.1 Contouring Plots 288
A.4.4.1.3.2 Volume-Weighted Averaging With Formulations 288
A.4.4.1.4 Discussion 289
A.4.4.1.5 Steps to Run nngridr 289
Chapter 5 LM-2 Toxic Calibration and Confirmation 291
4.5.1 Vertical Dispersion Coefficients Calibration 291
4.5.2 Organic Carbon Dynamics Calibration 293
4.5.2.1 Calibration Process/Procedure 293
4.5.2.2 Results and Discussion 293
4.5.3 PCB Dynamics Calibration 294
4.5.3.1 Calibration Procedures 304
4.5.3.2 Results and Discussion 305
4.5.4 The LM2-Toxic Confirmation 313
4.5.4.1 Mass Balance Checking 313
4.5.4.2 Chloride Model 313
4.5.4.3 137Cs and 239>24°pu Simulation and Results 313
4.5.4.4 Long-Term Organic Carbon Simulations 317
4.5.4.5 PCB Hindcast 319
4.5.4.5.1 Data and Procedure for the PCB Hindcast 319
4.5.4.5.2 Results From the LM2-Toxic PCB Hindcast 328
Appendix 4.5.1 Results From Thermal Balance Model 337
Appendix 4.5.2 Calibrated Results for Organic Carbons 349
Appendix 4.5.3 Calibration Results for PCB28+3, and ZPCBs 365
Appendix 4.5.4 Simulation Results From Chloride 418
Appendix 4.5.5 Primary Production for the LM2-Toxic 426
Chapter 6 The LM2-Toxic Application and Interpretation 430
4.6.1 Conversion of PCB Congener Results to Total PCBs 430
4.6.2 Mass Budget Diagnosis of the LM2-Toxic for the LMMBP Period 430
XIV
-------
4.6.3 LM2-Toxic Application for Long-Term Forecast and Sensitivity Scenarios 436
4.6.4 Results of the Forecast and Sensitivity Scenarios and Discussion . 437
4.6.5 Results Provided for the LM Food Chain Model 442
Chapter 7 LM2-Toxic Sensitivity Analysis 444
4.7.1 Primary Production Sensitivity 444
4.7.2 PCB Loads Sensitivity 450
Part 5 LM Food Chain 453
Chapter 1 Executive Summary 453
Chapter 2 Recommendations 455
5.2.1 Additional Model Validations 455
5.2.2 Model Applications 455
5.2.3 Future Improvements 456
Chapter 3 Model Description 457
5.3.1 Chemical Bioaccumulation in Fish 457
5.3.1.1 Chemical Uptake From Water 458
5.3.1.2 Chemical Uptake From Prey 459
5.3.1.3 Chemical Elimination Via Gills 459
5.3.1.4 Chemical Dilution by Growth 460
5.3.2 Chemical Bioaccumulation in the Base of Food Webs 460
5.3.2.1 Chemical Bioaccumulation in Zooplankton 460
5.3.2.2 Chemical Bioaccumulation in Diporeia 461
5.3.3 Model Description of Exposure Environment 462
Chapter 4 Description of Data, Constants, and Other Information Necessary
to Run Model 465
5.4.1 Chemical Properties of PCB Contaminants 465
5.4.2 Site-Specific Data 465
5.4.2.1 Fish Food Web Structures 465
5.4.2.1.1 Lake Trout Food Web 465
5.4.2.1.2 Coho Salmon Food Web 468
5.4.2.2 Fish Growth Rates 475
5.4.2.3 Energy Density of Food Web Components 476
5.4.2.4 Exposure Conditions 476
5.4.2.4.1 PCB Concentrations in Water 481
5.4.2.4.2 PCB Concentrations in Sediment 481
5.4.2.4.3 Exposure Temperature 483
5.4.2.4.4 Oxygen Concentration in Water 483
xv
-------
5.4.3 Physiological Data of Fish and Other Organisms 483
5.4.3.1 Species-Specific Respiration Rates 483
5.4.3.2 Respiration Rates Adjusted for Specific Dynamic Action (SDA) 486
5.4.4 Calibrated Model Parameters 487
Chapter 5 Calibration 492
5.5.1 Introduction 492
5.5.2 Description of Process 492
5.5.3 Calibration Results 493
5.5.4 Field Data for PCBs in Fish and Their Comparisons to Calibrated Model
Outputs 495
Appendix 5.5.1 PCB Concentrations 502
Appendix 5.5.2 Agreement Between Modeled and Observed PCB Concentrations 516
Chapter 6 Model Verification 526
5.6.1 Introduction 526
5.6.2 Model Applicability to Other Sites 526
Chapter 7 Model Sensitivity and Uncertainty 528
5.7.1 Introduction 528
5.7.2 Sensitivity Analysis 528
5.7.2.1 Chemical Assimilation Efficiency (a) 529
5.7.2.2 Food Assimilation Efficiency (P) 529
5.7.2.3 Chemical Relative Gill Transfer Coefficient (EJE0) 531
5.7.2.4 The Fraction of Ingested Energy for Specific Dynamic Action (SDA) 531
5.7.2.5 Fish Growth Rate 532
5.7.2.6 Octanol-Water Partition Coefficient Kow 533
5.7.2.7 Fish Diet 533
Chapter 8 Model Application 536
5.8.1 Introduction 536
5.8.2 Simulation of Fish PCB Levels Based on Hypothetical Exposure Inputs 536
5.8.2.1 Exposure Concentration Inputs Used for Model Simulations 536
5.8.2.2 Responses of Fish Models to Different Exposure Inputs 537
5.8.2.3 Discussion 542
Part 6 Comparison of Model Results 544
6.1 Summary 544
6.2 Comparison of Models 545
6.2.1 Model Similarities 545
6.2.2 Model Differences 545
6.3 Comparison of Model Results 547
6.3.1 Comparison of Mass Budget Analyses 547
6.3.2 Comparison of Model Forecast Scenarios 547
XVI
-------
Part 7 Appendices 552
Appendix 7.1 Lake Michigan Mass Balance Project (LMMBP) PCB Peer
Review Report 552
7.1.1 Executive Summary 552
7.1.2 LMMBP Peer Review Panel 553
7.1.3 LMMBP PCB Charge to Peer Reviewers 553
7.1.3.1 Overall Multimedia Ecosystem Modeling Approach 555
7.1.3.2 Overall Model Performance 555
7.1.3.3 Suitability for Management 555
7.1.4 Modelers' Responses to Peer Review Comments 555
7.1.5 Modelers' Responses to Specific Comments Made by Peer Review
Panel Member - James Martin 563
Appendix 7.2 Comments as Received From Dr. James Martin Peer Review
Summary: Lake Michigan Mass Balance Project 573
7.2.1 General Comments 573
7.2.1.1 Overall Multi-Media Ecosystem Modeling Approach 573
7.2.1.2 Overall Model Performance 573
7.2.1.3 Suitability for Management 574
7.2.2 Specific Recommendations 574
7.2.2.1 POM and Linkages 574
7.2.2.2 LM2-Eutro and LM3-Eutro 574
7.2.2.3 Level 1 Model 575
7.2.2.4 LM2-Toxic 575
7.2.2.5 LM Food Chain 575
7.2.2.6 LM3-Toxic 575
7.2.3 Specific Comments 575
7.2.3.1 Hydrodynamics and POM Linkage 575
7.2.3.2 LM2-Eutro and LM3-Eutro 577
7.2.3.3 Level 1 Model 577
7.2.3.4 LM2-Toxic 578
7.2.3.5 LM Food Chain 579
XVII
-------
List of Figures
1.1.1 Simplified mass balance approach 2
1.1.2 The LMMBP sampling locations 8
1.1.3 Flow of information in the LMMBP ig
1.2.1 Surface water segmentation for alternative Lake Michigan mass balance model levels 20
1.2.2 Model construct used for the LMMBP to model PCBs 22
1.4.1 Location of the NOAA's buoys in Lake Michigan 55
1.4.2 Monthly mean water temperatures in southern Lake Michigan 58
1.4.3 Monthly mean water temperatures in northern Lake Michigan 58
1.4.4 Mean June water temperatures in southern Lake Michigan 58
1.4.5 Mean June water temperatures in northern Lake Michigan 58
1.4.6 Monthly mean air temperatures in southern Lake Michigan 60
1.4.7 Monthly mean air temperatures in northern Lake Michigan 60
1.4.8 Mean June air temperatures in southern Lake Michigan 60
1.4.9 Mean June air temperatures in northern Lake Michigan 60
1.4.10 Record of mean monthly water levels for Lake Michigan 60
1.4.11 Annual precipitation to Lake Michigan between 1949 and 1998 60
1.4.12 Comparison of 1982, 1983, 1994, and 1995 monthly mean precipitation to the means
for the period of 1949 through 1998 61
1.4.13 Comparison of tributary flow for hydrodynamic model calibration (1982-1983) to the
historic means 61
XVIII
-------
1.4.14 Comparison of tributary flow for the study period (1994-1995) to the historic means 62
1.5.1 Median concentration of vapor phase PCBs in the atmosphere during 1994 and 1995
for all seasons of both years 68
1.5.2 Time variation of vapor phase PCBs in Lake Michigan 70
1.5.3 Time variation of vapor phase PCBs in Lake Michigan at Sleeping Bear Dunes based
on IADN data 70
1.5.4 Median concentration of wet (precipitation) PCBs in the atmosphere during 1994 and
1995 for all seasons of both years 72
1.5.5 Time variation of precipitation PCBs in Lake Michigan at Sleeping Bear Dunes based
on IADN data 73
1.5.6 Median concentration of particulate PCBs in the atmosphere during 1994 and 1995 for
all seasons of both years 74
1.5.7 Time variation of atmospheric particulate PCBs in Lake Michigan at Sleeping Bear
Dunes based on IADN data 75
1.5.8 Distribution of total PCBs (ng/L) in 1994-1995 Lake Michigan water 76
1.5.9 Time variation of total PCBs in Lake Michigan water .... 77
1.5.10 Time variation of total PCBs in Lake Michigan water since 1986 78
1.5.11 Distribution of dissolved PCBs (ng/L) in 1994-1995 Lake Michigan water 79
1.5.12 Distribution of dissolved PCBs (ng/L) in 1994-1995 summer hypolimnetic Lake Michigan
water 80
1.5.13 Time variation of dissolved PCBs in Lake Michigan water 81
1.5.14 Distribution of particulate PCBs (ng/L) in 1994-1995 Lake Michigan water 81
1.5.15 Time variation of particulate PCBs in Lake Michigan water 82
1.5.16 Relative loads of PCBs to Lake Michigan from tributaries 83
1.5.17 Total PCBs in 1994-1995 Lake Michigan surficial sediments (ng/g) 85
1.5.18 Vertical variation of PCBs in dated sediment cores collected for the LMMBP 85
1.5.19 Vertical variation of PCBs in dated sediment cores reported by Swackhamer and
Armstrong (1988) 87
1.5.20 Vertical variation of PCBs in dated sediment cores reported by Hermanson ef al. (1991) . . . 87
XIX
-------
1.5.21 Vertical variation of PCBs in dated sediment cores reported by Golden et al. (1993) 88
1.5.22 Vertical variation of PCBs in dated sediment cores reported by Schneider et al. (2001) 88
1.5.23 PCB concentrations in various members of the lake trout food web during the LMMBP 89
1.5.24 PCB concentrations in various age classes of lake trout during the LMMBP 90
1.5.25 Time variation of PCB concentration in five to six year-old lake trout from Lake
Michigan 90
1.5.26 Time variation of PCB concentrations in bloater from Lake Michigan 91
1.6.1 LM2 surface water segmentation and LMMBP biota boxes 94
1.6.2 Cumulative frequency distribution - PCB congeners in segment 2 vapor phase 98
1.6.3 Cumulative frequency distribution - PCB congeners in segment 2 dry deposition 98
1.6.4 Cumulative frequency distribution - PCB congeners in segment 2 wet deposition 98
1.6.5 Cumulative frequency distribution - dissolved PCB congeners in segment 2 water 98
1.6.6 Cumulative frequency distribution - particulate PCB congeners in segment 2 water 98
1.6.7 Cumulative frequency distribution - PCB congeners in segment 2 surficial sediment 98
1.6.8 Cumulative frequency distribution - dissolved PCB congeners in Kalamazoo River water .... 99
1.6.9 Cumulative frequency distribution - particulate PCB congeners in Kalamazoo River water ... 99
1.6.10 Cumulative frequency distribution - age 5-6 Saugatuck lake trout 99
1.6.11 Cumulative frequency distribution - PCB congeners in segment 2 vapor phase 99
1.6.12 Cumulative frequency distribution - PCB congeners in segment 2 dry deposition 99
1.6.13 Cumulative frequency distribution - dissolved PCB congeners in Kalamazoo River water 99
1.6.14 Cumulative frequency distribution - particulate PCB congeners in Kalamazoo River water ... 100
1.6.15 Cumulative frequency distribution - PCB congeners in segment 2 wet deposition 100
1.6.16 Cumulative frequency distribution - dissolved PCB congeners in Saugatuck water 100
1.6.17 Cumulative frequency distribution - particulate PCB congeners in Saugatuck water 100
1.6.18 Cumulative frequency distribution - PCB congeners in surficial sediment 100
xx
-------
1.6.19 Cumulative frequency distribution - PCB congeners in age 5-6 Saugatuck lake trout 100
1.6.20 PCB congeners in segment 2, Saugatuck 101
1.6.21 Air sampling locations 102
1.6.22 Lake Michigan high-resolution 5 km x 5 km grid with 19 sigma layers . 102
1.6.23 Cumulative frequency distribution (mean) - PCB congeners in atmospheric vapor phase .... 103
1.6.24 Cumulative frequency distribution (mean) - PCB congeners in atmospheric wet
deposition 103
t.6.25 Cumulative frequency distribution (mean) - PCB congeners in atmospheric dry
deposition 103
1.6.26 Cumulative frequency distribution (mean) - PCB congeners in segment 2 atmospheric
data 103
1.6.27 PCB congeners in Lake Michigan tributaries 104
1.6.28 Comparison of dissolved PCB congeners in Lake Michigan western tributaries to
segment 2 vapor phase 105
1.6.29 Comparison of dissolved PCB congeners in Lake Michigan eastern tributaries to
segment 2 water 106
1.6.30 Comparison of particulate PCB congeners in Lake Michigan segment 2 to tributaries 106
1.6.31 Cumulative frequency distribution (mean) - PCB congeners in ages 5 and 6 lake trout 107
1.6.32 Cumulative frequency distribution (mean) - PCB congeners in Saugatuck lake trout 107
1.6.33 Comparison of dissolved PCB congeners in west side-to-east side of Lake Michigan
tributaries 108
1.6.34 Comparison of particulate PCB congeners in west side-to-east side of Lake Michigan
tributaries - 108
1.7.1 Locations of dated cores analyzed for PCBs by Hermanson et al. (1981), Swackhamer
and Armstrong (1988), Schneider era/. (2001), Golden et al. (1993), and P. Van Hoof
(personal communication) for the LMMBP 113
1.7.2 Fit of concentration functions to observed data for core 15 114
1.7.3 Lake Michigan non-depositional (0-40 m), transitional (40-100 m), and depositional
(> 100 m) zones based on water depth and the depth of wind-wave interaction with
sediments 116
XXI
-------
1.7.4 Comparison of load function to 210Pb focusing factor corrected core 15 loads 117
2.4.1 The LMMBP sampling locations 143
2.4.2 Lake-wide phytoplankton carbon calculated from biovolume data and carbon-to-
chlorophyll a ratios for the eight LMMBP cruises 148
2.4.3 Level 2 model segmentation for LM3-Eutra 151
2.4.4 Lake-wide Secchi depths for the eight LMMBP cruises 153
2.5.1 Level 2 and Level 3 model segmentation 159
2.5.2 LM3-Eutro model versus laboratory primary production 161
2.5.3 Level 3 LM3-Eutro model predictions versus field data, lake-wide 164
2.5.4 Level 2 LM3-Eutro model output versus field data for selected segments 165
2.5.5 Level 3 LM3-Eutro model output versus field data for selected nearshore and
offshore cells 166
2.6.1 MICH1 versus LM3-Eutro model predictions and available field data 170
2.7.1 Scenario 1: Constant Conditions 172
2.7.2 Scenario 2: Virtual elimination 172
2.7.3 Historical total phosphorus loading - Lake Michigan 173
2.7.4 Scenario 3: Best estimate of current trends resulting from previous actions 174
2.7.5 Scenario 4: Scenario 1 with tributary load elimination 175
2.7.6 Scenario 5: Scenario 1 with atmospheric load elimination 175
2.7.7 Scenario 6: Scenario 1 with tributary and atmospheric loads increased 20% 176
2.7.8 Scenario 7: Application of the GLWQA loads to model 178
2.7.9 Scenario 8: Estimate of the TMDL to reach the IJC's target total phosphorus
concentration 178
2.7.10 Annual average (1994-1995) Lake Michigan total phosphorus loading (kg/year) 180
2.7.11 Annual average (1994-1995) Lake Michigan and Green Bay total phosphorus
loading (kg/year) 181
3.3.1 MICHTOX PCB mass balance schematic 187
XXII
-------
3.3.2 MICHTOX model segmentation 188
3.3.3 Long-term estimates of Lake Michigan total PCB vapor concentrations 192
3.3.4 Long-term estimates of Lake Michigan total PCB atmospheric deposition loadings 192
3.3.5 Long-term estimates of Lake Michigan total PCB tributary loadings 192
3.3.6 Long-term Scenario A predictions of main lake total PCB concentrations 192
3.3.7 Comparison of long-term Scenario A predictions to main lake sediment total PCB
concentrations (sediment cores collected in 1991-1992). 193
3.3.8 Comparison of long-term Scenario A predictions to GLNPO lake trout data 193
3.3.9 Comparison of MICHTOX Scenario A total PCB concentrations to Sheboygan Reef
data 193
3.3.10 Long-term Scenario B predictions of main lake total PCB concentrations 193
3.3.11 Comparison of long-term Scenario B predictions to the LMMBP deep water dissolved
total PCB concentrations 194
3.3.12 Comparison of Scenario B predictions to main lake sediment total PCB concentrations
(sediment cores collected in 1991-1992). 194
3.3.13 Comparison of long-term Scenario B predictions to average total PCB sediment
concentrations (LMMBP and GBMBP box core samples) 194
3.3.14 Comparison of long-term Scenario B predictions to GLNPO lake trout data 194
3.3.15 Comparison of long-term Scenario B total PCB concentrations to Sheboygan Reef
fish data 195
3.3.16 Long-term Scenario C predictions of main lake total PCB concentrations 195
3.3.17 Comparison of Scenario C predictions to main lake sediment total PCB concentrations
(sediment cores collected in 1991 -1992). 196
3.3.18 Comparison of long-term Scenario C predictions to GLNPO lake trout data 196
3.3.19 Comparison of MICHTOX Scenario C total PCB concentrations to Sheboygan Reef
fish data 196
3.3.20 Comparison of MICHTOX epilimnetic total PCB concentrations to the LMMBP
cruise data 197
3.3.21 Comparison of MICHTOX hypolimnetic total PCB concentrations to the LMMBP
cruise data 198
XXIII
-------
3.3.22 MICHTOX predicted mass balance fluxes and inventories (kg/year) for 1994-1995,
whole lake results 199
3.3.23 MICHTOX predicted mass balance fluxes and inventories (kg/year) for 1994-1995,
Green Bay and main lake results 199
3.4.1 The LMMBP biota sampling zones 202
3.4.2 Total PCB concentrations of organisms in Lake Michigan biota zones 206
3.4.3 Total PCB concentrations in 5.5 year-old lake trout at Saugatuck biota zone 207
3.4.4a Sensitivity scenario predicted total PCB concentrations in 5.5 year-old lake trout
from Saugatuck biota zone : 208
3.4.4b Sensitivity scenario total PCB predictions and the fish consumption target level 208
3.4.5a Model application scenario total PCB predictions in 5.5 year-old lake trout from the
Saugatuck biota zone 208
3.4.5b Application scenario total PCB predictions and the fish consumption target level 208
3.4.6 Sensitivity scenario total PCB concentration predictions for 5.5 year-old lake trout
at Saugatuck 209
A3.4.1 Whole fish to edible portion of fish PCBs and lipid ratios for lake trout 214
A3.4.2 Comparison of whole fish to fillet PCB ratios and lipid content for various fish species 214
4.1.1 Mass budget average for 1994-1995 total PCBs Lake Michigan system (including
Green Bay) 218
4.1.2 Annual long-term responses of total PCB concentrations in the water column of Lake
Michigan for the forecast and sensitivity scenarios 219
4.3.1 Water column segmentation for the LM2-Toxic model 225
4.3.2 Surface sediment segmentation for the LM2-Toxic model 226
4.3.3 Cross-sectional sediment segmentation and overlying water column segments for 10
Lake Michigan and four Green Bay water columns 227
4.3.4 Conceptual framework of organic carbon sorbent dynamics used in the LM2-Toxic model ... 232
4.3.5 Schematic of conceptualization for the steady-state mass balance analysis for PDC
vertical transport 234
4.3.6 Locations of the 30 deployments between 1994 and 2000 236
XXIV
-------
4.3.7 Regression analysis on the data set (resuspension observed only) 237
4.3.8 Conceptual framework used by the LM2-Toxic model for PCB congeners in
Lake Michigan 238
4.4.1 Locations of 10 tributaries whose flows were considered a part of the water transport
used in the LM2-Toxic 249
4.4.2 Locations of monitored and unmonitored tributaries during the LMMBP 253
4.4.3 Lake Michigan water sampling sites during the LMMBP 254
4.4.4 Primary production generated from the LM3-Eutro for Lake Michigan, including
Green Bay 257
4.4.5 Lake Michigan sediment sampling sites of organic carbon during the LMMBP 265
4.4.6 Distribution of POC in Lake Michigan surficial sediments (mg/gdw) 265
4.4.7 Distribution of POC in Lake Michigan surficial sediments (mg/L) 266
4.4.8 IPCBs tributary (11 monitored and 18 unmonitored tributaries) loads to Lake Michigan
during the LMMBP period 272
4.4.9 Estimated IPCBs atmospheric loads including dry and wet deposition into Lake Michigan
during the LMMBP period 273
4.4.10 Lake Michigan atmospheric sampling sites during the LMMBP 275
4.4.11 Seasonal variation of IPCB vapor phase concentrations observed during the LMMBP 276
4.4.12 Lake Michigan sediment sampling sites for PCBs during the LMMBP 277
4.4.13 Distribution of IPCBs in Lake Michigan surficial sediments (ng/gdw) 278
4.4.14 Distribution of ZPCBs in Lake Michigan surficial sediments (ng/L) 278
4.4.15 Comparison between the estimated log K'POc,a f°r the LMMBP selected PCB congeners
based on the two-phase partitioning model and Kow calculated by Hawker and Connell
(1988) for all 209 PCB congeners 281
4.5.1 a Comparison between the temporal profiles for temperature from the LM2-Toxic 292
4.5.1 b Comparison between the temporal profiles for temperature results from the
Princeton Ocean Hydrodynamic model 292
4.5.2 Temporal profiles of DOC in the Lake Michigan water column segments for the organic
carbon dynamics calibration of the LM2-Toxic and the LMMBP cruise mean data 295
xxv
-------
4.5.3 Temporal profiles of BIG in the Lake Michigan water column segments for the organic
carbon dynamics calibration of the LM2-Toxic and the LMMBP cruise mean data 298
4.5.4. Temporal profiles of PDC in the Lake Michigan water column segments for the organic
carbon dynamics calibration of the LM2-Toxic and the LMMBP cruise mean data 301
4.5.5 Observed data versus the LM2-Toxic predictions for DOC, BIG, and PDC for the
LMMBP period 304
4.5.6 Temporal profiles of PCB28+31 (dissolved phase + particulate phase) in Lake
Michigan water column segments for PCB dynamics calibration of the LM2-Toxic and
the LMMBP cruise mean data 306
4.5.7 Temporal profiles of IPCBs (dissolved phase + particulate phase) in Lake Michigan
water segments for PCB dynamics calibration of the LM2-Toxic and the LMMBP
cruise mean data 309
4.5.8 Observed data versus the LM2-Toxic predictions for PCB28+31 and IPCBs for the
LMMBP period 312
4.5.9 Lake-wide average concentrations of (a) 137Cs and (b) 239'240pu computed by the LM2
radionuclide model over 46 years (1950-1995) 318
4.5.10 Sediment 137Cs inventory comparison between the observed data and the LM2
radionuclide model outputs 318
4.5.11 Reconstructed historical total PCB loading time functions and sediment core
LM94-15A total PCB concentration profiles for Lake Michigan 323
4.5.12 Reconstructed total PCB vapor phase concentrations and total PCB loading time
functions for Lake Michigan 323
4.5.13 Reconstructed total organic carbon load (primary production + LMMBP tributary
loads) for Lake Michigan 325
4.5.14 The sampling sites of the sediment box core samples (LM94-15A, LM95-61A,
LM95-87A) taken during the LMMBP for which vertical PCB concentration profiles
were analyzed and available 327
4.5.15 Annual lake-wide average total PCB water column concentrations from the LM2-Toxic
PCB hindcast simulation 329
4.5.16 Monthly lake-wide average total PCB water column concentrations from the LM2-Toxic
PCB hindcast simulation 329
4.5.17 Annual average total PCB concentration profiles in the sediment depositional zone
from the LM2-Toxic PCB hindcast simulation 330
4.5.18 IPCB mass budget of Lake Michigan during the period of the LM2-Toxic PCB
hindcast (1949-1995) 332
XXVI
-------
4.5.5.1 Model data versus measured total phosphorus loads 427
4.5.5.2 Model output versus measured total phosphorus concentration data 428
4.5.5.3 Relationship between lake-wide total phosphorus concentration and primary
productivity based on model output from the LM3-Eutro model 428
4.6.1 Mass budget average for 1994-1995 total RGBs in the Lake Michigan system
(including Green Bay) 432
4.6.2 1994-1995 total PCB Lake Michigan and Green Bay mass budget (averaged) 433
4.6.3a Annual long-term responses to total PCB concentrations in the water column of
Lake Michigan for the forecast scenarios and USEPA water quality criteria for the
protection of wildlife (U.S. Environmental Protection Agency, 2005) and human
health (U.S. Environmental Protection Agency, 1997) in the Great Lakes system 438
4.6.3b Annual long-term responses to total PCB concentrations in the water column of
Lake Michigan for the sensitivity scenarios 438
4.6.4a Monthly long-term responses to total PCB concentrations in the water column of
Lake Michigan for the forecast scenarios and USEPA water quality criteria for the
protection of wildlife (U.S. Environmental Protection Agency, 2005) and human
health (U.S. Environmental Protection Agency, 1997) in the Great Lakes system 439
4.6.4b Monthly long-term responses to total PCB concentrations in the water column of
Lake Michigan for the sensitivity scenarios 439
4.6.5a Annual long-term responses to total PCB concentrations in the surficial sediment of
Lake Michigan for the forecast scenarios 441
4.6.5b Annual long-term responses to total PCB concentrations in the surficial sediment of
Lake Michigan for the sensitivity scenarios 441
4.7.1 Short-term (1994-1995) variations of lake-wide (Green Bay included) organic carbon
concentrations for ± 50% primary production changes without adjusting settling
and resuspension rates 445
4.7.2 Long-term (1994-2055) variations of lake-wide (Green Bay included) organic carbon
concentrations for ± 50% primary production changes without adjusting settling
and resuspension rates 446
4.7.3 Short-term (1994-1995) variations of lake-wide (Green Bay included) PCB28+31
(dissolved + particulate) concentrations for ± 50% primary production changes
without adjusting settling and resuspension rates , 448
4.7.4 Long-term (1994-2055) variations of lake-wide (Green Bay included) PCB28+31
(dissolved + particulate) concentrations for ± 50% primary production changes
without adjusting settling and resuspension rates 449
XXVII
-------
4.7.5 Sensitivity analysis of the LM2-Toxic predictions to varying PCB loads .................. 451
5.3.1 Primary chemical exchange processes between a fish and its environment ............... 458
5.3.2 Comparison of modeling approaches for exposure temperatures in food web models ....... 463
5.4. 1 Biota zones in Lake Michigan .................................................. 46?
5.4.2a Typical annual cycles of exposure temperature for Lake Michigan food webs at
Saugatuck and Sturgeon Bay .................................................. 485
5.4.2b Typical annual cycles of exposure temperature for Lake Michigan food web at
Sheboygan Reef [[[ 485
5.4.2c Typical annual cycles of exposure temperature for coho salmon in Lake Michigan ......... 486
5.5.1 Agreement between modeled and observed fish PCB concentrations in coho
salmon using Saugatuck food web (1 994 and 1 995) ................................. 497
5.5.2 Individual comparison between modeled and observed data for PCB congeners in
lake trout at Saugatuck (1 994 and 1 995) ......................................... 499
5.5.3 Comparison between modeled and observed total PCBs for lake trout at Saugatuck
(1994 and 1995) [[[ 500
5.7. 1 Sensitivity of PCBs in lake trout (age four) to chemical assimilation efficiency
presented as ratios of model outputs with modified chemical assimilation efficiency
to model outputs with the calibrated chemical assimilation efficiency .................... 530
5.7.2 Effect of changes in food assimilation efficiency on the computed PCB data for lake
trout in Lake Michigan presented as ratios of model outputs with modified food
assimilation efficiency to model outputs with the calibrated food assimilation efficiency ...... 530
5.7.3 Effect of changes in chemical relative gill transfer coefficient (E^Eo) on the computed
PCB data for lake trout in Lake Michigan presented as ratios of model outputs with
modified chemical relative gill transfer coefficient to model outputs with the
calibrated chemical relative gill transfer coefficient ................................ 531
5.7.4 Effect of changes in SDA on the computed PCB data for lake trout in Lake Michigan
presented as ratios of model outputs with modified SDA parameter to model outputs
with the calibrated SDA value .................................................. 532
5.7.5 Sensitivity of PCBs in lake trout (age four) to fish growth rate presented as the ratios
of model outputs with zero lake trout growth rate to model outputs with field estimated
growth rate [[[ 533
5.7.6 Sensitivity of PCBs in lake trout (age four) to octanol-water partition coefficient (logKow)
-------
5.7.7 Sensitivity of PCBs in lake trout (age four) to fish diet presented as ratios of model
outputs with modified fish diet to model outputs with field estimated fish diet 534
5.8.1 a PCB congener-specific exposure concentrations at Sturgeon Bay predicted by
LM2-Toxic for Scenario A - PCBs in suspended particles of the water column 538
5.8.1 b PCB congener-specific exposure concentrations at Sturgeon Bay predicted by
LM2-Toxic for Scenario A - PCBs in the surface sediment 539
5.8.2 Total PCB concentrations of the lake trout in response to the exposure concentration
inputs associated with various loading scenarios 540
5.8.3 Total PCB concentrations of the lake trout in response to the exposure concentration
inputs associated with various loading scenarios 541
6.1 Supporting models and links for MICHTOX and the LM models 545
6.2 Comparison of the Lake Michigan total PCB mass balance analyses results,
1994-1995 547
6.3 Comparison of model output annual average total PCB water concentrations for
the Constant Conditions Scenario 548
6.4 Comparison of model output annual average total PCB sediment concentrations for
the Constant Conditions Scenario 549
6.5 Comparison of model output annual average total PCB water concentrations for
the Continued Recovery - Fast Scenario 549
6.6 Comparison of the bioaccumulation model annual average total PCB concentration
results for the Constant Conditions Scenario 550
6.7 Comparison of the bioaccumulation model annual average total PCB concentration
results for the Continued Recovery - Fast Scenario 551
XXIX
-------
List of Tables
1.1.1 Characteristics of the LMMBP Modeled Pollutants 4
1.1.2 The LMMBP Parameters 7
1.3.1 Revised Regression Equations for the LMMBP Total PCBs in All Media 29
1.3.2 Ratio of Measured Field Data/Model Results for Congeners 84+92 and 99 in Water 30
1.4.1 Summary of Lake Michigan Ice Cover Based Upon Asset (2003) 57
1.4.2 Description Wave Statistics for POM Calibration Years (1982-1983) and Study
Years (1994-1995) Compared to the Period of Record for NOAA's Buoys in
Northern and Southern Lake Michigan 64
1.5.1 Significant Dates in the History of PCBs in the Lake Michigan Basin 67
1.5.2 Monthly Composite Concentrations of Vapor Phase Total PCBs Measured in Samples
Collected Around Lake Michigan From April 1994 to October 1995 69
1.5.3 Monthly Composite Concentrations of Total PCBs Measured in Precipitation Samples
Collected Around Lake Michigan From April 1994 to October 1995 73
1.5.4 Monthly Composite Concentrations of Particulate Phase Total PCBs Measured in
Samples Collected Around Lake Michigan From April 1994 to October 1995 75
1.5.5 Monthly Composite Concentrations of PCBs Measured in Dry Deposition 76
1.5.6 Concentrations of PCBs in 1994-1995 Lake Michigan Water (ng/L) 77
1.5.7 Concentrations of PCBs Measured in Tributaries 82
1.5.8 Comparison of PCB Concentrations in Samples Collected From Tributaries in 1994-1995
With Those in Samples Collected From Tributaries in 1980-1983 84
1.5.9 Concentrations of Total PCBs in Lake Michigan Surficial Sediment (ng/g) 84
1.5.10 Physical Parameters Associated With LMMBP Cores 86
xxx
-------
1.5.11 Mean Concentrations of PCBs Measured in the 1994-1995 Lake Michigan
Food Web 89
1.6.1 Comparison of Congeners Available for Analysis in All LMMBP Media 95
1.7.1 Significant Dates in the History of PCBs in the Lake Michigan Basin 111
1.7.2 Sedimentary Zones of Lake Michigan 115
2.3.1 Nutrient State Variables 126
2.4.1 1994-1995 Monthly Atmospheric Total Phosphorus Loads 140
2.4.2 Tributary Total Phosphorus Loads (kg/year) 141
2.4.3 Sediment Masses, Fluxes, and Loads 142
2.4.4 The LMMBP Sampling Cruises 142
2.4.5 The LMMBP Open Lake Nutrient Data Summary Statistics 144
2.4.6 Relationship of Field Measurements and Model State Variables 152
2.4.7 Important LM3 Model Coefficients 153
2.5.1 Coefficients Used in the LM3 Model (Units Correspond to Required LM3 Model Output) .... 162
2.5.2 Summary of Statistical Results of the Calibration 165
2.7.1 Final Eutrophication Scenario Results 179
3.3.1 MICHTOX Segment Geometry 188
3.3.2 Model Parameters and Coefficients 189
3.3.3 Cruise and Segment-Specific Average Dissolved Total PCB Concentrations (ng/L) 190
3.3.4 Cruise and Segment-Specific Average Particulate Total PCB Concentrations (ng/L) 190
3.3.5 Segment-Specific Average Surficial Sediment Total PCB Concentrations (ng/g) 190
3.4.1 Average Total PCB Concentrations in Fish in the Saugatuck Biota Zone 203
3.4.2 Average Total PCB Concentrations in Fish in the Sheboygan Reef Biota Zone 204
3.4.3 Average Total PCB Concentrations in Fish in the Sturgeon Bay Biota Zone 204
3.4.4 MICHTOX Food Chain Age- and Species-Specific Weight, Growth Rate, and Lipid
Concentrations 205
XXXI
-------
3.4.5 MICHTOX Food Chain Model Parameters and Coefficients 206
A3.4.1 Comparison of the LMMBP Lake Trout to MDEQ Lake Superior Lake Trout 213
4.3.1 Geometry Data for Water Column Segments and Lake Michigan (Total) 229
4.3.2 Initial Geometry Data for Surficial Sediment Segments and Surficial Sediment
Layer (Total) 230
4.3.3 Processes Considered in Organic Carbon Sorbent Dynamics Constructed for
the LM2-Toxic 233
4.4.1 Average Annual Flows of the 10 Monitored Tributaries 249
4.4.2 Monthly Average Flows Across the Straits of Mackinac 251
4.4.3 Initial Temperatures in Water Column Segments for the Thermal Balance Model 251
4.4.4 Cruise-Segment Mean Temperatures for the LMMBP Period 252
4.4.5 The LMMBP Sampling Cruises 255
4.4.6 Initial Chloride Concentrations in Water Column Segments for the Chloride Model 255
4.4.7 Cruise-Segment Mean Chloride Concentrations for the LMMBP Period 256
4.4.8 Annual Average Organic Carbon Loads From 11 Monitored Tributaries to Lake
Michigan During the LMMBP 259
4.4.9 Annual Average Organic Carbon Loads From 18 Unmonitored Tributaries to Lake
Michigan During the LMMBP 259
4.4.10 Annual Average Organic Carbon Internal Loads Generated From the LM3-Eutro
for Lake Michigan During the LMMBP 26°
4.4.11 Initial Concentrations of Organic Carbon Sorbents in Water Column Segments for
the LM2-Toxic 261
4.4.12 Cruise-Segment Mean Concentrations of DOC (mg/L) for the LMMBP Period 262
4.4.13 Cruise-Segment Mean Concentrations of BIG (mg/L) for the LMMBP Period 263
4.4.14 Cruise-Segment Mean Concentrations of PDC (mg/L) for the LMMBP Period 264
4.4.15 Concentration of Organic Carbon Sorbents in Surficial Sediments for the LM2-Toxic 267
4.4.16 Organic Carbon Sorbent Biotransformation Parameters Specified for the LM2-Toxic 268
4.4.17 Segment-Specific Settling Rates (m/d) for Organic Carbon Sorbents (BIG and PDC)
Specified for the LM2-Toxic 269
XXXII
-------
4.4.18 Segment-Specific Effective PDC Concentrations (Cw) Used in the Steady-State
Resuspension Calculation Based on the LMMBP Data 270
4.4.19 Segment-Specific Sediment Accumulation Rates (vb) and Thickness of Mixing Layer
(z) Used in the Steady-State Resuspension Calculation 270
4.4.20 Segment-Specific Critical Wave Heights (Wcr) and Empirical Wave Coefficients (a)
Used in the Wave-Induced Resuspension Calculation Based on the LMMBP Data 271
4.4.21 List of PCB State Variables Modeled in the LM2-Toxic 272
4.4.22 Annual Average IPCB Loads From 11 Monitored and 18 Unmonitored Tributaries to
Lake Michigan During the LMMBP 274
4.4.23 Annual Average IPCB Atmospheric Dry and Wet Loads in the 10 Surface Water
Column Segments of Lake Michigan During the LMMBP 274
4.4.24 Annual Average Boundary Conditions of IPCB Vapor Phase Concentrations for
Lake Michigan During the LMMBP 276
4.4.25 Initial Concentrations of IPCBs in Water Column Segments for Lake Michigan 278
4.4.26 Cruise-Segment Mean Concentration of IPCBs (ng/L) for the LMMBP Period 279
4.4.27 Initial Concentrations of IPCBs in Sediment Segments for Lake Michigan 280
4.4.28 Final Partition Coefficients for the LMMBP Selected PCBs Used in the LM2-Toxic 282
4.4.29 Values of Parameters Used for Air-Water Exchange in the LM2-Toxic for the LMMBP
Selected PCB Congeners 283
4.5.1 Results of the Regression Between the LM2-Toxic Calibration Results and the Cruise
Mean Data for the LMMBP Selected PCB Congeners 314
4.5.2 Results of the LM2-Toxic Mass Balance Checking for a 62-Year Simulation of an
Assumed Conservative Tracer 315
4.5.3 Comparison Between the LMMBP Field-Generated and the LM2-Toxic-Generated
Sediment Accumulation Rates (cm/year) 320
4.5.4 Available Historical Water Column Total PCB Concentrations for Lake Michigan 325
4.5.5 Sediment PCB Concentration Vertical Profiles Analyzed for Three Sediment Box
Cores Taken During the LMMBP 326
4.5.6 Available Inventories of PCBs in Lake Michigan Sediments 327
4.5.7 Calculations in PCB Mass Budget Checking for the LM2-Toxic PCB Hindcast 332
XXXIII
-------
4.6.1 Regression Equations Used for Converting IPCBs to Total RGBs for the LM2-Toxic
Results 431
4.6.2 Results of Total PCB Mass Budget Analysis for Lake Michigan and Green Bay 434
4.6.3 Mean and Medium Particulate RGBs/Organic Carbon and Field Data and Scaling
Factor for Hypolimnetic Level 2, Segments 21, 30, and 37, and for Saugatuck
Biota Box Hypolimnion 442
4.7.1 Annual Average Concentrations of Water Column Carbon Solids and Annual
Average Change in Percentage for Water Column Carbon Solids Concentrations
Resulting From the LM2-Toxic Model Runs for Both the Short-Term (1994-1995)
and the Long-Term (1994-2055) Simulations with 50% Increase and 50%
Decrease of the LM3-Eutro Produced Primary Production 447
4.7.2 PCB28+31 Mass Fluxes and Inventories for Lake Michigan System Results From
the LM2-Toxic Sensitivity Analysis on Primary Production for the Short-Term (Two-
Year Period: 1994-1995) Simulations 448
4.7.3 PCB28+31 Mass Fluxes and Inventories for Lake Michigan System Results From
the LM2-Toxic Sensitivity Analysis on Primary Production for the Last Two Years
of the Long-Term (62-Year Period: 1994-2055) Simulations 449
4.7.4 PCB28+31 Average Inventories of Water Column and Surficial Sediment Results
From the LM2-Toxic Simulations for the Primary Production Sensitivity Analysis,
and Changes in Percentage for These Inventories Compared to the Inventories
From the Original Base Runs 451
5.4.1 Targeted PCB Congeners and Their Kow 466
5.4.2a Annual Dietary Composition of Lake Trout at Saugatuck (1994-1995) 469
5.4.2b Annual Dietary Composition of Lake Trout at Sheboygan Reef (1994-1995) 470
5.4.2c Annual Dietary Composition of Lake Trout at Sturgeon Bay (1994-1995) 472
5.4.3 Dietary Composition of Alewife in Lake Michigan (1994-1995) 473
5.4.4. Dietary Composition of Bloater in Lake Michigan (1994-1995) 474
5.4.5 Dietary Composition of Rainbow Smelt in Lake Michigan (1994-1995) 474
5.4.6 Dietary Composition of Slimy Sculpin in Lake Michigan (1994-1995) 474
5.4.7 Dietary Composition of Deepwater Sculpin in Lake Michigan (1994-1995) 475
5.4.8 Dietary Composition of Coho Salmon in Lake Michigan (1994-1995) 475
5.4.9a Average Weight-Age Relationships for Lake Trout in Lake Michigan (1994-1995) 477
xxxiv
-------
5.4.9b Average Weight-Age Relationships for Coho Salmon in Lake Michigan (1994-1995) 477
5.4.9c Average Weight-Age Relationships of Forage Fish in Lake Michigan (1994-1995) 478
5.4.9d Estimated Weight-Age Relationships of Mysis in Lake Michigan 478
5.4.1 Oa Average Lipid and Protein Fractions (%) of Lake Trout in Lake Michigan (1994-1995) 478
5.4.1 Ob Average Lipid and Protein Fractions (%) of Coho Salmon in Lake Michigan (1994-1995) .... 479
5.4.1 Oc Average Lipid and Protein Fractions (%) of Alewife in Lake Michigan (1994-1995) 479
5.4.1 Od Average Lipid and Protein Fractions (%) of Bloater in Lake Michigan (1994-1995) 480
5.4.1 Oe Average Lipid and Protein Fractions (%) of Rainbow Smelt in Lake Michigan
(1994-1995) 480
5.4.1 Of Average Lipid and Protein Fractions (%) of Slimy Sculpin in Lake Michigan
(1994-1995) 480
5.4.10g Average Lipid and Protein Fractions (%) of Deepwater Sculpin in Lake Michigan
(1994-1995) 481
5.4.1 Oh Average Lipid and Protein Fractions (%) of Zooplankton, Mysis, and Diporeia in Lake
Michigan (1994-1995) 481
5.4.11 PCB Concentrations in Lake Michigan Water Column (1994-1995) 482
5.4.12 PCB Concentrations in Lake Michigan Surface Sediment (1994-1995) 484
5.4.13 Bioenergetic Parameters of Lake Michigan Fishes 487
5.5.1 Calibrated Parameter Values for Diporeia Submodel 494
5.5.2 Calibrated Model Parameters for PCBs in the Sturgeon Bay and Saugatuck Lake
Trout Food Webs 494
5.5.3 Calibrated Model Parameters for PCBs in the Sheboygan Reef Lake Trout Food Web 494
5.5.4 Calibrated Model Parameters for PCBs in Lake Michigan Coho Salmon 495
7.1.1 Agenda - Lake Michigan Mass Balance PCB Modeling Peer Review 554
7.1.2 Significant Dates in the History of PCBs in the Lake Michigan Basin 562
xxxv
-------
Abbreviations
AOCs Areas of Concern
AREAL Atmospheric Research and Exposure Assessment Laboratory
BIG Biotic carbon
BMC Bayesian Monte Carlo
BNL Benthic nepheloid layer
CCC Criterion continuous concentration
CMC Criterion maximum concentration
CPE Catch per unit of effort
CTF Contaminant transport and fate model
DDE Dichlorodiphenyldichloroethylene
DEA Deethylatrazine
DIA Deisopropylatrazine
DIN Dissolved inorganic nitrogen
DOC Dissolved organic carbon
DON Dissolved organic nitrogen
DOP Dissolved organic phosphorus
DQO Data quality objectives
DSi Dissolved silica
EEGLE Episodic Events-Great Lakes Experiments
EMP Enhanced Monitoring Program
GLERL Great Lakes Environmental Research Laboratory
GLNPO Great Lakes National Program Office
GLWQA Great Lakes Water Quality Agreement
GBMBP Green Bay Mass Balance Project
HOC Hydrophobic organic chemicals
IADN Integrated Atmospheric Deposition Network
IDL Instrument detection limit
IDW Inverse distance weighted
LaMP Lake-wide Management Plan
LLRFRB Lake Lakes and Rivers Forecasting Research Branch
LLRS Large Lakes Research Station
LMMBP Lake Michigan Mass Balance Project
LOG Labile organic carbon
LON Labile organic nitrogen
LOP Labile organic phosphorus
MCL Maximum contaminant level
MDEQ Michigan Department of Environmental Quality
xxxvi
-------
MDL Method detection limit
MED Mid-Continent Ecology Division
MQO Measurement quality objectives
NHEERL National Health and Environmental Effects Research Laboratory
NOAA National Oceanic and Atmospheric Administration
ORD Office of Research and Development
PCBs Polychlorinated biphenyls
PDC Particulate detrital carbon
PI Principal Investigator
POC Particulate organic carbon
POM Princeton Ocean Model
POP Persistent organic pollutants
QA Quality assurance
QAPP Quality Assurance Project Plan
QC Quality control
RAPs Remedial Action Plans
RDMQ Research Data Management and Quality Control System
RFS Routine field sample
ROC Refractory organic carbon
RON Refractory organic nitrogen
ROP Refractory organic phosphorus
SA Available silica
SDA Specific dynamic action
SDL System detection limit
SRP Soluble reactive phosphorus
SU Biogenic silica
TKN Total Kjeldahl nitrogen
TMDL Total maximum daily load
USDOI United States Department of Interior
USEPA United States Environmental Protection Agency
USFWS Unites States Fish and Wildlife Service
USGS United States Geological Survey
VWA Volume-weighted average
XXXVII
-------
Acknowledgments
Special thanks to the United States Environmental Protection Agency, Great Lakes National Program Office
for leadership, support, and collaboration on the Lake Michigan Mass Balance Project. The multiple efforts
by the Principal Investigators for providing data, necessary for the modeling, are greatly appreciated. Thank
you to David H. Miller, Kenneth R. Rygwelski, Timothy J. Feist, Xiaomi Zhang, and James P. Pauer for
providing internal technical reviews of various parts of this document. Thanks to Kay Morrison for the graphic
renditions and figures and to Debra L. Caudill for formatting and word processing. Finally, thanks to Robert
B. Ambrose, Jr., Joel E. Baker, Ken G. Drouillard, Barry Lesht, and James L. Martin for serving on the peer
review panel.
XXXVIII
-------
Executive Summary
The Lake Michigan Mass Balance Project (LMMBP) provided an opportunity to improve the quality of
polychlorinated biphenyl (PCB) mass balance models used to represent large, freshwater ecosystems. A
rigorously quality-assured large supporting data set derived from samples collected during eight cruises in
1994-1995 was used to establish atmospheric and tributary loads, estimate initial conditions, perform model
calibration and confirmation and, to a lesser extent, to assist in estimating a number of kinetic coefficients.
A significant aspect of this modeling effort was modeling PCBs at a congener-level basis.
Lake Michigan is acted upon by a number of physical parameters that impact the hydrology, chemistry, and
biology of the lake. For a lake the size of Lake Michigan, changes in these parameters can lead to significant
changes, especially when models are used in long-term predictions to predict the outcome of various
scenarios. The primary driving forces are wind, air temperature, and precipitation. These impact tributary
flows, lake levels, waves, water circulation, water temperature, and ice cover. For the period of record, these
driving forces vary from year-to-year. The period of 1982 to 1983 was used to calibrate the hydrodynamic
models. For this period of time, hydrodynamic conditions were not at any extreme. This is also true for the
period of 1994 and 1995 when the models were applied.
Major physical forcing functions were not extreme during the sampling period of 1994-1995 or the
hydrodynamic model calibration period of 1982-1983. Precipitation was within the normal range for all years
of modeling interest, resulting in lake levels and tributary flows that were within normal bounds.
Temperature will impact the eutrophication and contaminant modeling. Air temperature impacts how quickly
the lake warms in any one year. Water temperature is critical to the timing of algae blooms, especially the
spring diatom bloom. It also impacts the volatilization of contaminants. There appears to be a four-year cycle
of quicker warming which exists within a trend of general warming of the lake. The trend of warming may be
part of a longer term, undocumented cycle, or may be related to climate change.
Models developed at the United States Environmental Protection Agency's Large Lakes Research Station,
including LM3-Eutro, MICHTOX, LM2-Toxic, and LM Food Chain, utilized results from a hydrodynamic model
to describe the lake's physics and results from air and tributary models to provide loadings to the lake.
LM3-Eutro uses state-of-the-science eutrophication kinetics to simulate the interactions between plankton and
nutrients. LM3-Eutro is a high-resolution (44,042 cells and 19 sigma layers) carbon based model that provides
a highly resolved description of areas such as near and off shore zones, bays, river confluences, and the
thermocline. Its nutrient variables include dissolved, labile particulate, and refractory particulate forms which
provide a more realistic description of phytoplankton-nutrient interactions. Improvements were made to the
light calculation by using a three-hour rather than 24-hour (one day) average estimate of solar radiation. The
model is driven by the Princeton Ocean hydrodynamics Model which simulates water movements. LM3-Eutro
xxxix
-------
has 17 state variables, including a single zooplankton class, two phytoplankton classes, and several participate
and dissolved nutrient (including carbon) states.
The model was calibrated on the high-resolution (44,042 cells) Level 3 framework as well as the 41 segment
Level 2 framework. The Level 2 calibration enabled us to visually observe known spatial and temporal trends
such as the spring diatom bloom and phytoplankton concentration gradients between the epilimnion and
hypolimnion. The Level 3 calibration was performed on a whole-lake basis. The 1994-1995 LMMBP field data
were used to calibrate the model. The final calibration was chosen based on the best Level 3 calibration, but
Level 2 output was visually inspected to ensure that expected phytoplankton and nutrients trends were
reflected. Model confirmation was performed by comparing the model to limited total phosphorus data for
1998 and 2000 and to a historical model, MICH1, which was developed and calibrated in the 1970s and
modified more recently. All comparisons were done on a whole-lake basis, and LM3-Eutro fits the 1998 and
2000 data well. LM3-Eutro and MICH1 compared surprisingly well, especially given the fact that they are
based on very different frameworks, kinetics, and segmentation. Compared to field data and LM3-Eutro
predicted, MICH1 underpredicted both total phosphorus concentrations. This was probably due to the fact that
MICH1 does not have any phosphorus sediment recycling. Lower phosphorus values also cause MICH1 to
under-predict chlorophyll a concentrations in the lake.
MICHTOX is a toxic chemical mass balance and food chain bioaccumulation model developed in the early
1990s. A Bayesian Monte Carlo uncertainty analysis demonstrated that MICHTOX predicted PCB
concentrations should be within a factor of two measured data. During the early part of the LMMBP, MICHTOX
was updated and used as a preliminary assessment tool of the LMMBP PCB data and to provide a screening-
level analysis of the potential future trends in total PCB concentrations in Lake Michigan water, sediment, and
fish under a variety of contaminant load scenarios. Unmonitored tributary inputs were added to the model and
the applicability of MICHTOX for predicting Lake Michigan total PCB concentrations in water, sediment, and
fish was reconfirmed. MICHTOX was applied using the previously developed parameterization and LMMBP
data and forcing functions. The model fit to data was acceptable with no adjustments to the model
parameters. The model also provided a comparison of an older, "off-the-shelf model with the more complex
models developed as part of the Lake Michigan Mass Balance Project (LMMBP). MICHTOX was run for seven
scenarios to help evaluate the impacts on PCB trends caused by various loading sources and to evaluate
loading scenarios. Results of the MICHTOX modeling indicate that atmospheric exchange is a dominant loss
process of total PCBs in Lake Michigan, and that the reservoir of total PCBs in the sediment has a significant
impact on the future trends in concentrations of total PCBs in lake trout.
LM2-Toxic is a sophisticated and state-of-the-art toxic chemical fate and transport model for Lake Michigan.
It is a coupled mass balance of organic carbon solids and toxic chemical (PCBs) dynamics. Using the LMMBP
generated field data, the organic carbon solids dynamics were first calibrated. This was followed by the
independent calibration of PCB dynamics. The temporal variations of both biotic carbon (BIC) and particulate
detrital carbon (PDC) resulted from an algal bloom in late spring and early summer. Primary production was
the dominant organic carbon load to Lake Michigan. The eutrophication model (LM3-Eutro) generated primary
production accounted for over 90 percent of the total particulate organic carbon load to the lake.
The main focus of this model is to address the relationship between sources of toxic chemicals and their
concentrations in water and sediments of Lake Michigan and to provide the PCB exposure concentrations to
the bioaccumulation model (LM2 Food Chain) to predict PCBs concentrations in lake trout tissue. LM2-Toxic
is a revision of the USEPA supported WASP4 water quality modeling framework. It incorporates the organic
carbon dynamics featured in G BTOX and the sediment transport scheme, a quasi-Lagrangian framework, used
in the IPX. Both GBTOX and IPX were WASP4-type models and major components in the Green Bay Mass
Balance Study modeling framework. Another important modification was the addition of updated air-water
exchange formulations to the model.
xl
-------
The results at 5 x 5 km2 grid generated by Princeton Ocean Model for the Great Lakes (POMGL) were linked
to the transport fields for LM2-Toxic. Due to an affinity of PCBs for organic carbon, three organic carbon
sorbents were simulated as state variables in LM2-Toxic. They were BIG, PDC and dissolved organic carbon
(DOC). The model simulated 54 PCB congeners which accounted for 63% to 85% of the total PCB mass in
various media for Lake Michigan. This was an enormous effort because individual congeners or co-eluting
congeners were modeled as separate state variables in the mass balance, each with their own
physical/chemical properties. Four phases were simulated in LM2-Toxic for the congeners. The four phases
were dissolved, sorbed to PDC, sorbed to BIG, and bound to DOC.
To reduce uncertainties associated with water transport, settling and resuspension, and sedimentation, a
thermal balance model, a chloride model, a long-term simulation using a 137Cs and 239'240pu model, and a long-
term organic carbon simulation using LM2-Toxic were developed and run for LM2-Toxic confirmation.
Air-water exchange of PCBs was the most important process for Lake Michigan. Net sediment resuspension
was the second largest net source. Both the water column and the surficial sediment layer of the lake were
not at steady-state during the LMMBP period. The model was also applied for forecasting the long-term
responses (60-year simulation, starting on January 1, 1996) of the PCBs in Lake Michigan under various
forcing functions and load reduction scenarios. The results indicate that the PCB mass in the surficial sediment
is large and thus could support PCB concentrations in the water column for a very long time.
LM Food Chain is the food web bioaccumulation model developed for the LMMBP. The model established
dynamic relationships between PCBs concentrations in the exposure environments and resulting PCBs levels
in the lake trout food webs of Lake Michigan. The model was based upon available theory and data
characterizing the bioaccumulation of toxic chemicals in fish and other aquatic organisms. Samples collected
for the LMMBP were used to generate data on lake trout and coho salmon food webs in Lake Michigan and
to facilitate refinement of model parameters to site-specific conditions for forty PCB congeners or co-eluters
that represented toxic chemicals covering a wide range of hydrophobicity.
The food web model was calibrated with PCB data collected in 1994 and 1995 for lake trout food webs at
Sturgeon Bay, Sheboygan Reef, and Saugatuck. The lake trout sub-populations in these three biota zones
were believed to be appropriate representations of lake trout in Lake Michigan. Model calibration was also
performed for a lake-wide coho salmon food web. During the model calibration, model parameters were
refined to achieve an adequate agreement between model calculations and observed PCB data for a food
web. The focus of model calibration was not limited to top predators or to toxics with a certain hydrophobicity.
The model parameters were systematically optimized for all species at various trophic levels and for PCB
congeners of a wide range of hydrophobicity. Extra care was taken to ensure the refined parameter values
were consistent with the hydrophobicity of individual PCB congeners and with the trophic position of individual
species. Satisfactory calibration results were achieved for the lake trout food webs at Sturgeon Bay and
Saugatuck. The model parameters calibrated with data from the Sturgeon Bay food webs were independently
tested and validated with data from the Saugatuck food web, and vice versa.
The availability of a complete account of observed data for each food web made this model calibration
probably the most thorough process among similar efforts. Although PCB concentrations in lake trout or coho
salmon were the endpoint of the model computation and the focus of most model applications, the food web
model with parameters "fine-tuned" for species at all trophic levels can be used to target any desirable species
in the food web with a high degree of confidence. Also, the food web model can be used to model toxics with
various hydrophobicities. No food web model intended to simulate as many toxic chemicals with diverse
hydrophobicity has been previously developed.
xli
-------
The validated food web model was applied to the lake trout food webs at Sturgeon Bay and Saugatuck and
inferred for the southeastern and northwestern regions of the lake to predict future PCB concentrations.
Several model simulations were performed to predict the expected changes in future lake trout PCB
concentrations in response to different exposure scenarios. Hypothetical long-term PCB exposure scenarios
in the post-1994/1995 period for the food webs at the Sturgeon Bay and Saugatuck biota zones were
generated by the water quality model LM2-Toxic. For each lake trout food web, the resulting concentrations
of individual PCB congeners in fish were predicted. Similar model predictions were observed for these two
biota zones under each reduction scenario. For the continued fast recovery scenario, current simulations
indicate that the total PCB concentrations in adult lake trout (5.5 years old) were expected to reach the target
level of 0.075 ppm in 2030 for the Saugatuck biota zone, 2033 for southeastern Lake Michigan, 2036 for
northwestern Lake Michigan, and 2036 for the Sturgeon Bay biota zone.
xlii
-------
PART1
INTRODUCTION
Chapter 1. Project Overview
Harry B. McCarty, Ken Miller, Robert N. Brent, and
Judy Schofield
DynCorp (a CSC Company)
601 Stevenson Avene
Alexandrea, Virginia 22304
and
Ronald Rossmann and Kenneth R. Rygwelski
U.S. Environmental Protection Agency
Office of Research and Development
National Health and Environmental Effects
Research Laboratory
Mid-Continent Ecology Division
Large Lakes and Rivers Forecasting Research
Branch
Large Lakes Research Station
9311 Groh Road
Grosse lie, Michigan 48138
The United States Environmental Protection
Agency's (USEPA) Great Lakes National Program
Office (GLNPO) instituted the Lake Michigan Mass
Balance Project (LMMBP) to measure and model the
concentrations of representative pollutants within
important compartments of the Lake Michigan
ecosystem. For the LMMBP, concentrations of
polychlorinated biphenyls (PCBs), frans-nonachlor,
atrazine, and mercury in tributaries, lake water,
sediments, food webs, and the atmosphere
surrounding Lake Michigan were measured. This
document contains the PCB modeling results
reported by staff and contractors of the
USEPA/Office of Research and Development
(ORD)/National Health and Environmental Effects
Research Laboratory (NHEERL)/Large Lakes and
Rivers Forecasting Research Branch (LLRFRB) staff
and contractors located at the Large Lakes Research
Station (LLRS).
1.1.1 Background
The Great Lakes, which contain 20% of the world's
freshwater, are a globally important natural resource
currently threatened by multiple stressors. While
significant progress has been made to improve the
quality of the lakes, pollutant loads from point, non-
point, atmospheric, and legacy sources continue to
impair ecosystem functions and limit the attainability
of designated uses of these resources. Fish
consumption advisories and beach closings continue
to be issued, emphasizing the human health
concerns from lake contamination. Physical and
biological stressors, such as invasion of non-native
species and habitat loss, also continue to threaten
the biological integrity of the Great Lakes.
The United States and Canada have recognized the
significance and importance of the Great Lakes as a
natural resource and have taken steps to restore and
protect the lakes. In 1978, both countries signed the
Great Lakes Water Quality Agreement (GLWQA).
This Agreement calls for the restoration and
maintenance of the chemical, physical, and biological
integrity of the Great Lakes by developing plans to
monitor and limit pollutant flows into the lakes.
The GLWQA, as well as Section 118(c) of the Clean
Water Act, require the development of a Lake-wide
Management Plan (LaMP) for each Great Lake. The
purpose of these LaMPs is to document an approach
to reduce inputs of critical pollutants to the Great
-------
Lakes and restore and maintain Great Lakes
integrity. To assist in developing these LaMPs and
to monitor progress in pollutant reduction, Federal,
State, Tribal, and local entities have instituted
Enhanced Monitoring Plans (EMPs). Monitoring is
essential to the development of baseline conditions
for the Great Lakes and provides a sound scientific
base of information to guide future toxic load
reduction efforts.
The LMMBP is a part of the EMPs for Lake Michigan.
The LMMBP was a coordinated effort among
Federal, State, and academic scientists to monitor
tributary and atmospheric pollutant loads, develop
source inventories of toxic substances, and evaluate
the fates and effects of these pollutants in Lake
Michigan. A mass balance modeling approach
provides the predictive ability to determine the
environmental benefits of specific load reduction
scenarios for toxic substances and the time required
to realize those benefits. This predictive ability will
allow Federal, State, Tribal, and local agencies to
make more informed load reduction decisions.
1.1.2 Description
The LMMBP used a mass balance approach to
evaluate the sources, transport, and fate of
contaminants in the Lake Michigan ecosystem. A
mass balance approach is based on the law of
conservation of mass, which states that the amount
of a pollutant entering a system is equal to the
amount of that pollutant leaving, trapped in, and
chemically changed in the system (Figure 1.1.1). In
the Lake Michigan system, pollutant inputs may
come from atmospheric deposition or tributary loads.
Simple Mass Budget for Conservative Substances
1
source
mass ;
in
water system
mass out = mass jn + Zsources
t
source
Mass Balance Modeling Approach
mass i
in
air system
-N-
water system
sediment system
air sources
mass out = mass jn + ^sources
± air-water exchange
± sediment-water exchange
± ^internal processes
Figure 1.1.1. Simplified mass balance approach.
-------
Pollutants may leave the system through burial in
bottom sediments, volatilization to the atmosphere, or
discharge into Lake Huron through the Straits of
Mackinaw. Pollutants within the system may be
transformed through degradation or stored in
ecosystem compartments such as the water column,
sediments, or biota.
For the LMMBP, contaminant concentrations in
various inputs and ecosystem compartments over
spatial and temporal scales were measured.
Mathematical models that track the transport and fate
of contaminants within Lake Michigan were
developed and calibrated using these field data. The
LMMBP was the first lake-wide application of a mass
balance determination for the Great Lakes and will
serve as a basis for future mass budget/mass
balance efforts.
1.1.3 Scope
1.1.3.1 Modeled Pollutants
When the USEPA published the Water Quality
Guidance for the Great Lakes System (58 FR
20802), the Agency established water quality criteria
for 29 pollutants. Those criteria were designed to
protect aquatic life, terrestrial wildlife, and human
health. PCBs, frans-nonachlor, and mercury are
included in the list of 29 pollutants. The water quality
criteria and values proposed in the guidance apply to
all of the ambient waters of the Great Lakes System,
regardless of the sources of pollutants in those
waters. The proposed criteria provide a uniform
basis for integrating Federal, State, and Tribal efforts
to protect and restore the Great Lakes ecosystem.
The number of pollutants that can be intensively
monitored and modeled in the Great Lakes System
is limited by the resources available to collect and
analyze thousands of samples, assure the quality of
the results, manage the data, and develop and
calibrate the necessary models. Therefore, the
LMMBP focused on constructing mass balance
models for a limited group of pollutants. PCBs, trans-
nonachlor, atrazine, and mercury were selected for
inclusion in the LMMBP because these pollutants
currently or potentially pose a risk to aquatic and
terrestrial organisms (including humans) in the Lake
Michigan ecosystem (Table 1.1.1). These pollutants
also were selected to cover a wide range of chemical
and physical properties and represent other classes
of compounds which pose current or potential
problems. Once a mass budget for selected
pollutants is established and a mass balance model
calibrated, additional contaminants can be modeled
with limited data and future resources can be
devoted to activities such as emission inventories
and dispersion modeling.
1.1.3.1.1 PCBs
Polychlorinated biphenyls (PCBs) are a class of man-
made, chlorinated, organic chemicals that include
209 congeners, or specific PCB compounds. The
highly stable, nonflammable, non-conductive
properties of these compounds made them useful in
a variety of products including electrical transformers
and capacitors, plastics, rubber, paints, adhesives,
and sealants. PCBs were produced for such
industrial uses in the form of complex mixtures under
the trade name "Aroclor" and were commercially
available from 1930 through 1977, when the USEPA
banned their production due to environmental and
public health concerns. PCBs also may be produced
by combustion processes, including incineration, and
can be found in stack emissions and ash from
incinerators.
Because they were found by the USEPA in the
effluents from one or more wastewater treatment
facilities, seven Aroclor formulations were included in
the Priority Pollutant List developed by the USEPA
Office of Water under the auspices of the Clean
Water Act. Aroclors may have entered the Great
Lakes through other means, including spills or
improperdisposal of transformerfluids, contaminated
soils washing into the watershed, or discharges from
ships. The PCBs produced by combustion
processes may be released to the atmosphere where
they are transported in both vapor and particulate
phases and enter the lakes through either dry
deposition or precipitation events (e.g., rain).
The stability and persistence of PCBs, which made
them useful in industrial applications, have also made
these compounds ubiquitous in the environment.
PCBs do not readily degrade and thus accumulate in
water bodies and aquatic sediments. PCBs also
bioaccumulate, or build up, in living tissues. Levels
of PCBs in some fish from Lake Michigan exceed
U.S. Food and Drug Administration tolerances,
-------
Table 1.1.1. Characteristics of the LMMBP Modeled Pollutants
Pollutant
RGBs
Sources
• Waste incinerators
(unintentional
byproducts of
combustion)
• Industrial
dischargers
• Electrical power
Uses
• Electrical
transformers and
capacitors
• Carbonless copy
paper
• Plasticizers
• Hydraulic fluids
Toxic Effects
• Probable human
carcinogen
• Hearing and vision
impairment
• Liver function alterations
• Reproductive impairment
and deformities in fish and
wildlife
Biocon-
centration
Factor1
1 ,800 to
180,000
USEPA
Regulatory
Standards2
MCL = 0.5 ug/L
CCC = 14 ng/L
HH=0.17ng/L
frans-Non-
achlor3
Atrazine
Application to crops
and gardens
Application to crops
Pesticide on com
and citrus crops
Pesticide on
lawns and
gardens
Herbicide for corn
and sorghum
production
Probable human 4,000 to MCL = 2 ug/L
carcinogen 40,000 CMC = 2.4 (jg/L
• Nervous system effects CCC = 4.3 ng/L
Blood system effects HH = 2.1 ng/L
Liver, kidney, heart, lung,
spleen, and adrenal gland
damage
Weight loss 2 to 100 MCL = 3 ug/L
Cardiovascular damage CMC4 = 350
Muscle and adrenal ug/L
degeneration CCC4 = 12 ug/L
Congestion of heart,
lungs, and kidneys
Toxic to aquatic plants
Mercury
• Waste disposal
• Manufacturing
processes
• Energy production
• Ore processing
• Municipal & medical
waste incinerators
• Chloralkali factories
• Fuel combustion
Battery cells
Barometers
Dental fillings
Thermometers
Switches
Fluorescent lamps
• Possible human 63,000 to
carcinogen 100,000
• Damage to brain and
kidneys
• Adverse affects on the
developing fetus, sperm,
and male reproductive
organs
MCL = 2 ug/L
CMC =1.4 ug/L
CCC = 0.77 ug/L
HH = 50 ng/L
FWA5 = 2.4 pg/L
FWC5 = 12ng/L
Wildlife6 =1.3
ng/L
1From: U.S. Environmental Protection Agency, 1995a, National Primary Drinking Water Regulations, Contaminant Specific
Fact Sheets, Inorganic Chemicals, Technical Version, EPA 811/F-95/002-T, USEPA, Office of Water, Washington, D.C.;
and U.S. Environmental Protection Agency, 1995b, National Primary Drinking Water Regulations, Contaminant Specific
Fact Sheets, Synthetic Organic Chemicals, Technical Version, EPA 811/F-95/003-T, USEPA, Office of Water,
Washington, D.C.
2MCL = Maximum Contaminant Level for drinking water. CMC = Criterion Maximum Concentration for protection of aquatic
life from acute toxicity. CCC = Criterion Continuous Concentration for protection of aquatic life from chronic toxicity. HH
= water quality criteria for protection of human health from water and fish consumption. Data from: U.S. Environmental
Protection Agency, 1999, National Recommended Water Quality Criteria-Correction, EPA 822/Z-99/001, USEPA, Office
of Water, Washington, D.C.
Characteristics presented are for chlordane. frans-Nonachlor is a principle component of the pesticide chlordane.
"Draft water quality criteria for protection of aquatic life. From: U.S. Environmental Protection Agency, 2001 b, Ambient
Aquatic Life Water Quality Criteria for Atrazine, USEPA, Office of Water, Washington, D.C.
5FWA = Freshwater acute water quality criterion. FWC = Freshwater chronic water quality criterion. From National Toxics
Rule (58 FR 60848).
6Wildlife criterion. From the Stay of Federal Water Quality Criteria for Metals (60 FR 22208), 40 CFR 131.36 and the
Water Quality Guidance for the Great Lakes System (40 CFR 132).
-------
prompting closure of some commercial fisheries and
issuance of fish consumption advisories. PCBs are
a probable human carcinogen, and human health
effects of PCBs exposure include stomach, kidney,
and liver damage; liver and biliary tract cancer; and
reproductive effects, including effects on the fetus
after exposure of the mother.
PCB congeners exhibit a wide range of physical and
chemical properties (e.g. vapor pressures,
solubilities, boiling points), are relatively resistant to
degradation, and are ubiquitous. These properties
make them ideal surrogates for a wide range of
organic compounds from anthropogenic sources.
1.1.3.1.2 Isomer trans-Nonachlor
The isomer frans-nonachlor is a component of the
pesticide chlordane. Chlordane is a mixture of
chlorinated hydrocarbons that was manufactured and
used as a pesticide from 1948 to 1988. Prior to
1983, approximately 3.6 million pounds of chlordane
were used annually in the United States. In 1988,
the USEPA banned all production and use of
chlordane in the United States.
Like PCBs, chlordane is relatively persistent and
bioaccumulative. The taans-nonachlor is the most
bioaccumulative of the chlordanes, and is a probable
human carcinogen. Other human health effects
include neurological effects, blood dyscrasia,
hepatoxicity, immunotoxicity, and endocrine system
disruption.
Historically, frans-nonachlor may have entered the
Great Lakes through a variety of means related to
the application of chlordane, including improper or
indiscriminate application, improper cleaning and
disposal of pesticide application equipment, or
contaminated soils washing into the watershed. In
the LMMBP, frans-nonachlor served as a model for
the cyclodiene pesticides.
1.1.3.1.3 Atrazine
Atrazine is a triazine herbicide based on a ring
structure with three carbon atoms alternating with
three nitrogen atoms. Atrazine is the most widely
used herbicide in the United States for corn and
sorghum production. Atrazine has been used as an
agricultural herbicide since 1959, and 64 to 75 million
pounds of atrazine are used annually in the United
States. Atrazine is extensively used in the upper
Midwest, including the Lake Michigan watershed,
where it is primarily associated with corn crops.
Unlike PCBs and fra/is-nonachlor, atrazine is not
bioaccumulative. It is persistent in water; however,
it is moderately susceptible to biodegradation in soils
with a half-life of about 60-150 days. Atrazine rarely
exceeds the 3 ppb maximum contaminant level
(MCL) set by the USEPA as a drinking water
standard, but localized peak values can exceed the
MCL following rainfall events after atrazine
application.
In January 31,2003, the U.S. EPA issued an Interim
Reregistration Eligibility Decision (IRED) for atrazine.
In an October 2003 addendum to the IRED, the
agency concluded that there is sufficient evidence to
formulate a hypothesis that atrazine exposure may
impact gonadal development in amphibians, but
there are currently insufficient data to either confirm
or refute the hypothesis. Based on available test
data, atrazine is not likely to be a human carcinogen.
The Agency does have concern in regards to the
potential hormonal effects observed in laboratory
animals exposed to atrazine. Above certain
concentration thresholds, atrazine is toxic to aquatic
plants. In the LMMBP, atrazine served as a model to
describe the transport and fate of a water-soluble
pesticide in current use.
1.1.3.1.4 Mercury
Mercury is a naturally-occurring toxic metal. Mercury
is used in battery cells, barometers, thermometers,
switches, fluorescent lamps, and as a catalyst in the
oxidation of organic compounds. Global releases of
mercury in the environment are both natural and
anthropogenic (caused by human activity). It is
estimated that about 11,000 metric tons of mercury
are released annually to the air, soil, and water from
anthropogenic sources. These sources include
combustion of various fuels such as coal; mining,
smelting and manufacturing activities; wastewater;
and agricultural, animal, and food wastes.
As an elemental metal, mercury is extremely
persistent in all media. Mercury also bioaccumulates
with reported bioconcentration factors in fish tissues
in the range of 63,000 to 100,000. Mercury is a
-------
possible human carcinogen and causes the following
human health effects: stomach, large intestine, brain,
lung, and kidney damage; blood pressure and heart
rate increase; and fetal damage. In the LMMBP,
mercury served as a model for bioaccumulative
metals.
1.1.3.2 Other Measured Parameters
In addition to the four chemicals modeled in the
LMMBP, many other chemicals and parameters were
measured in the LMMBP as part of the EMPs. A
survey of these chemicals and parameters aids in the
understanding of the overall ecological integrity of
Lake Michigan. These additional parameters include
various biological indicators; meteorological
parameters; and organic, metal, and conventional
chemicals in Lake Michigan. A complete listing of
parameters included in this study is provided in Table
1.1.2. A comprehensive listing of parameters and
compartments may be found in Chapter 3 (Appendix
1.3.1).
1.1.3.3 Measured Compartments
In the LMMBP, contaminants were measured in the
following compartments:
• Open Lake Water Column: The water column in
the open lake was sampled and analyzed for the
modeled pollutants.
• Tributaries: Major tributaries were sampled and
analyzed for the modeled pollutants.
• Fish: Top predators and forage base species
were sampled and analyzed for diet analysis and
contaminant burden.
• Lower Pelagic Food Chain: Phytoplankton and
zooplankton were sampled and analyzed for
species diversity, taxonomy, and contaminant
burden.
• Sediments: Cores were collected and trap
devices were used to collect sediment for
determination of contaminants and sedimentation
rates.
• Atmosphere: Vapor, particulate, and precipitation
phase samples were collected and analyzed for
the modeled pollutants.
For the modeled pollutants, more than 20,000
samples were collected at more than 300 sampling
locations and analyzed, including more than 9,000
quality control (QC) samples (Figure 1.1.2). Field
data collection activities were initially envisioned as
a one-year effort. However, it became evident early
into the project that a longer collection period would
be necessary to provide a full year of concurrent
information on contaminant loads and ambient
concentrations for modeling purposes. Therefore,
field sampling occurred from April 1994 to October
1995.
1.1.4 Objectives
The goal of the LMMBP was to develop a sound,
scientific base of information to guide future toxic
load reduction efforts at the Federal, State, Tribal,
and local levels. To meet this goal, the four following
LMMBP objectives were developed:
> Estimate pollutant loading rates: Environmental
sampling of major media will allow estimation of
relative loading rates of critical pollutants to the
Lake Michigan Basin.
»• Establish baseline: Environmental sampling and
estimated loading rates will establish a baseline
against which future progress and contaminant
reductions can be gauged.
- Predict benefits associated with load
reductions: The completed mass balance model
will provide a predictive tool that environmental
decision-makers and managers may use to
evaluate the benefits of specific load reduction
scenarios.
» Understand ecosystem dynamics: Information
from the extensive LMMBP monitoring and
modeling efforts will improve our scientific
understanding of the environmental processes
governing contaminant cycling and availability
within relatively closed ecosystems.
-------
Table 1.1.2. The LMMBP Parameters
Conventionals (Continued)
acenaphthene
acenaphthylene
aldrin
anthracene
atrazine
cc-BHC
P-BHC
5-BHC
Y-BHC
benzo[a]anthracene
benzo[g,/v]perylene
benzo[b]fluoranthene
benzo[/f]fluoranthene
benzo[e]pyrene
benzo[a]pyrene
a-chlordane
Y-chlordane
chrysene
coronene
p,p'-DDE
p,p'-DDD
Organics
p.p'-DDT
endosulfan sulfate
endosulfan I
endosulfan II
endrin
endrin aldehyde
endrin ketone
fluoranthene
fluorene
heptachlor
heptachlor epoxide
hexachlorobenzene (HCB)
indeno[1,2,3-cc(|pyrene
mirex
frans-nonachlor
oxychlordane
PCBs congeners
phenanthrene
pyrene
retene
toxaphene
Metals
aluminum
arsenic
calcium
cadmium
chromium
cesium
copper
iron
mercury
potassium
magnesium
manganese
sodium
nickel
lead
selenium
thorium
titanium
vanadium
zinc
Conventionals
alkalinity
ammonia
bromine
chloride
chlorine
sulfate
conductivity
dissolved organic
carbon
dissolved oxygen
dissolved phosphorus
dissolved reactive silica
particulate organic carbon
percent moisture
PH
phosphorus
silica
silicon
temperature
total Kjeldahl nitrogen
total organic carbon
total phosphorus
total suspended
particulates
dry weight fraction
element carbon
nitrate
ortho-phosphorus
total hardness
turbidity
Biologicals
fish species
fish age
fish maturity
chlorophyll a
fish lipid amount
zooplankton
fish weight
fish length
fish taxonomy
fish diet analysis
primary productivity
Meteorological
air temperature
relative humidity
barometric pressure
weather conditions
wind direction
wind speed
visibility
wave height and direction
1.1.5 Design
1.1.5.1 Organization
The GLNPO proposed a mass balance approach to
provide coherent, ecosystem-based evaluation of
toxics in Lake Michigan. GLNPO served as the
program sponsor for the LMMBP. GLNPO formed
two committees to coordinate study planning, the
Program Steering Committee and the Technical
Coordinating Committee. These committees were
comprised of Federal, State, and academic
laboratories as well as commercial laboratories (see
Section 1.1.5.2, Study Participants). The committees
administered a wide variety of tasks including:
planning the project, locating the funding, designing
the sample collection, coordinating sample collection
activities, locating qualified laboratories, coordinating
analytical activities, assembling the data, assuring
the quality of the data, assembling skilled modelers,
developing the models, and communicating interim
and final project results. The Mid-Continent Ecology
Division (MED) at Duluth, in cooperation with the
National Oceanic and Atmospheric Administration
(NOAA) Great Lakes Environmental Research
Laboratory (GLERL) and the Atmospheric Sciences
Modeling Division, supported the modeling
component of the mass balance study by developing
a suite of integrated mass balance models to
-------
Menominee
£Bfewi
Manistique
.e&ktete^
n atmospheric monitoring
» stations
^ sediment samples
Q water survey stations
tributary monitoring
stations
monitored tributary
basins
unmonitored tributary
basins
biota survey boxes
Grand Calumet
St. Joseph
Figure 1.1.2. The LMMBP sampling locations.
8
-------
simulate the transport, fate, and bioaccumulation of
the study target analytes.
1.1.5.2 Study Participants
The LMMBP was a coordinated effort among
Federal, State, and academic scientists; and
commercial laboratories. The following agencies and
organizations have all played roles in ensuring the
success of the LMMBP. Except for the three
organizations indicated with an asterisk (*), all of the
participants were members of the LMMBP Steering
Committee.
Federal and International
- USEPA GLNPO (Program Sponsor)
» USEPA Region V Water Division (WD)
> USEPA Region V Air Division
•> USEPA/ORD/NHEERL/MED/LLRFRB
> ORD National Exposure Research Laboratory
•• U.S. Department of Interior (USDOI) U.S.
Geological Survey (USGS) Water Resources
Division (WRD)
»• USDOI/USGS Biological Resources Division Great
Lakes Science Center (GLSC)
- U.S. Fish and Wildlife Service (USFWS)
* U.S. Department of Energy
" U.S. Department of Commerce NOAA/GLERL
- USEPA Office of Air and Radiation*
> USEPA Office of Water*
>• Environment Canada*
>• U.S. Department of Energy Battelle NW
State
" Illinois Department of Natural Resources
>• Illinois Water Survey
*• Indiana Department of Environmental
Management
*• Michigan Department of Natural Resources
*• Michigan Department of Environmental Quality
(MDEQ)
" Wisconsin Department of Natural Resources
>• Wisconsin State Lab of Hygiene
Academic and Commercial
»• Indiana University
*• Rutgers University
>• University of Maryland
* University of Michigan
*• University of Minnesota
*• University of Wisconsin
»• Grace Analytical
1.1.5.3 Workgroups
Eleven workgroups were formed to provide oversight
and management of specific project elements. The
workgroups facilitated planning and implementation
of the study in a coordinated and systematic fashion.
The workgroups communicated regularly through
participation in monthly conference calls and annual
"all-hands" meetings. Workgroup chairs were
selected and were responsible for managing tasks
under the purview of the workgroup and
communicating the status of activities to other
workgroups. The workgroups and workgroup chairs
are listed below.
• Program Steering Committee - Paul Horvatin
(USEPA/GLNPO)
• Technical Coordinating Committee-Paul Horvatin
(USEPA/GLNPO)
• Modeling Workgroup - William Richardson
(USEPA/ORD/NHEERL/MED/LLRFRB)
• Air Monitoring Workgroup - Jackie Bode
(USEPA/GLNPO)
• Biota Workgroup - Paul Bertram (USEPA/
GLNPO)andJohnGannon(USDOI/USGS/GLSC)
• Chemistry Workgroup - David Anderson
(USEPA/GLNPO)
• Data Management Workgroup - Kenneth Klewin
and Philip Strobel (USEPA/GLNPO)
• Lake Monitoring Workgroup - Glenn Warren
(USEPA/GLNPO)
• Tributary Monitoring Workgroup - Gary Kohlhepp
(USEPA/Region V/WD) and Robert Day (MDEQ)
• Quality Assurance Workgroup - Louis Blume and
Michael Papp (USEPA/GLNPO)
• Sediment Monitoring Workgroup - Brian Eadie
(NOAA/GLERL)
1.1.5.4 Information Management
As program sponsor, GLNPO managed information
collected during the LMMBP. Principal Investigators
(Pis) participating in the study reported field and
analytical data to GLNPO. GLNPO developed a data
standard for reporting field and analytical data and a
database for storing and retrieving study data.
-------
GLNPO was also responsible for conducting data
verification activities and releasing verified data to the
study modelers and the public. The flow of
information is illustrated in Figure 1.1.3.
1.1.5.4.1 Data Reporting
Over 20 organizations produced LMMBP data
through the collection and analysis of more than
20,000 samples. In the interest of standardization,
specific formats (i.e., file formats and codes to
represent certain data values) were established for
reporting the LMMBP data. Each format specified
the "rules" by which data were submitted, and, in
many cases, the allowable values by which they were
to be reported. The data reporting formats were
designed to minimize the number of data elements
reported from the field crews and laboratory analysis.
Data reporting formats and the resulting Great Lakes
Environmental Monitoring Database (GLENDA, see
Sectionl .1.5.4.2) were designed to be applicable to
projects outside the LMMBP as well.
Principal Investigators (including sampling crews and
the analytical laboratories) supplied sample collection
and analysis data following the standardized
reporting formats, if possible. The LMMBP data were
then processed through an automated SAS-based
data verification system, Research Data
Management and Quality Control System (RDMQ),
for quality assurance (QA)/QC checking. After
verification and validation by the PI, the data sets
were output in a form specific for upload to GLENDA.
Finally, these data sets were uploaded to GLENDA.
Principal
Investigator (PI)
Collect and Analyze
Samples
I
Report Field and
Analytical Data
(According to LMMB Data
Standard)
A
No
II
^^~- -___
*\.
GLNPO Data
Management
Workgroup
Receive, Store, and
Transmit Data
^ ^^
•^
Store, Transmit, and
GLENDA
s~
Fulfill Request
data)
1
-^
— *
GLNPO QA
Workgroup
Conduct Data
Verification (Merge Field
and Analytical Data using
RDMQ)
1
<5a1a^r7^e>>
<\!^!^>
No
Produce final verified
data file and provide to
PI for review and
approval
Produce Final Verified
Data File and Transmit
(in GtENDA-compatible
Format)
1
LMMB Study
Modelers
^ Input Data to Study
* Models
External Parties
Roquo-t
A
Figure 1.1.3. Flow of information in the LMMBP.
10
-------
1.1.5.4.2 Great Lakes Environmental Monitoring
Database
Central to the data management effort is a
computerized database system to house LMMBP
and other project results. That system, the Great
Lakes Environmental Monitoring Database
(GLENDA), was developed to provide data entry,
storage, access, and analysis capabilities to meet the
needs of mass balance modelers and other potential
users of Great Lakes data.
Development of GLENDA began in 1993 with a
logical model based on the modernized STORET
concept and requirements analysis. GLENDA was
developed with the following guiding principles:
• True multi-media scope - Water, air, sediment,
taxonomy, fish tissue, fish diet, and meteorology
data can all be housed in the database.
• Data of documented quality - Data quality is
documented by including results of quality control
parameters.
• Extensive contextual indicators - Ensure data
longevity by including enough information to allow
future or secondary users to make use of the data.
• Flexible and expandable - Database is able to
accept data from any Great Lakes monitoring
project.
• National compatibility - GLENDA is compatible
with STORET and allows ease of transfer between
these large databases.
In an effort to reduce the data administration burden
and ensure consistency of data in this database,
GLNPO developed several key tools. Features
including standard data definitions, reference tables,
standard automated data entry applications, and
analytical tools are (or will soon be) available.
1.1.5.4.3 Public Access to LMMBP Data
All LMMBP data that have been verified (through the
QC process) and validated (accepted by the PI) are
available to the public. Currently, GLNPO requires
that written requests be made to obtain the LMMBP
data. The data sets are available in several formats
including WK1, DBF, and SD2.
The primary reason for requiring an official request
form for the LMMBP data is to keep track of
requests. This allows GLNPO to know how many
requests have been made, who has requested data,
and what use they intend for the data. This
information assists GLNPO in managing and
providing public access to Great Lakes data and
conducting public outreach activities. In the future,
after all data are verified and validated, GLNPO
intends to make condensed versions of the data sets
available on the LMMBP web site for downloading.
This will allow easy public access to the LMMBP
data.
Further information on the information management
for the LMMBP can be found in The Lake Michigan
Mass Balance Study Quality Assurance Report (U.S.
Environmental Protection Agency, 2001 a).
1.1.5.5 Quality Assurance Program
At the outset of the LMMBP, managers recognized
that the data gathered and the models developed
from the study would be used extensively by
decision-makers responsible for making
environmental, economic, and policy decisions.
Environmental measurements are never true values
and always contain some level of uncertainty.
Decision-makers, therefore, must recognize and be
sufficiently comfortable with the uncertainty
associated with data on which their decisions are
based. In recognition of this requirement, the
LMMBP managers established a QA program goal
ofensuring that data produced under the LMMBP
would meet defined standards of quality with a
specified level of confidence.
The QA program prescribed minimum standards to
which all organizations collecting data were required
to adhere. Data quality was defined, controlled, and
assessed through activities implemented within
various parameter groups (e.g., organic, inorganic,
and biological parameters). QA activities included
the following:
11
-------
•• QA Program - Prior to initiating data collection
activities, plans were developed, discussed, and
refined to ensure that study objectives were
adequately defined and to ensure that all QA
activities necessary to meet study objectives were
considered and implemented.
•> QA Workgroup - USEPA established a QA
Workgroup whose primary function was to ensure
that the overall QA goals of the study were met.
•• QA Project Plans (QAPPs) - USEPA worked
with Pis to define program objectives, data quality
objectives (DQOs), and measurement quality
objectives (MQOs) for use in preparing Quality
Assurance Project Plans (QAPPs). Principal
investigators submitted QAPPs to the USEPA for
review and approval. USEPA reviewed each
QAPPfor required QA elements and soundness of
planned QA activities.
> Training - Before beginning data collection
activities, Pis conducted training sessions to
ensure that individuals working on the project were
capable of properly performing data collection
activities for the LMMBP.
«• Monthly Conference Calls and Annual
Meetings - USEPA, Pis, and support contractors
participated in monthly conference calls and
annual meetings to discuss project status and
objectives, QA issues, data reporting issues, and
project schedules.
*• Standardized Data Reporting Format-Pis were
required to submit all data in a standardized data
reporting format that was designed to ensure
consistency in reporting and facilitate data
verification, data validation, and database
development.
>• Intercomparison Studies - USEPA conducted
studies to compare performance among different
Pis analyzing similar samples. The studies were
used to evaluate the comparability and accuracy
of program data.
* Technical Systems Audits - During the study,
USEPA formally audited each Pi's laboratory for
compliance with their QAPPs, the overall study
objectives, and pre-determined standards of good
laboratory practice.
+ Data Verification - Pis and the USEPA evaluated
project data against pre-determined MQOs and
DQOs to ensure that only data of acceptable
quality would be included in the program
database.
»• Statistical Assessments - USEPA made
statistical assessments of the LMMBP data to
estimate elements of precision, bias, and
uncertainty.
*• Data Validation - USEPA and modelers
evaluated the data against the model objectives.
Comparability of data among Pis participating in the
LMMBP was deemed to be important for successful
completion of the study. Therefore, MQOs for
several data attributes were developed by the Pis
and defined in the QAPPs. MQOs were designed to
control various phases of the measurement process
and to ensure that the total measurement uncertainty
was within the ranges prescribed by the DQOs.
MQOs were defined in terms of six attributes:
* Sensitivity/Detectability - The determination of
the low-range critical value that a method-specific
procedure can reliably discern for a given
pollutant. Sensitivity measures included, among
others, method detection limits (MDLs) as defined
in 40 CFR Part 136, system detection limits
(SDLs), or instrument detection limits (IDLs).
*• Precision - A measure of the degree to which
data generated from replicate or repetitive
measurements differ from one another. Analysis
of duplicate samples was used to assess
precision.
»• Bias - The degree of agreement between a
measured and actual value. Bias was expressed
in terms of the recovery of an appropriate
standard reference material or spiked sample.
* Completeness - The measure of the number of
samples successfully analyzed and reported
compared to the number that were scheduled to
be collected.
12
-------
•• Comparability - The confidence with which one
data set can be compared to other data sets.
- Representativeness - The degree to which data
accurately and precisely represent characteristics
of a population, parameter variations at a
sampling point, a process condition, or an
environmental condition.
The Pi-defined MQOs also were used as the basis
for the data verification process. GLNPO conducted
data verification through the LMMBP QA Workgroup.
The workgroup was chaired by GLNPO's Quality
Assurance Manager and consisted of QC
Coordinators that were responsible for conducting
review of specific data sets. Data verification was
performed by comparing all field and QC sample
results produced by each PI with their MQOs and
with overall LMMBP objectives. If a result failed to
meet predefined criteria, the QC Coordinator
contacted the PI to discuss the result, verify that it
was correctly reported, and determine if corrective
actions were feasible. If the result was correctly
reported and corrective actions were not feasible, the
results were flagged to inform data users of the
failure. These flags were not intended to suggest
that data were not useable; rather they were intended
to caution the user about an aspect of the data that
did not meet the predefined criteria. Data that met all
predefined requirements were flagged to indicate that
the results had been verified and were determined to
meet applicable MQOs. In this way, every data point
was assigned one or more validity flags based on the
results of the QC checks. GLNPO also derived data
quality assessments for each LMMBP data set for a
subset of the attributes listed above, specifically
sensitivity, precision, and bias. The LMMBP
modelers and the LLRS Database Manager also
performed data quality assessments prior to inputting
data into study models. Such activities included
verifying the readability of electronic files, identifying
missing data, checking units, and identifying outliers.
A detailed description of the QA program is included
in The Lake Michigan Mass Balance Project Quality
Assurance Report (U.S. Environmental Protection
Agency, 2001 a). A brief summary of quality
implementation and assessment is provided in each
of the following parts.
1.1.6 Project Documents and Products
During project planning, LMMBP participants
developed study tools including work plans, a
methods compendium, QAPPs, and data reporting
standards. Through these tools, LMMBP participants
documented many aspects of the study including
information management and QA procedures. Many
of these documents are available on GLNPO's
website at http://www.epa.gov/glnpo/lmmb.
The LMMBP Work Plan
Designers of the LMMBP have documented their
approach in a report entitled Lake Michigan Mass
Budget/Mass Balance Work Plan (U.S.
Environmental Protection Agency, 1997a). The
essential elements of a mass balance study and the
approach used to measure and model these
elements in the Lake Michigan system are described
in the work plan. This document was developed
based upon the efforts of many Federal and State
scientists and staff who participated in the initial
planning workshop, as well as Pis.
QA Program/Project Plans
The Lake Michigan Mass Balance Project: Quality
Assurance Plan for Mathematical Modeling, Version
3.0 (Richardson et al., 2004) documents the quality
assurance process for the development and
application of LMMBP models, including
hydrodynamic, sediment transport, eutrophication,
transport chemical fate, and food chain
bioaccumulation models.
The Enhanced Monitoring Program QA Program
Plan
The Enhanced Monitoring Program Quality
Assurance Program Plan (U.S. Environmental
Protection Agency, 1997b) was developed in 1993 to
ensure that data generated from the LMMBP
supported its intended use.
The LMMBP Methods Compendium
The Lake Michigan Mass Balance Project Methods
Compendium (U.S. Environmental Protection
Agency, 1997c, 1997d) describes the sampling and
analytical methods used in the LMMBP- The entire
13
-------
three volumes are available on GLNPO's website
mentioned above.
The LMMBP Data Reporting Formats and Data
Administration Plan
Data management for the LMMBP was a focus from
the planning stage through data collection,
verification, validation, reporting, and archiving. The
goal of consistent and compatible data was a key to
the success of the project. The goal was met
primarily through the development of standard
formats for reporting environmental data. The data
management philosophy is outlined on the LMMBP
website mentioned above.
Lake Michigan LaMP
"Annex 2" of the 1972 Canadian-American Great
Lakes Water Quality Agreement (amended in 1978,
1983, and 1987) prompted development of a Lake-
wide Area Management Plan (LaMP) for each Great
Lake. The purpose of these LaMPs is to document
an approach to reducing input of critical pollutants to
the Great Lakes and restoring and maintaining Great
Lakes integrity. The Lake Michigan LaMP calls for
basin-wide management of toxic chemicals.
GLENDA Database
Central to the data management effort is a
computerized data system to house LMMBP and
other project results. That system, the Great Lakes
Environmental Monitoring Database (GLENDA), was
developed to provide, data entry, storage, access,
and analysis capabilities to meet the needs of mass
balance modelers and other potential users of Great
Lakes data.
References
Richardson, W.L., D.D. Endicott, R.G. Kreis, Jr., and
K.R. Rygwelski (Eds.). 2004. The Lake Michigan
Mass Balance Project Quality Assurance Plan for
Mathematical Modeling. Prepared by the
Modeling Workgroup. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
EPA/600/R-04/018, 233 pp.
U.S. Environmental Protection Agency. 1995a.
National Primary Drinking Water Regulations,
Contaminant Specific Fact Sheets, Inorganic
Chemicals, Technical Version. U.S.
Environmental Protection Agency, Office of
Water, Washington, D.C. EPA/811/F-95/002-T.
U.S. Environmental Protection Agency. 1995b.
National Primary Drinking Water Regulations,
Contaminant Specific Fact Sheets, Synthetic
Organic Chemicals, Technical Version. U.S.
Environmental Protection Agency, Office of
Water, Washington, D.C. EPA/811/F-95/003-T.
U.S. Environmental Protection Agency. 1997a. Lake
Michigan Mass Budget/Mass Balance Work Plan.
U.S. Environmental Protection Agency, Great
Lakes National Program Office, Chicago, Illinois.
EPA/905/R-97/018, 155 pp.
U.S. Environmental Protection Agency. 1997b. The
Enhanced Monitoring Program Quality Assurance
Program Plan. U.S. Environmental Protection
Agency, Great Lakes National Program Office,
Chicago, Illinois. EPA/905/R-97/017, 61 pp.
U.S. Environmental Protection Agency. 1997c. Lake
Michigan Mass Balance Study (LMMB) Methods
Compendium, Volume 1: Sample Collection
Techniques. U.S. Environmental Protection
Agency, Great Lakes National Program Office,
Chicago, Illinois. EPA/905/R-97/012a, 1,440pp.
U.S. Environmental Protection Agency. 1997d. Lake
Michigan Mass Balance Study (LMMB) Methods
Compendium, Volume 2: Organic and Mercury
Sample Analysis Techniques. U.S.
Environmental Protection Agency, Great Lakes
National Program Office, Chicago, Illinois.
EPA/905/R-97/012b, 532 pp.
U.S. Environmental Protection Agency. 1999.
National Recommended Water Quality Criteria-
Correction. U.S. Environmental Protection
Agency, Office of Water, Washington, D.C.
EPA/822/Z-99/001, 25 pp.
14
-------
U.S. Environmental Protection Agency. 2001 a. The U.S. Environmental Protection Agency. 2001 b.
Lake Michigan Mass Balance Study Quality Ambient Aquatic Life Water Quality for Atrazine.
Assurance Report. U.S. Environmental U.S. Environmental Protection Agency, Office of
Protection Agency, Great Lakes National Water, Washington, D.C. EPA/822/D-01/002,
Program, Chicago, Illinois. EPA/905/R-01/013. 230pp.
15
-------
PART1
INTRODUCTION
Chapter 2. PCBs Modeling Overview
Douglas D. Endicott
Great Lakes Environmental Center
Traverse City, Michigan
and
William L. Richardson
Retired
and
Ronald Rossmann
U.S. Environmental Protection Agency
Office of Research and Development
National Health and Environmental Effects
Research Laboratory
Mid-Continent Ecology Division
Large Lakes and Rivers Forecasting Research
Branch
Large Lakes Research Station
9311 Groh Road
Grosse lie, Michigan 48138
1.2.1 Background
The mass balance project was based upon the
Enhanced Monitoring Program (EMP), a
comprehensive, two-year synoptic survey for
selected toxic chemicals in the Lake Michigan
ecosystem. The EMP included tributary load and
atmospheric deposition monitoring; ambient water
column, biota, and sediment sampling; and additional
measurements to define and confirm transport and
fate processes. The toxics studied for the Lake
Michigan Mass Balance Project (LMMBP) included
polychlorinated biphenyls (PCBs), atrazine, trans-
nonachlor, and mercury. The project was led by the
United States Environmental Protection Agency
(USEPA)/Great Lakes National Program Office
(GLNPO). Modeling support to the project was
provided by the USEPA/Mid-Continent Ecology
Division (MED)/Office of Research and Development
(ORD)/Large Lakes Research Station (LLRS) in
cooperation with the Atmospheric Research and
Exposure Assessment Laboratory (AREAL); the
National Oceanic and Atmospheric Administration
(NOAA)/Great Lakes Environmental Research
Laboratory (GLERL); and other cooperators. The
research developed a suite of integrated mass
balance models to simulate the transport, fate, and
bioaccumulation of toxic chemicals in Lake Michigan.
1.2.2 Modeling Objectives
Development of effective strategies for toxics
management requires a quantitative understanding
of the relationships between sources, inventories,
concentrations, and effects of contaminants in the
ecosystem. A mass balance modeling approach was
used to address the relationship between sources of
toxic chemicals and concentrations in air, water,
sediment, and biota. This approach integrated load
estimation, ambient monitoring, and research efforts
within a modeling framework that was compatible
with both scientific as well as ecosystem
management objectives. The mass balance
approach estimated the magnitude of mass fluxes
that constitute the pathways for toxics transport into
and out of the lake, that distribute toxics within the
lake water column and sediment, and that lead to
bioaccumulation of the aquatic food webs. Based
upon these estimates, the mass balance was used to
determine the rate of change in concentrations and
inventories of toxics as inputs such as atmospheric
16
-------
and tributary loadings change. Thus, the mass
balance can serve as a useful tool to estimate or
predict the outcome of alternatives under
consideration for toxics management.
Modeling efforts associated with the LMMBP had the
following objectives:
1. Provide a consistent framework for integrating
load estimates, ambient monitoring data, process
research efforts, and prior modeling efforts,
leading to a better understanding of toxic
chemical sources, transport, fate, and
bioaccumulation in Lake Michigan.
2. Estimate the loading of priority toxics, solids, and
nutrients from all major tributaries to Lake
Michigan for the duration of the study.
3. Estimate the atmospheric deposition and air-
water exchange of priority toxics, including spatial
and temporal variability over Lake Michigan.
4. Calibrate and confirm mass balance models for
priority toxics using project data, based upon
modelsforhydrodynamicandsedimenttransport,
eutrophication/organic carbon dynamics, toxics
transport and fate, and food web
bioaccumulation.
5. Based upon the mass balance models, evaluate
the magnitude and variability of toxic chemical
fluxes within and between lake compartments,
especially between the sediment and water
column and between the water column and the
atmosphere.
6. Apply the calibrated mass balance models to
forecast contaminant concentrations in water and
sediment throughout Lake Michigan, based upon
meteorological forcing functions and future
loadings based upon load reduction alternatives.
7. Predict the bioaccumulation of persistent toxic
chemicals through the food web leading to top
predator fish (lake trout and coho salmon) for
location-specific fish populations in the lake, in
order to relate mass balance predictions of water
and sediment exposure to this significant
impaired use.
8. Estimate (quantify) the uncertainty associated
with estimates of tributary and atmospheric loads
of priority toxics, and model predictions of
contaminant concentrations.
9. Identify and prioritize further monitoring,
modeling, and research efforts to (1) address
additional toxic substances, (2) further reduce
uncertainty and improve accuracy of predictions,
(3) establish additional cause-effect linkages,
such as ecological risk endpoints and feedbacks,
and (4) evaluate additional source categories,
such as non-point sources in the watershed.
The purpose of PCBs modeling was to simulate their
transport, fate, and bioaccumulation in Lake
Michigan. PCBs are a group of persistent,
bioaccumulative hydrophobic organic chemicals
(HOCs) that are ubiquitous in the Great Lakes.
Although anthropogenic inputs from production and
disposal largely ceased following their ban in the
1970s,atmosphericandwatershedtributarytransport
pathways to the lake continue the import of PCBs. In
addition, a large in-lake sediment inventory
represents an internal source of PCBs, which are
recycled annually. PCBs have been consistently
identified as the contaminant of greatest concern to
human and ecosystem health in the Great Lakes
(Ludwig et a/., 1993; Swain, 1991; Gilbertson et al.,
1991).
1.2.3 Historical Modeling
The modeling design and approach for the LMMBP
reflects a progression of prior modeling efforts in
Lake Michigan and throughout the Great Lakes.
These include eutrophication and toxic substance
mass balance models, food web bioaccumulation
models, and predictive hydrodynamic and sediment
transport models. Although not a comprehensive
review, several of these prior modeling efforts are
discussed below.
1.2.3.1 Lake-1
A eutrophication model for Lake Michigan was
developed by Rodgers and Salisbury (1981), based
upon the Lake-1 model which was also applied to
Lakes Erie, Huron, and Ontario. The model was
calibrated and tested using data from 1976 and 1977.
17
-------
The importance of climatic factors on limnological
(including eutrophication) processes in Lake
Michigan was demonstrated, as the severe winter
and extensive ice cover of 1976-1977 dramatically
reduced total phosphorus concentrations in the
second year. This work also identified several
refinements necessary for accurate modeling of
eutrophication: phosphorus availability to
phytoplankton and particle transport including
shoreline erosion and sediment resuspension were
apparently significant influences upon nutrient and
phytoplankton dynamics observed in Lake Michigan.
1.2.3.2 Completely-Mixed Model
A lakes-in-series model for conservative substances
was developed by Sonzogni et al. (1983) and applied
to forecast chloride concentrations in each of the
Great Lakes as a function of expected future
loadings. This model demonstrated that
concentrations of non-reactive substances would
substantially "lag" the history of their input. This was
especially the case for Lake Michigan, where
maximum chloride concentrations were not predicted
to occur until the 22nd Century despite declining
loads after the 1970s. Similarly strong, non-steady-
state behavior may be expected for other chemicals
which are non-reactive and weakly associated to
particles.
1.2.3.3 General Mass Balance Framework for
Toxic Chemicals in the Great Lakes
At about the same time, models were being
developed which would serve as the foundation for
describing and simulating the transport and fate of
hydrophobic chemicals in the Great Lakes. Thomann
and Di Toro (1983) and Robbins (1985)
demonstrated that the lake-wide, annual
concentration trend of contaminants including
cesium-137, plutonium-239/240, and PCBs, were
dependent upon particle transport between the water
column and a resuspendable sediment compartment.
The principal loss mechanisms from the lakes were
found to be burial by sedimentation and (for PCBs)
volatilization. The somewhat paradoxical behavior of
these models was that the water column contaminant
dynamics were largely controlled by sediment
parameters.
1.2.3.4 Food Web Bioaccumulation Model
A food web bioaccumulation model was developed
by Thomann and Connolly (1984) and applied to
simulate bioaccumulation of PCBs in Lake Michigan
lake trout. The model was confirmed with an
extensive data set collected in 1971, including nine
age classes of trout, diet characterization by gut
contents analysis, and alewife. The model was
successful in predicting bioaccumulation for mature
age classes of lake trout, although not for juveniles.
Dietary transfer was demonstrated to be the
predominant route of PCBs accumulation, in
comparison to direct chemical uptake from water.
Substantial residual variance in lake trout PCBs
concentrations within age class CV = 1 was not
explained by this lake-wide, average-individual
model.
1.2.3.5 MICHTOX
An integrated mass balance and bioaccumulation
model for PCBs (modeled as two homologs) and 10
other toxic chemicals was developed as a planning
tool for the LMMBP (Endicott et at., 2005). The
MICHTOX mass balance was calibrated to
suspended solids and plutonium data for the
southern lake basin, while the bioaccumulation model
combined Thomann and Connolly's (1984) effort with
chemical-specific parameterization from Lake
Ontario. MICHTOX demonstrated that reasonable
predictions of PCB concentration trends in water,
sediment, and biota could be developed although
significant uncertainties regarding sediment-water
and air-water contaminant transport remained.
These are the most significant transport fluxes for
PCBs and presumably other hydrophobic
contaminants. Major data gaps for other priority
toxics allowed only order-of-magnitude estimates of
load-concentration relationships. When this model
was developed and run, available monitoring data for
toxic chemical concentration in tributaries, air, lake
water, sediment, and biota were not adequate to
define loading trends or to relate the distribution of
loadings to contaminant gradients observed for
sediment and biota. Credible model predictions of
toxic chemical transport, fate, and bioaccumulation
would depend upon developing a comprehensive
data set quantifying loadings, sediment inventories,
18
-------
concentrations, and transport fluxes on a spatially-
resolved basis, and localized descriptions of food
web structures.
1.2.3.6 Green Bay Mass Balance Project
The Green Bay Mass Balance Project (GBMBP)
demonstrated the feasibility of applying mass
balance principles to manage toxic chemicals in the
Great Lakes ecosystem. A two-year (1989-1990)
synoptic sampling program was designed to collect
appropriate and complete data for the mass balance
study. A suite of integrated mass balance and
bioaccumulation models were developed which,
together, provided an ecosystem-level simulation of
sources, transport, fate, and bioaccumulation of
PCBs throughout the Fox River and Green Bay. This
study advanced the state-of-the-art of mass balance
modeling, particularly the ability to construct a fairly
complete and accurate description of contaminant
mass transport.
Several aspects of the Green Bay modeling effort
were noteworthy. Particle transport and sorption
processes were found to be of fundamental
importance as bases for contaminant modeling.
Resuspension of contaminated sediments in the Fox
River constituted the major source of PCBs to the
river as well as the bay. In the bay, particle sorbent
dynamics were strongly affected by phytoplankton
production and decay. The relative significance of
hydraulic transport, sediment transport, burial,
volatilization, and open lake boundary exchange
processes upon the PCBs mass balance varied
considerably with location in Green Bay.
Radionuclide tracers were again essential for
calibration of particle fluxes and confirmation of long-
term contaminant transport predictions. The
significance of contaminant accumulation at the base
of the food web, and fish movement in relation to
exposure gradients, were demonstrated in the
bioaccumulation model. The LMMBP demonstrated
the linked submodel approach to ecosystem model
development and application, and the feasibility of
using such a model for assessing the effectiveness
of toxics management control alternatives.
1.2.3.7 SEDZL
The GBMBP also provided data to test a predictive
two-dimensional, hydrodynamic and sediment
transport model of the Fox River, SEDZL (Gailani et
al., 1991). SEDZL incorporates realistic descriptions
of cohesive sediment resuspension, flocculation, and
deposition processes, and contaminant sorption,
which are critical for accurate prediction of
hydrophobic contaminant transport. These process
descriptions were based on laboratory and field
experiments with river, bay, and lake sediments. A
three-dimensional bed submodel was used to
describe sediment bed properties which varied with
depth as well as location. The fine spatial resolution
of the model allowed detailed simulation of in-p!ace
pollutant transport in both the water column and
sediment bed. Although computationally intensive
and requiring specialized data, SEDZL has
substantially advanced the state-of-the-art for
sediment and contaminant transport modeling in the
Great Lakes. SEDZL had also been applied to the
Buffalo and Saginaw Rivers as part of the
ARCS/RAM project (Gailani et al., 1994; Cardenas
and Lick, 1996). These applications included long-
term forecasts (10-25 years) of sediment and
contaminant transport. SEDZL had also been
applied to large water bodies such as Lake Erie, and
marine coastal waters including Santa Barbara
Channel, and Atchafalaya Bay (Lick et al., 1994;
Pickens, 1992) where wave action as well as
currents force sediment resuspension.
1.2.4 Model Resolution
Model resolution is the spatial and temporal scale of
predictions, as well as the definitions of model state
variables. While factors such as data availability,
model sophistication, and computer resources
constrain resolution to a degree, different levels of
model resolution are possible and, are in fact,
necessary. Three "levels" of spatial resolution,
indicated by the segmentation grid of the lake
surface, are illustrated in Figure 1.2.1. Level 1 was
resolved at the scale of lake basins (characteristic
length, L = 150 km), with an associated seasonal
temporal resolution. This was a screening-level
model resolution used in MICHTOX. Level 2 was
19
-------
LEVEL 1 - MICHTOX
(Screening)
6 surface segments
9 water segments
LEVEL 2 - LM-2
10 surface segments
41 water segments
LEVEL 3 - LM-3
(High resolution 5km X 5km grid)
2318 surface segments
44,042 water segments
19 sigma layers
Figure 1.2.1. Surface water segmentation for alternative Lake Michigan mass balance model levels.
resolved at a regional scale defined by food webs (L
= 40 km) including gross resolution of the nearshore
and offshore regions; temporal resolution was
weekly-to-monthly. This resolution was roughly
comparable to that achieved by models developed in
the GBMBP. Level 3 was a hydrodynamic scale
resolution (L = 5 km), with associated daily temporal
resolution. Level 3 was scaled to resolve and predict
particle transport processes as well as hydrodynamic
transport.
Although the Lake-wide Management Plan (LaMP)
and the Great Waters Program objectives are "lake-
wide," both of these emphasize biotic impairments
occurring primarily in localized, nearshore regions.
LaMP objectives also require that the transport of
contaminants from tributaries and other nearshore
sources to the open lake be resolved. Therefore, the
Level 1 model was not adequate for the study
objectives. Level 2 resolution was adequate for most
modeling objectives, but not for resolution of
significant hydrodynamic and sediment transport
events. Level 3 resolution was required for accurate
hydrodynamic and sediment transport modeling and
was desirable for predicting nearshore gradients,
especially those formed by transients such as
thermal bars, upwelling, and storm-induced
resuspension; as well as more persistent features
such as tributary plumes, thermal stratification, and
the benthic nepheloid layer. Level 3 transport
resolution was also valuable in relating toxics loading
from the 10 Areas of Concern (AOCs) adjoining Lake
Michigan, which must be addressed by the Remedial
Action Plan (RAP) process, to the LaMP via the
LMMBP.
20
-------
The modeling design for the LMMBP was based on
the development of a number of models at three
levels of resolution. For the contaminant transport
and fate (CTF) models, MICHTOX was resolved at
Level 1, and LM2-Toxic was resolved at Level 2. For
the eutrophication models, MICH1 was resolved at
Level 1 and LM3-Eutro was resolved at Level 3. The
Princeton Ocean Model (POM) and atmospheric
loading models were resolved at Level 3. Results of
the hydrodynamic model were spatially and
temporally averaged prior to coupling to the Level 2
model. The rationale for specifying different
resolutions was the hydrodynamic models require a
Level 3 resolution to offer the best capability for
transport simulation and forecasting. A lower
resolution was specified for LM2-Toxic because this
model had been demonstrated at this resolution.
1.2.5 Models Developed and Applied
The model design for the LMMBP was based upon
the linked submodel approach used in the GBMBP.
Models developed, refined, and applied by the Large
Lakes and Rivers Forecasting Research Branch
(LLRFRB) included eutrophication/sorbent dynamics
(MICH1 and LM3-Eutro), contaminant transport and
fate (MICHTOX and LM2-Toxic), and food web
bioaccumulation (LM Food Chain) models (Figure
1.2.2). Models developed and run elsewhere
included a hydrodynamics model (POM) (Schwab
and Beletsky, 1998), an atmospheric loading model
(Green et al., 2000; Miller et al., 2001), and a
tributary loading model (Hall and Robertson, 1998).
Only the models developed, refined, and applied at
LLRFRB will be discussed in detail within this
document.
1.2.5.1 Lake Process Models
The mass balance for toxics in Lake Michigan was
comprised of linked hydrodynamic (POM),
eutrophication/sorbent dynamics (LM3-Eutro),
contaminant transport and transformation (LM2-
Toxic), and bioaccumulation simulations (LM Food
Chain). In addition, Level 1 eutrophication/sorbent
dynamics (MICH1) and contaminant transport and
transformation/bioaccumulation (MICHTOX)
simulations were run for comparison to Level 3 and
2 results, respectively. Each of these models
represented significant processes affecting the mass
balance for toxic chemicals. The hydrodynamic
model predicted water movements necessary to
describe the three-dimensional transport of dissolved
and particulate constituents in the water column. The
eutrophication models described the production,
respiration, grazing, and decomposition of planktonic
biomass within the lake. The contaminant transport
and fate models described contaminant partitioning
between dissolved and sorbed phases, mass transfer
between media (air, water, sediment), and
biogeochemical transformations. The
bioaccumulation models simulated contaminant
accumulation from water and sediments to predator
fish via direct exposure and trophic transfer through
benthic and pelagic food webs. Together, these
submodels formed an integrated description of toxic
chemical cycling in the aquatic ecosystem with which
to predict the relationship between loadings and
concentrations of PCBs.
1.2.5.2 Hydrodynamics (POM)
The Princeton Ocean Model (POM) (Blumberg and
Mellor, 1980, 1987) was used to compute three-
dimensional current fields in the lake. The POM
simulated large- and medium (km)-scale circulation
patterns, vertical stratification and velocity
distribution, seiche, and surface waves. This model
was also used to simulate a thermal balance for the
lake. The POM is a primitive equation, numerical
hydrodynamic circulation model that predicts three-
dimensional water column transport in response to
wind stress, temperature, barometric pressure, and
Coriolis force. The POM has been demonstrated to
accurately simulate the predominant physics of large
water bodies (Blumberg and Mellor, 1983, 1985;
Blumberg and Goodrich, 1990). This model was
used to develop year-long simulations on a 5 km
horizontal grid, with nine sigma vertical layers, at
one-hour intervals for Lake Michigan (Schwab and
Beletsky, 1998). Observed and simulated
meteorological data were used to define model
forcing functions. Extensive measurements of
temperature and current distributions collected in
Lake Michigan during 1982-1983 were used to
provide the necessary data for model calibration;
measurements of daily surface temperature and
current distributions were used to confirm
hydrodynamic simulations for 1994-1995.
21
-------
model type
Level 1 models
Level 2 models
Level 3 models
Hydrodynamic and
load models
POM
advective/
dispersive
transport and
bottom
,shear stress
Eutrophication/
sorbent dynamics
internal carbon
aggregated to
Level 2
Contaminant
transport and fate
transport
"aggregated
to Level 2
Food web
bioaccumulation
Figure 1.2.2. Model construct used for the LMMBP to model PCBs.
1.2.5.3 Eutrophication/Sorbent Dynamics (LM3-
Eutro)
The eutrophication/sorbent dynamics (LM3-Eutro)
model predicted the production, transformation, and
decay of plankton biomass in response to seasonal
dynamics of temperature, light, and nutrient
concentrations. In the open lake, living and dead
plankton comprise the majority of suspended
particles and generate significant autochthonous
loads of particulate and dissolved organic carbon
(POC and DOC) to which PCBs and other
contaminants preferentially partition (Richardson et
al., 1983; DePinto era/., 1993). LM3-Eutrosimulated
the non-conservative, seasonally-variable dynamics
of the biotic organic pool, which has a significant
influence upon partitioning of HOCs (Dean et al.,
1993). A similar, less resolute model was applied to
simulate the dynamics of organic carbon states in
Green Bay as part of the GBMBP (DePinto et al.,
1993). Model outputs included autochthonous solids
loads and transformation and decay rates that were
used as inputs for LM2-Toxic.
1.2.5.4 Contaminant Transport and Fate (LM2-
Toxic)
The mass balance for toxic chemicals in the lake was
computed in a contaminant transport and fate model
(LM2-Toxic) which described contaminant transport,
intermedia exchange, phase distribution, and
biogeochemical transformations in both the water
column and sediments. LM2-Toxic was calibrated
and confirmed for selected individual congeners and
the sum of PCBs congeners. Mass balance analyses
were performed for total PCBs to evaluate the
significant source, transport, and loss pathways. The
effectiveness of alternative load reduction scenarios
upon reducing total PCB concentrations were
forecast.
22
-------
1.2.5.5 Food Web Bioaccumulation (LM Food
Chain)
A bioaccumulation model simulated chemical
accumulation in the food web in response to
chemical exposure, based upon chemical mass
balances for aquatic biota. The general form of the
bioaccumulation equation was well-defined, and
equated the rate of change in chemical concentration
within a fish (or other aquatic organism) to the sum of
chemical fluxes into and out of the animal. These
fluxes included direct uptake of chemical from water,
the flux of chemical into the animal through feeding,
and the loss of chemical due to elimination
(desorption and excretion) and dilution due to growth.
To predict bioaccumulation for top predator fish, the
bioaccumulation mass balance was repeatedly
applied to animals at each trophic level to simulate
chemical biomagnification from primary and
secondary producers, through forage species to top
predators. Food web bioaccumulation models have
been successfully applied for PCBs and other HOCs
in several large-scale aquatic ecosystems (Thomann
and Connolly, 1984; Connolly and Tonelli, 1985) and,
most recently, for the GBMBP (Connolly etal., 1992).
The model developed for that project, FDCHN, was
adapted for use in Lake Michigan (LM Food Chain).
FDCHN is a time-variable, population-based age
class model, incorporating realistic descriptions of
bioenergetic, trophodynamic, and toxicokinetic
processes. The general features of FDCHN were
well-suited to a modeling application such as the
LMMBP. For Lake Michigan, bioaccumulation of
PCB congeners was modeled for lake trout and coho
salmon food webs. Food web bioaccumulation was
simulated for sub-populations of lake trout in three
distinct biotic zones.
1.2.6 Model Quality Assurance
A Quality Assurance Project Plan (QAPP) was
prepared and implemented for the PCBs modeling
(Richardson et al., 2004). The QAPP specified
procedures for code development, testing,
modification, and documentation; as well as methods
and measures applied in model calibration,
confirmation, and uncertainty analysis.
1.2.7 Model Application and
Computational Aspects
1.2.7.1 Annual Simulations
Annual simulations were run with the models for the
period of 1994-1995. Results were analyzed in terms
of regional and lake-wide contaminant loads, fluxes
and inventories, and spatial and temporal gradients
of contaminant concentrations. Bioaccumulation
simulations were analyzed in terms of relative
accumulation pathways, spatial and temporal
variability of contaminant concentration ratios
(bioconcentration factor, bioaccumulation factor,
biota/sediment accumulation factor, predator/prey),
and influence of diet, age, and migration factors.
1.2.7.2 Long-Term Simulations
Long-term simulations were used to forecast the
impact of various management scenarios. Forecasts
were performed to determine time to steady-state for
both continuing and discontinued loads. Forecasts
were also run to evaluate reductions in exposure
concentrations resulting from elimination of tributary
and/or atmospheric loading. These forecasts were
propagated through the food web bioaccumulation
model for PCBs to estimate time for sport fish
contaminant concentrations to decline below criteria
limits.
References
Blumberg, A.F. and G.L. Mellor. 1980. A Coastal
Ocean Numerical Model. In: J. Sunderman and
K.P. Holtz (Eds.), Mathematical Modeling of
Estuarine Physics, pp. 203-214, Proceedings of
the International Symposium, Hamburg,
Germany, August 1978.
Blumberg, A.F. and G.L. Mellor. 1983. Diagnostic
and Prognostic Numerical Circulation Studies of
the South Atlantic Bight. J. Geophys. Res.,
88(C8):4579-4592.
Blumberg, A.F. and G.L. Mellor. 1985. A Simulation
of the Circulation in the Gulf of Mexico. Israel J.
Earth Sci., 34:122-144.
23
-------
Blumberg, A.F.andG.L Mellor. 1987. A Description
of a Three-Dimensional Coastal Ocean
Circulation Model. In: N.S. Heaps (Ed.), Three-
Dimensional Coastal Ocean Models, Coastal and
Estuarine Sciences, pp. 1-16. American
Geophysical Union, Washington, D.C.
Blumberg, A.F. and D.M. Goodrich. 1990. Modeling
of Wind-Induced Destratification in Chesapeake
Bay. Estuaries, 13:1236-1249.
Cardenas, M. and W. Lick. 1996. Modeling the
Transport of Sediment and Hydrophobic
Contaminants in the Lower Saginaw River. J.
Great Lakes Res., 22(3):669-682.
Connolly, J.P. and R. Tonelli. 1985. Modeling
Kepone in the Striped Bass Food Chain of the
James River Estuary. Estuarine Coast. Shelf
Sci., 20:349-366.
Connolly, J.P., T.F. Parkerton, J.D. Quadrini, ST.
Taylor, and A.J. Thurmann. 1992. Development
and Application of PCBs in the Green Bay, Lake
Michigan Walleye and Brown Trout and Their
Food Webs. Report to the U.S. Environmental
Protection Agency, Office of Research and
Development, ERL-Duluth, Large Lakes
Research Station, Grosse lie, Michigan. 300 pp.
Dean, K.E., M.M. Shafer, and D.E. Armstrong. 1993.
Particle-Mediated Transport and Fate of a
Hydrophobic Organic Contaminant in Southern
Lake Michigan: The Role of Major Water Column
Particle Species. J. Great Lakes Res.,
19(2):480-496.
DePinto, J.V., R. Raghunathan, P. Sierzenga, X.
Zhang, V.J. Bierman, Jr., P.W. Rodgers, and T.C.
Young. 1993. Recalibration of GBTOX: An
Integrated Exposure Model for Toxic Chemicals
in Green Bay, Lake Michigan. Final Report. U.S.
Environmental Protection Agency, Office of
Research and Development, ERL-Duluth, Large
Lakes Research Station, Grosse He, Michigan.
132pp.
Endicott, D.D., W.L. Richardson, and D.J. Kandt.
2005. 1992 MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan. In: R. Rossmann (Ed.),
MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan, Part 1. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
EPA/600/R-05/158, 140 pp.
Gailani, J., C.K. Ziegler, and W. Lick. 1991.
Transport of Suspended Solids in the Lower Fox
River. J. Great Lakes Res., 17(4):479:494
Gailani, J.Z., W. Lick, M.K. Pickens, C.K. Ziegler, and
D.D. Endicott. 1994. Sediment and Contaminant
Transport in the Buffalo River. U.S.
Environmental Protection Agency, Office of
Research and Development, ERL-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
54pp.
Gilbertson, M., T. Colborn, and A. Duda. 1991.
Focus on International Joint Commission
Activities, Volume 16, Number 2, July/August
1991, Windsor, Ontario, Canada.
Green, M.L., J.V. DePinto, C.W. Sweet, and K.C.
Hornbuckle, 2000. Regional Spatial and
Temporal Interpolation of Atmospheric PCBs:
Interpretation of Lake Michigan Mass Balance
Data. Environ. Sci. Technol., 34(9): 1833-1841.
Hall, D. and D. Robertson. 1998. Estimation of
Contaminant Loading from Monitored and
Unmonitored Tributaries to Lake Michigan for the
USEPA Lake Michigan Mass Balance Study.
Quality Systems and Implementation Plan.
Submitted October 23,1998. U.S. Environmental
Protection Agency, Great Lakes National
Program Office, Chicago, Illinois. 19 pp.
Lick, W., J. Lick, and C.K. Ziegler. 1994. The
Resuspension and Transport of Fine-Grained
Sediments in Lake Erie. J. Great Lakes Res.,
20(4):599-612.
24
-------
Ludwig, J.P., J.P. Giesy, C.L Summer, W.
Bowerman, R. Aulerich, S. Bursian, H.J. Auman,
P.O. Jones, L.L. Williams, D.E. Tillett, and M.
Gilbertson. 1993. A Comparison of Water
Quality Criteria for the Great Lakes Based on
Human and Wildlife Health. J. Great Lakes Res.,
19(4):789-807.
Miller, S.M., M.L Green, J.V. DePinto, and K.C.
Hornbuckle. 2001. Results from the Lake
Michigan Mass Balance Study: Concentrations
and Fluxes of Atmospheric Polychlorinated
Biphenyls and frans-Nonachlor. Environ. Sci.
Technol., 35(2):278-285.
Pickens, K., C. Donna, M. Chen, J. Lick, and W. Lick.
1992. Sediment Transport in a Stratified Estuary.
In: H. Alder and J.C. Heinrich (Eds.), Numerical
Methods in Engineering and Applied Sciences.
Centro Internacional de Methodos Numericos en
Ingerieria, Barcelona, Spain.
Richardson, W.L., V.E. Smith, and R. Wethington.
1983. Dynamic Mass Balance of PCB and
Suspended Solids in Saginaw Bay - A Case
Study. In: D. Mackay, S. Patterson, and S.J.
Eisenreich (Eds.), Physical Behavior of PCBs in
the Great Lakes, pp. 329-366. Ann Arbor
Science Publishers, Ann Arbor, Michigan.
Richardson, W.L., D.D. Endicott, R.G. Kreis, Jr., and
K.R. Rygwelski(Eds.). 2004. The Lake Michigan
Mass Balance Project Quality Assurance Plan for
Mathematical Modeling. Prepared by the
Modeling Workgroup. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
EPA/600/R-04/018, 233 pp.
Robbins, J.A. 1985. The Coupled Lakes Model for
Estimating the Long-Term Response of the G reat
Lakes to Time-Dependent Loadings of Particle-
Associated Contaminants. National Oceanic and
Atmospheric Administration, Great Lakes
Environmental Research Laboratory, Ann Arbor,
Michigan. NOAA Technical Memorandum ERL
GLERL-57, 41 pp.
Rodgers, P.W. and D. Salisbury. 1981. Water
Quality Modeling of Lake Michigan and
Consideration of the Anomalous Ice Cover of
1976-1977. J. Great Lakes Res., 7(4):467-480.
Schwab, D. and D. Beletsky. 1998. Lake Michigan
Mass Balance Study: Hydrodynamic Modeling
Project. National Oceanic and Atmospheric
Administration, Great Lake Environmental
Research Laboratory, Ann Arbor, Michigan.
NOAA Technical Memorandum ERLGLERL-108,
53pp.
Sonzogni, W.C., W. Richardson, P. Rodgers, and
T.J. Monteith. 1983. Chloride Pollution of the
Great Lakes. Water Pollut. Contr. Fed. J.,
55(5):513-521.
Swain, W.R. 1991. Effects of Organochlorine
Chemicals on the Reproductive Outcome of
Humans Who Consumed Contaminated Great
Lakes Fish: An Epidemiological Consideration. J.
Toxic. Environ. Health, 33:587-639.
Thomann, R.V. and D.M. Di Toro. 1983. Physico-
Chemical Model of Toxic Substances in the Great
Lakes. J. Great Lakes Res., 9(4):474-496.
Thomann, R.V. and J.P. Connolly. 1984. An Age
Dependent Model of PCB in a Lake Michigan
Food Chain. U.S. Environmental Protection
Agency, Office of Research and Development,
ERL-Duluth, Large Lakes Research Station,
Grosse lie, Michigan. EPA/600/S3-84/026, 3 pp.
25
-------
PART1
INTRODUCTION
Chapter 3. Information Management
David A. Griesmer
Computer Sciences Corporation
Large Lakes Research Station
9311 Groh Road
Grosse lie, Michigan 48138
To support the modeling efforts of the Lake Michigan
Mass Balance Project (LMMBP), large amounts of
data were collected and analyzed by a number of
State and government agencies and universities
(Appendix 1.3.1). Data were collected, analyzed,
and sent to the United States Environmental
Protection Agency (USEPA) Great Lakes National
Program Office (GLNPO) in Chicago, Illinois.
GLNPO staff, under the direction of Lou Blume, were
responsible for quality assurance (QA) assessment,
organization, and consolidation of all data. To
facilitate the QA assessment process, a SAS
application Research Data Management and Quality
Assurance System (RDMQ), developed by Syd Allen,
a private contractor, was used to automate the QA
process (Sukloff, 1995). RDMQ is a menu-driven
SAS program. It has capabilities for loading data,
applying quality control checks, adding validity flags,
viewing and editing data, producing user-defined
tables and graphs, and exporting data in ASCII files.
These tasks are performed through a set of menu-
driven SAS programs and macros. Data which had
been put through the assessment process and
approved for release by both GLNPO and the
Principal Investigator (PI) were then sent to USEPA,
Office of Research and Development (ORD)/National
Health and Environmental Effects Research
(NHEERL)/Large Lakes and Rivers Forecasting
Research Branch (LLRFRB)/Large Lakes Research
Station (LLRS) for use by the modeling staff.
1.3.1 Overview of Information
Management at the LLRS
Data received from GLNPO were usually in the form
of electronic media. Data were typically E-mailed,
but sometimes they were downloaded from GLNPO
databases or received on CD-ROM. Data were
reformatted by GLNPO into a form facilitating entry
into database programs at the LLRS. Upon arrival,
raw data were copied to the "Immb" folder on Dave
Griesmer's personal network space ("n:\" drive). In
addition, data were imported into one of several
Microsoft Access databases in the
"\Access_2000\lmmb" folder on Mr. Griesmer's "n:\"
drive. The "n:\" drive was used to facilitate data
security because this file space is backed up
regularly and is only available to Mr. Griesmer. Data
were placed in the Microsoft Access databases to
facilitate data review/assessment and later retrieval
for the modeling team.
Prior to use, several reviews were done of the data
received to look for errors in the data sets. At the
LLRS, this review was broken up into two parts.
First, an initial review was made to check for
completeness of information, to look for transcription
errors, programming errors, and formatting errors,
and to review comments added by collection and
analysis personnel. Second, a review was done by
the data users to determine if the data made
environmental sense. This type of review was
conducted for the open lake, surficial sediment and
sediment trap, lower food chain, and fish
26
-------
polychlorinated biphenyl (PCB) data sets.
Atmospheric PCB fluxes/loadings and tributary PCB
loadings did not go through this review process at the
LLRS, but they were assessed by study members
assigned with providing loading values. Tributary
PCB loading assessment was done by David Hall,
U.S. Geological Survey (USGS). All atmospheric
PCB loading/concentration data were assessed by
Keri Hornbuckle, University of Iowa. The
assessment process used by these individuals is
unknown.
For data reviewed at LLRS, samples which GLNPO
determined had failed the RDMQ QA process were
flagged with the value of -9999. GLNPO preserved
the values in the data sets that were received and
flagged the analytical remark field for that parameter.
Flagging these values as -9999 facilitated processing
by analytical software such as IDL. Parameter
values with analytical remark flags of "INV" (invalid
data, as determined by GLNPO QA evaluation), and
"NAI" (no result reported - interference) were
changed to -9999. Samples with the analytical
remark flag of "LAC" (no results reported, laboratory
accident) were removed.
Documentation associated with the data was studied.
RDMQ data warning fields (RS_NMAND,RS_WARN,
RS_UPDAT) were checked to verify that there were
no problems flagged by RDMQ which were
inadvertently included in the database. Every routine
field sample (RFS) and field duplicate (FD#) was
checked to verify that a valid station name, sampling
date, and depth collection information were included.
The value ranges (minimum, maximum, average) for
all congeners was checked to look for any obvious
errors. Data ranges of all data were also checked for
obvious errors. Data were checked to verify units
and to confirm whether blank, dilution, or surrogate
corrections were done. Quality Control (QC)
Coordinator(RECSTAT),stationnotes(STNNOTES),
and record (RECSTAFF) comment fields were
checked for comments associated with a sample. All
of this information was recorded on a Data
Verification Checklist (Appendix 1.3.2). If questions
or errors were found, they were referred back to
GLNPO for resolution.
Upon completion of this initial data check, readme
files were created to describe the data, and the raw
data set(s) and readme files were copied to a data
archive on the LLRS Unix systems. This archive is
located at \usr\lmmbdata on the Alpha workstation
named Ilrssrv2 and is available to modeling staff at
the LLRS. Each study has its own directory
(LMI0001 LMI0041) within the Immbdata archive.
PCB data for the LMMBP can be found in directories:
LMI0029 (daily gas phase congener, total PCBs, and
frans-nonachlor for each surface cell in the LMMBP
5 km grid); LMI0032 (particulate and precipitation
congener, total PCBs, and frans-nonachlor data for
eight onshore sampling stations around Lake
Michigan and for shipboard sampling from the Lake
Guardian); LMI0035 (open lake congener, total
PCBs, and frans-nonachlor data collected during the
eight LMMBP cruises conducted in 1994-1995);
LMI0036 (phytoplankton, zooplankton, Mysis and
Diporeia congener, total PCBs, and frans-nonachlor
data collected during eight LMMBP cruises);
LMI0037 (forage and predator (lake trout, coho
salmon) fish congener, total PCBs, and frans-
nonachlor data collected for the LMMBP); LMI0040
(surficial sediment and sediment trap congener, total
PCBs, and frans-nonachlor data); and LMI0041 (daily
tributary congener, total PCBs, and frans-nonachlor
loading data from 11 monitored and 18 unmonitored
Lake Michigan tributaries).
At the same time, information about data received
(metadata) was stored in a searchable Microsoft
Access database. The database is found on the
LLRS common drive "\\giord2\grlcommon", which is
also known as the "r:\" drive. This database is
named "Imtrack2000.mdb" and is found in the
r:\access2000A folder. This database is available to
all staff. This database can be searched by library
number (consecutive number assigned when data
are logged in, corresponds to LMI folder name in
Immbdata archive), PI, parameter, PI and parameter,
or library number and parameter (Appendix 1.3.3.).
After initial review of a data set was completed, data
were retrieved from the Microsoft Access databases
and exported into files (usually Microsoft Excel) for
assessment by the modeler who would be using the
data set. Water and sediment PCB data were given
to Xiaomi Zhang. Lower food chain data were
assessed by Xin Zhang and Katie Taunt. Forage and
predator fish data were assessed by Xin Zhang.
Initially, only routine field samples and field duplicates
were given to the data assessors. If issues or
problems were found, the person assessing the data
27
-------
would then request additional QA data. If
questions/problems could not be resolved by looking
at QA data, they were referred back to GLNPO for
resolution.
In several instances, data which passed GLNPO QA
checks from the analytical point of view were rejected
during assessment because values were not
environmentally reasonable. For example,
particulate and dissolved water PCB values from
station MB63 from the October 1995 cruise were
orders of magnitude higher than values for
surrounding stations. In addition, their values were
orders of magnitude higher than values from the
same station collected on different cruises.
Environmentally these results were unreasonable,
and they were not used by the modelers. GLNPO
was informed whenever we rejected data.
After the assessment process was completed, files
were created which could be used in IDL, which is a
software package used for visualization and analysis
of LMMBP data. Standard formats were developed
for water, sediment, and fish data (Appendices 1.3.4,
1.3.5, 1.3.6). All files were fixed format ASCII text
files. One of the principal uses of IDL was to develop
volume-weighted averages (VWA) estimates of
parameter concentrations for each cell in the
modeling grid. These VWA estimates could then be
compared to model results.
1.3.2 Calculation of Total PCBs
In general, total PCBs were calculated by the PI
reporting the data. In the case of tributary loads for
total PCBs, total PCBs were calculated by the PI, and
loads were calculated by David Hall, USGS. In a
similar fashion, total PCBs were calculated by the PI,
and atmospheric loads were calculated by Keri
Hornbuckle, University of Iowa. Open lake total
PCBs were calculated by GLNPO contractor staff
(Marcia Kuehl). A GLNPO contractor, DynCorp,
verified total PCB values. However, the method used
to calculate total PCBs was not consistent from Pl-to-
Pl. Some Pis blanked corrected data; some included
invalid data (samples with INV analytical remark
field); and some did not surrogate correct data. In
those instances when invalid samples were included
in the total PCB calculation or surrogate correction
was not done, total PCBs were recalculated by
DynCorp to correct these problems. Attached are
documents from Marcia Kuehl, PCBs QA
Coordinator, describing how total PCBs were
calculated by each PI (Appendix 1.3.7).
1.3.3 Regression Analysis of Measured
Congener, Total PCB Data
As the modeling study was originally devised, all
modeling was to be done at the congener level;
however, at a later date it was decided that
simulation of total PCBs would also be desirable.
The Level 2 (LM2) and Level 3 (LM3) LMMBP
models did not model total PCBs; therefore, a
method was devised to calculate total PCB
concentrations for model results based on the set of
congeners modeled. Regressions and ratios were
calculated comparing the Pis' measured total PCB
field values (the independent variable) to the Pis'
measured sum of the congeners that were modeled
at the LLRS (the dependent variable) in all media
modeled (atmospheric vapor phase, wet and dry
deposition, dissolved and particulate tributary water,
total, dissolved and particulate water, surficial
sediment, phytoplankton, zooplankton, Diporeia,
Mysis, and all forage and predator fish species).
Note that total PCBs in water were not measured, but
were derived by adding up dissolved and particulate
PCBs for each sample. With R**2 values of .90 or
greater, these regression analyses produced very
good results (Table 1.3.1).
Additional analysis was then performed to produce
an uncertainty estimate for the regression equations.
A mean was calculated for the slope of the line in the
linear regression (z), and 95 percent confidence
intervals were calculated for z using the formula:
z = x/y
where
x = total of PCB congener subset that was
modeled
y = true total PCBs as calculated by the PI.
David Miller, statistician at LLRS, verified that the z
values were generally normally distributed. This
allowed us to calculate a mean, standard deviation,
28
-------
Table 1.3.1. Revised Regression Equations for the LMMBP Total PCBs in All Media
Media
Atmospheric Vapor Phase
Atmospheric Dry Deposition
Atmospheric Wet Deposition
Tributary Loading Data
Dissolved Water
Particulate Water
Dissolved + Particulate Water
Surficial Sediment
Phytoplankton
Zooplankton
Diporeia
Mysis
Alewife < 1 20 mm
Alewife > 1 20 mm
Bloater < 1 60 mm
Bloater > 1 60 mm
Deepwater Sculpin
Slimy Sculpin
Adult Smelt
Hatchery Coho
Coho Yearling
Coho Adult
Adult Lake Trout
Ratio of PI Calculated
Total PCBs to Summed
Modeling Congeners
1 .2944
1 .3597
1 .5775
1 .2476
1 .4822
1 .2948
1.4147
1.1805
1 .3842
1 .3923
1.3652
1.3162
1 .4458
1 .4281
1 .4761
1 .4827
1.5157
1 .4976
1 .4447
1 .2836
1.497
1.444
1 .4897
Pis' Calculated Total PCBs
Versus Summed Modeling
Congeners: Regression
Equation
y = 1 .2707x + 0.0891
y= 1.3204X + 0.2159
y= 1. 691 7x- 0.0322
y= 1. 21 34x + 0.7752
y = 1 .2738x + 0.0268
y= 1 . 2251 x + 0.0051
y = 1 .2427x + 0.0347
y= 1.1668X + 0.6125
y= 1. 2871 x + 3.621 6
y = 1 .2058x + 22.833
y= 1.3763X-3.4124
y= 1.3829X- 12.842
y= 1.4534X- 1.296
y= 1.3338X + 38.145
y = 1. 431 7x + 19.505
y = 1.4146X + 38.14
y = 1 .3752x + 38.735
y = 1 .5272x - 8.4009
y = 1 .46693x - 5.2828
y = 1 .4009x - 1 1 .024
y = 1.6263X- 16.984
y = 1.4392x + 2.7179
y = 1 .4875x + 3.7424
R2
0.9997
0.9623
0.9672
0.991
0.9413
0.9992
0.9829
0.997
0.9584
0.9595
0.9795
0.9833
0.9784
0.947
0.9851
0.9426
0.9897
0.9257
0.9044
0.994
0.9835
0.9927
0.9977
and 95 percent confidence intervals for the z value in
all media modeled (Appendix 1.3.8).
A comparison of field measured water congeners
values to model results from Xiaomi Zhang, WelSo
modeler at LLRS, indicated that there were some
problems with measured field results for congeners
84+92 and 99. In both dissolved and particulate
water fractions, concentrations for both of these
congeners were much higher than model results.
Investigation into this issue revealed a contamination
issue with congener 99. The analytical technique
used to measure the water congener could not
adequately separate congener 99 from trans-
nonachlor, which caused a co-elution problem. The
reason for the high field values for congener 84+92
was unclear, but it also is believed to be caused by a
co-elution problem. Xiaomi Zhang developed ratios
comparing field data to model results (Table 1.3.2).
These ratios were used to correct 84+92 and 99
congener values by the following formula:
Corrected congener nn value = Measured congener
nn value/ratio
where
nn = congener 84+92 or 99.
29
-------
Table 1.3.2. Ratio of Measured Field Data/Model
Results for Congeners 84+92 and 99 in Water
Congener
84+92
99
Ratio in
Dissolved
Water
2.08
1.15
Ratio in
Particulate
Water
3.41
1.94
Regression analysis was then redone for dissolved,
paniculate, and dissolved + particulate water.
These revised regression equations were then
applied to summed modeled congeners to calculate
modeled total PCBs. Regression equations for
dissolved and particulate fractions of total PCBs have
a positive y-intercept. This means that when these
regressions are used to calculate total PCBs, the
value will never drop to zero even if modeled
congeners drop to zero. Meetings were held with the
modeling staff to discuss this issue. It was believed
that this bias was caused by 1) lack of blank
correction of congener data, 2) detection limits, and
3) inherent uncertainty in the regression process.
Since it was not possible to correct these problems,
the decision was made to use the regression
equations, and carefully explain these difficulties
when documenting modeling scenarios.
1.3.4 Summary
The LMMBP data received at the LLRS were
carefully evaluated prior to use to insure that the
field data being used by the modelers were as
accurate as possible. In addition, data were archived
and cataloged to protect these valuable data sets
and make it easier for users to find the information.
Incorporation of this information into LLRS Microsoft
Access databases has given us flexibility in retrieving
the information needed by the modeling staff at the
LLRS. These data were used to develop regression
equations which were used to approximate total PCB
concentrations for modeled data.
References
Sukloff, W.B., S. Allan, and K. Ward. 1995. RDMQ
User Manual. EnvironmentCanada, Atmospheric
Environment Service, North York, Ontario,
Canada. 91 pp.
30
-------
PART1
INTRODUCTION
Appendix 1.3.1. List of Parameters Analyzed and Principal Investigators for the
LMMBP
Parameter
Focus
Group
Media
Notes
Principal Investigator
Atrazine, Deethylatrazine IUAA
(DEA),
Deisopropylatrazine
(DIA), Terbuthylazine
Atrazine, DEA, DIA,
d5-Atrazine
Atrazine
Atrazine
WSAA
Atmospheric
Vapor and
Particulate Phase,
Precipitation
Atmospheric
Vapor and
Particulate Phase,
Precipitation
RULA Open Lake
RUTA Tributary
Sleeping Bear
Dunes site
only.
All stations
except
Sleeping Bear
Dunes site.
Ron Hites, Indiana
University. Keri
Hornbuckle, U. of Iowa
used this data to
calculate loadings.
Clyde Sweet, Illinois
State Water Survey.
Keri Hornbuckle, U. of
Iowa used this data to
calculate loadings.
Steven Eisenreich,
Rutgers University
Steven Eisenreich,
Rutgers University.
David Hall, USGS used
this data to calculate
loads.
Alkalinity, Conductivity, GPLN Open Lake
Hardness, pH, Turbidity
Alkalinity, Chloride,
Conductivity, NO2+NO3,
Organic Carbon, pH,
Total Phosphorus, TKN
GRAN Atmospheric
Marvin Palmer,
GLNPO. Analysis by
Grace Analytical Labs.
Glenn Warren, GLNPO.
Analysis by Grace
Analytical Labs.
31
-------
Parameter
Conductivity, DO, pH,
Temperature
Alkalinity, Chloride,
Focus
Group
USTN
LHTN
Media
Tributary
Tributary
Notes Principal Investigator
Peter Hughes, USGS
George Bowman,
Conductivity, NH4,
NO2+NO3J
Orthophosphorus,
SiCySiOg, Sulfate, Total
Phosphorus, TSS, TKN
Chloride,
Ammonium-Nitrogen
(NH4N), N02+N03,
Dissolved Phosphorus,
Orthophosphorus, Total
Phosphorus, Dissolved
Silica , TKN
Phosphorus (Base
Extractable as PO4),
Total Phosphorus,
Biogenic Silica (SiO2-bio)
Total Organic Carbon,
Total Organic Nitrogen
Organic Carbon (DOC,
POC), TSS
Organic Carbon (DOC,
POC)
Mercury
Mercury
Mercury
GRLN
GLSN
MNPH
Open Lake
Sediment,
Sediment Traps
NASN
BALN
WWTN
Sediment,
Sediment Traps
Open Lake
Tributary
MIAH Atmospheric
Vapor, Particulate,
and Precipitation
Monthly
deposition/
concentration
calculated by
Matt Landis,
University of
Michigan
MDLH Open Lake
Plankton
Wisconsin State Lab of
Hygiene. David Hall,
USGS used this data to
calculate loads.
Glenn Warren, GLNPO,
Analysis by Grace
Analytical Labs.
Thomas Johengen,
NOAA/GLERL
Brian Eadie,
NOAA/GLERL
Eric Crecelius, Battelle
Marine Sciences
Martin Shafer,
University of Wisconsin
Water Quality
Laboratory
Jerry Keeler, University
of Michigan
Robert Mason, U. of
Maryland. David Hall,
USGS used this data to
calculate loads.
Edward Nater, U. of
Minnesota
32
-------
Parameter
Focus
Group
Media
Notes
Principal Investigator
Mercury
LLSH Sediment,
Sediment Trap
Ronald Rossmann,
LLRS
Mercury, Methylmercury WWTH Tributary
Mercury
Congener PCBs,
frans-nonachlor
Congener PCBs,
frans-nonachlor
Congener PCBs,
frans-nonachlor
Congener PCBs,
frans-nonachlor
Congener PCBs,
frans-nonachlor
Congener PCBs,
frans-nonachlor
Congener PCBs,
frans-nonachlor
Congener PCBs,
frans-nonachlor
MIFH Fish
I DAP Atmospheric
Vapor and
Particulate Phase,
Precipitation
WSAP Atmospheric
Vapor and
Particulate Phase,
Precipitation
RUAP Atmospheric Dry
Deposition
BALP Open Lake
LHTP Tributary
NASP Sediment,
Sediment Trap
MNPP Plankton, Mysis,
Diporeia
MNFP Forage Fish, Lake
Trout, Coho
Salmon
Sleeping Bear
Dunes site
only.
All stations
except
Sleeping Bear
Dunes site.
Includes both
land and over-
lake sampling
sites.
Jim Hurley, U. of
Wisconsin, Water
Quality Laboratory.
David Hall, USGS used
this data to calculate
loads.
Jerome Nriagu, U. of
Michigan
Ron Hites, Indiana
University. Keri
Hornbuckle, U. of Iowa
used this data to
calculate loadings.
Clyde Sweet, Illinois
State Water Survey.
Keri Hornbuckle, U. of
Iowa used this data to
calculate loadings.
Steven Eisenreich,
Rutgers University
Eric Crecelius, Battelle
Marine Sciences
Laboratory
William Sonzogni,
Wisconsin State Lab of
Hygiene
Pat Van Hoof,
NOAA/GLERL
Deborah Swackhamer,
U. of Minnesota
Deborah Swackhamer,
U. of Minnesota
33
-------
Parameter
Focus
Group
Media
Notes
Principal Investigator
Seabird Temperature,
Chlorophyll a,
Transmissivity
Chlorophyll a
N/A
Primary Productivity
Abundance/Biomass
Diet Information
Diet Information
Diet Information
Cs-137andPb-210
Sediment Bulk Density,
Fraction Dry weight,
Porosity, Sediment
Mixing Depth, Vertical
Sediment Transport, Net
Mass Accumulation Rate
BSDB
FSDB
NASR
N/A
Open Lake
WSLH Tributary
GRLY Open Lake
Phytoplankton
GRLP Phytoplankton
Abundance/Biomass GRLZ Zooplankton
BSDB Forage Fish
Lake Trout
Coho Salmon
Sediment
Sediment
Chlorophyll a
calculated from
fluorescence
data
Glenn Warren, GLNPO
George Bowman,
Wisconsin State Lab of
Hygiene. David Hall,
USGS used this data to
calculate loads.
Glenn Warren, GLNPO.
Analyzed by Grace
Analytical Laboratory
Glenn Warren, GLNPO.
Analyzed by Grace
Analytical Laboratory.
Glenn Warren, GLNPO.
Analyzed by Grace
Analytical Laboratory.
John Gannon/
Jacqueline Savino,
USGS, National
Biological Survey
John Gannon/Edward
Brown, USGS, National
Biological Survey
Mark Holey, U.S. Fish
and Wildlife Service
John Robbins,
NOAA/GLERL
John Robbins,
NOAA/GLERL
34
-------
PART1
INTRODUCTION
Appendix 1.3.2. Example of Data Verification Checklist
Used for the LMMBP
Data Verification Checklist
FOCUS Version Number Date Received
Description:
1. Read any documentation which came with data files:
2. Make sure I understand field names in RDMQ files:
3. Check fields which according to RDMQ should not be flagged/or indicate some question, with data (e.g.
RS_NMAND, RS_WARN, RSJJPDAT).
RS_NMAND
RS_WARN
RS UPDAT
4. Make sure every RFS and field duplicate has station, date, depth collected information.
5. Check to make sure every sample has station name that is valid.
35
-------
6. Check number of RFS and field duplicates for every analyte. Total Samples
Analyte RFS FDn
Analyte RFS FDn
Analyte RFS FDn
Analyte RFS FDn
Analyte RFS FDn
Analyte RFS FDn
7. Analysis Results for RFS and field duplicates for every analyte.
Analyte Avg Min Max Count
Analyte
Analyte
Analyte
Analyte
Analyte
Analyte
Analyte
Analyte
Analyte
Analyte
Analyte
Analyte
Analyte
Analyte
Analyte
Analyte
Avq
Ava
Ava
Avg
Ava
Ava
Ava
Ava
Ava
Ava
Ava
Ava
Ava
Ava
Ava
Ava
Min
Min
Min
Min
Min
Min
Min
Min
Min
Min
Min
Min
Min
Min
Min
Min
Max
Max
Max
Max
Max
Max
Max
Max
Max
Max
Max
Max
Max
Max
Max
Max
Count
Count
Count
Count
Count
Count
Count
Count
Count
Count
Count
Count
Count
Count
Count
Count
36
-------
Analvte
Analyte
Analvte
Analvte
Avg
Avg
Ava
Ava
Min
Min
Min
Min
Max
Max
Max
Max
Count
Count
Count
Count
8. Check date ranges of data to see if they are reasonable.
Analyte Min Max.
Analyte Min Max.
Analyte Min Max.
Analyte Min Max.
Analyte Min Max.
Analyte Min Max.
Analyte Min Max
Analyte Min Max
Analyte Min Max.
Analyte Min Max.
Analyte Min Max.
Analyte Min Max.
9. Check to verify units information looks alright.
10. Number of significant digits for each analyte.
11. Number of negative values for each analyte.
37
-------
12. Check flags on RFS and field duplicates.
13. Core slice range (sediment)/species, age, length, weight (fish).
14. Check blank correction, dilution, and surrogate correction fields.
15. Questions about QC Coordinator remarks (RECSTAT). Check flags for whole record (RECSTATF).
Questions about Station Notes (STNNOTES), Field Remarks (FREMARK), and Sample Description
(SAMPDESC).
18. Additional Questions.
38
-------
PART1
INTRODUCTION
Appendix 1.3.3. Printout of Information Stored in the LMMBP Tracking
Database (R.-\\Access2000\\lmmb\lmtrack.mdb)
LMMBP DATA ARCHIVE - QUICK REPORT
Note: All Data Archived on Ilrssrv2 in /usr/lmmbdata.
Library No.
Description
Library No.
Description
Library No.
Description
Library No.
Description
Library No.
Description
LM10001
PI:
David Scwab
Hourly Lake Michigan wind, wave, and atmospheric data (5 km grid) for 1982, 1983, 1994, 1995.
Original data files were converted to SEDZL and POM formerly by M. Settles. Also, bathymetric
data for Lake Michigan.
LM 10002
PI:
William Richardson
STORET conventional and general chemistry data for Lake Michigan, April 1962 - August 1993.
Note: Date range varies by parameter, includes original file, reformatted spreadsheet, and MS
Access file.
LM 10003
PI:
David Scwab
2D and 3D GLERL hydrodynamics data for the Lake Michigan 5 km grid. 2D data: January 1982-
September 1983; 3D: covers January-July 1982. Program//llrssrv2/~model/dev/PATRIC2D/RCS is
for 2D processing, no 3D programming yet.
LM 10004
PI:
Steven Eisenreich
Open Lake (RULA) and tributary (RUTA), atrazine, DEA, DIA data for LMMBP Open lake 325
samples (1/17/94 - 4/17/95). Tributary: 126 samples (4/4/95 - 5/15/96). Revised version of data
sent 2/19/98.
LM 10005
PI:
Angela Bandemehr
Hourly meteorological data (air temperature, solar radiation, relative humidity, wind speed and
direction, and precipitation) from 13 air sampling sites both in and outside of the Lake Michigan
basin. 11/30/90 -12/31/96 (Dates vary by site).
39
-------
Library No.
Description
Library No.
Description
Library No.
Description
LM10006
PI:
Glenn Warren
Seabird water temperature data for seven LMMBP surveys, April 1994 - October 1995. Data
collected at 0.5 m intervals. Does not include January 1994 survey. Note: Data received was
extensively revised from original version.
LM10007
PI:
David Hall
Tributary flow data for 11 tributaries to Lake Michigan (Fox, Grand, Indiana Harbor, Kalamazoo,
Manistique, Menominee, Milwaukee, Muskegon, Pere Marquette, Sheboygan, St. Joseph), 1/1/94-
12/31/95. Some data estimated.
LM10008
PI:
Glenn Warren
Open lake organic carbon (dissolved and particulate), and solids data for eight LMMBP cruises.
Sampling date was 4/14/94 -10.13/96. Data also received in D-base (dbf) format. Focus: BALN
Library No.
Description
Library No.
Description
Library No.
Description
Library No.
Description
LM10009
PI:
Debra Piper
Open Lake Michigan nutrient data (chlorophyll, ammonia, NO2+NO3, total phosphorus, dissolved
phosphorus, orthophosphate, silica, and TKN), 4/24/94 -10/13/95 (orthophosphate, NH4-N started
10/14/94). Focus GRLN, collected and analyzed by Grace Labs. Focus: GRLN.
LM10010
PI:
David Hall
Total and dissolved mercury loading estimates for monitored and unmonitored tributaries to Lake
Michigan -1/1/94 -12/31/95. Note: Associated flow data is included in an earlier release of this
data.
LM10011
PI:
David Schwab
Lake Michigan final report, hourly circulation, meteorology, and wave data (5 km grid) for 1982,
1983,1994, 1995. Includes intake, cruise, mooring, water level data. Also, HTML files and images,
model results (XDR format), Fortran and IDL programs.
LM10012
PI:
David Rockwell
Open lake conventional data (alkalinity, conductivity, total hardness, pH, turbidity) from eight
LMMBP cruises, 4/24/94 -10/13/95. Data received in D-base, and Lotus formats. Focus: GPLN.
Library No.
Description
Library No.
Description
LM10013
PI:
Peter Hughes
Conventional data (conductivity, dissolved oxygen, pH, temperature) collected from 11 Lake
Michigan tributaries, 3/29/94 -12/5/95. Note: Five dissolved oxygen samples were flagged as
invalid (INV) and should not be used. Files in D-base and Lotus formats. Focus: USTN.
LM10014
PI:
Robert Mason
Open lake mercury data (particulate and total), collected 6/17/94 -10/10/95. Note: There are 11
invalid samples (flagged as INV) in this data set which should not be used. Data received in D-
base, Lotus, SAS, and tabular delimited formats. Focus: MDLH.
Library No. LM10015 PI: David Rockwell
Description Secchi depth data collected during eight LMMBP cruise, 1/16/94 -10/11/95. Focus: GPLS.
Library No.
Description
LM10016
PI:
Matt Landis
Mercury deposition/concentration data estimated into the NOAA/GLERL 5 km over water grid. All
data are monthly averages. Covers time period: July 1994 - October 1995.
40
-------
Library No.
Description
LM10017
PI:
Glenn Warren
By species and by functional group, abundance, and biomass value, for open lake zooplankton
data collected during eight LMMBP cruses (4/24/94 -10/10/95). Focus Group: GRLZ.
Library No. LM10018 PI: George Bowman
Description Tributary chlorophyll a data, collected 3/24/94 -10/31/95. Focus Group: LHTL.
Library No.
Description
Library No.
Description
Library No.
Description
Library No.
Description
Library No.
Description
Library No.
Description
Library No.
Description
Library No.
Description
LM10019
PI:
Tom Johengen
Sediment nutrient data (NaOH extractable PO4, total phosphorus, biogenic silica), 6/24/94 -
5/22/96. Note: One sample was flagged as invalid (INV) for all parameters and should not be used.
Focus Group: GLSN.
LM10020
PI:
Keri Hornbuckle
Atmospheric atrazine and nutrient (NO3, total phosphorus, TKN) wet deposition loading data for
Lake Michigan 5 km grid cells used in hydrodynamic model. Atrazine wet deposition and
particulate monthly concentration data. Data for 10/94 -10/95 (nutrient) and 5/94 -10/95
(atrazine).
LM10021
PI:
David Hall
Tributary inorganic/nutrient loading data for 10 parameters (alkalinity, ammonia, chloride, NO2+NO3,
orthophosphate, dissolved silica, TKN, total nitrogen, total phosphorus, total suspended solids)
collected from 11 monitored Lake Michigan tributaries. Data collected 1/1/94 -12/31/95.
LM 10022
PI:
David Hall
Atrazine, DEA, DIA tributary loading data for 11 monitored tributaries and atrazine data for
unmonitored tributaries to Lake Michigan. Data covers the time period: 1/1/94 -12/31/95.
LM 10023
PI:
Glenn Warren
Primary productivity data collected during eight surveys for the LMMBP study. Data covers the
time period: 4/24/94 -10/13/95. Data did not go through RDMQ, but was QA'd by Deb Piper, Grace
Analytical. Focus Group: GRLY.
LM 10024
PI:
Glenn Warren
Abundance and biomass for plankton samples collected during the LMMBP. Data covers the time
period: 4/24/94 -10/10/95. Data is reported both by individual species and by functional group.
Focus: GRLP.
LM 10025
PI:
John Robbins
Sediment radiochemistry (Pb-210, Cs-137), physical properties (mass, fraction dry weight, soluble
fraction, bulk density, porosity, age) of Lake Michigan sediment collected 1994-1996. Station
location and modeled data (mixing rates, settling) also included. Focus: NASR.
LM 10026
PI:
Nathan Hawley
Current velocity, water transparency, temperature from three stations, 10/31 /94 -10/11 /95. In-situ
sediment resuspension from sediment flume experiments (8/12/95 - 9/23/98). Also profile data -
temperature, dissolved oxygen, conductivity, BAT, pH, fluorescence, TSM data from six stations in
Lake Michigan (1/4/95 -11/29/95).
41
-------
Library No.
Description
Library No.
Description
Library No.
Description
LM10027
PI:
Barry Lesht
Current velocity and direction, bottom wave orbital velocity, temperature, beam attenuation, and
TSM data collected from Tripod Station 98(latitude 42 52.18, longitude 87 42.41), during the
EEGLE project, 4/2/98 -12/1/98. Data collected every 30 minutes.
LM 10028
PI:
Michael Settles
NEMA and NOAA wind speed and direction, wave height and period data for six stations in Lake
Michigan, retrieved from ACOE Web Site (http://bigfoot.wes.army.mil/c300.html). 1980-1998 (not
all stations cover entire date range). NEMO-Daily data, NOAA-Hourly data.
LM 10029
PI:
Keri Hornbuckle
Daily gas phase congener, total PCBs, and frans-nonachlor concentration data for LMMBP 5 km
grid. Covers time period: 1/1/94 - 9/30/95.
Library No.
Description
Library No.
Description
Library No.
Description
Library No.
Description
Library No.
Description
Library No.
Description
LM 10030
PI:
Catherine Taylor
Sedflume data gathered from Catherine Taylor's masters thesis. Excel files contain erosion data at
different energy levels and graphs of shear stress at different erosional stresses.
LM 10031
PI:
Ronald Rossmann
Surficial sediment/sediment trap mercury data collected during eight LMMBP cruises, 7/18/94 -
8/28/95. Surficial sediment samples collected by R. Rossmann, trap samples by Brian Eadie,
NOAA. All analysis by R. Rossmann. D-base files from RDMQ output. GLNPO Focus LLSH.
LM 10032
PI:
Keri Hornbuckle
Congener, total PCBs, frans-nonachlor monthly and annual loading and flux data (particulate and
precipitation data) for eight land stations around Lake Michigan and from Lake Guardian. Monthly
and annual precipitation loading for LMMBP 5 km grid (text files). 4/94 - 9/95.
LM 10033
PI:
Ken Klewin
Nutrient data (chlorophyll, NO2+NO3, silica, total dissolved phosphorus, and total phosphorus for
1998 spring and summer GLNPO surveys of all five Great Lakes. Lake Michigan data was
collected 4/25/98 - 5/10/98 and August 1998 - first week in September 1998.
LM 10034
PI:
Rich Quintal
1994-1995 Lake Michigan diet, length, weight, age, and migration data (coho only) for forage fish,
coho salmon and lake trout. Data compiled by Lauri Davis based on data received from R. Quintal.
LM 10035
PI:
Marcia Kuehl
Open water congener PCBs and frans-nonachlor data collected during eight LMMBP cruises,
4/24/94 -10/13/95. Data analyzed by Battelle Marine Science Lab (Focus: BALP).
Library No.
Description
Library No.
Description
LM 10036
PI:
Deborah Swackhamer
Lower food chain (phytoplankton, zooplankton, Mysis, Diporeia) congener PCBs, total PCBs, frans-
nonachlor, lipids and moisture data collected during LMMBP cruises (6/17/94 -10/10/95). Focus
group: MNPP.
LM 10037
PI:
James Mickey
Lipids, moisture, congener PCBs data for composite samples of two predator fish species (coho
salmon, lake trout) and five forage fish (alewife, bloater, deepwater sculpin, slimy sculpin, and
smelt). Length/weight data also. Data collected during LMMBP
42
-------
Library No.
Description
Library No.
Description
Library No.
Description
Library No.
Description
LM10038
PI:
Ken Klewin
Nutrient data (chlorophyll, dissolved and total phosphorus, dissolved silica, NO2+N03) collected and
analyzed by GLNPO for spring (4/9 - 4/30/00) and summer (8/2 - 8/28/00) 2000 cruises for selected
stations for all five Great Lakes. Note: File is a Lotus 123 file, the rest are Excel files.
LM 10039
PI:
Glenn Warren
Seabird chlorophyll a transmissivity profiles for eight LMMBP cruises (4/25/94 -10/13/95).
Chlorophyll a calculated by John Goldsmith, GLNPO. Most 1/2 m depth intervals.
LM 10040
PI:
Pat Van Hoof
Surficial sediment and sediment trap PCB congeners, total PCBs, and frans-nonachlor data
collected 10/5/94 - 5/22/96 (surficial), 7/4/94 - 8/28/95 (traps), analyzed by Pat Van Hoof. Samples
are from top 1 cm of box and gravity cores, and ponar samples.
LM10041
PI:
David Hall
Daily tributary loading data (1/1/94 -12/31/95) for PCB congeners, total PCBs, and frans-nonachlor
for 11 monitored and 18 unmonitored tributaries to Lake Michigan. Data analyzed by Wisconsin
Laboratory of Hygiene. Loading estimates by David Hall and Faye Blondin.
43
-------
PART1
INTRODUCTION
Appendix 1.3.4. Generalized Format for the LMMBP Water Data to be
Analyzed With IDL Programs
Beginning -
Ending
Columns
1 7
8-8
9- 14
15-15
16-22
23-23
24-35
36-36
37-44
45-45
46-53
54-54
55-58
59-59
Variable Description
Cruise Name
Blank Space
Latitude (ddd.ddd)
Blank Space
Longitude (-ddd.ddd)
Blank Space
Station Name
Blank Space
Depth Sampled
Blank Space
Sampling Start Date
(mm/dd/yy)
Blank Space
Sampling Start Time
(24-hour clock)
Blank Space
Format
(A = Alpha, F =
Floating Point No.,
1 = Integer, X = Skip)
A7
1X
F6.3
1X
F7.3
1X
A12
1X
F8.0
1X
A8
1X
A4
1X
Sort Order
(A = Ascending,
D = Descending,
Blank = None)
A
N/A
N/A
N/A
A
N/A
A
N/A
A
N/A
N/A
Missing Data
Code
Blank
N/A
Blank
N/A
Blank
N/A
Blank
N/A
Blank
N/A
Blank
N/A
Blank
N/A
44
-------
Beginning -
Ending
Columns
60-67
68-68
69-72
73-73
74-75
76-76
77-79
80-80
81 -88
89-103
104-111
112-126
Variable Description
Sampling End Date
(mm/dd/yy)
Blank Space
Sampling End Time
(24-hour clock)
Blank Space
Filter Fraction
Blank Space
Sample Type
Blank Space
Value Parameter 1
Parameter 1 Flags
Value Parameter 2
Parameter 1 Flags
1
t
Value Parameter n
Parameter n Flags
Format
(A = Alpha, F =
Floating Point No.,
1 = Integer, X = Skip)
A8
1X
A4
1X
A2
1X
A3
1X
F8.0
A15
F8.0
A15
F8.0
A15
Sort Order
(A = Ascending,
D = Descending,
Blank = None)
A
N/A
N/A
A
N/A
D
N/A
Missing Data
Code
Blank
N/A
Blank
N/A
Blank
N/A
Blank
N/A
-9999
Blank
-9999
Blank
-9999
Blank
45
-------
PART1
INTRODUCTION
Appendix 1.3.5. Generalized Format for the LMMBP Sediment Data to be
Analyzed With IDL Programs
Beginning -
Ending
Columns
1 -6
7-7
8-14
15-15
16-27
28-28
29-36
37-37
38-47
48-48
49-58
59-59
60-65
66-66
Variable Description
Latitude (ddd.ddd)
Blank Space
Longitude (-ddd.ddd)
Blank Space
Station Name
Blank Space
Station Depth
Blank Space
Sampling Start Date
(mm/dd/yyyy)
Blank Space
Sampling End Date
(mm/dd/yy)
Blank Space
Top of Core Slice
Blank Space
Format
(A = Alpha, F =
Floating Point No.,
1 = Integer, X = Skip)
F6.3
1X
F7.3
1X
A12
1X
F8.0
1X
A10
1X
A8
1X
F6.0
1X
Sort Order
(A = Ascending,
D = Descending,
Blank = None)
N/A
N/A
A
N/A
N/A
N/A
N/A
A
N/A
Missing
Data Code
Blank
N/A
Blank
N/A
Blank
N/A
Blank
N/A
Blank
N/A
Blank
N/A
Blank
N/A
46
-------
Beginning -
Ending
Columns
67-72
73-73
74-75
76-76
77-79
80-80
81 -87
88-88
89-96
97 - 1 1 1
112-119
120-134
Variable Description
Bottom of Core Slice
Blank Space
Filter Fraction
Blank Space
Sample Type
Blank Space
Collection Method
Blank Space
Value Parameter 1
Parameter 1 Flags
Value Parameter 2
Parameter 1 Flags
1
t
Value Parameter n
Parameter n Flags
Format
(A = Alpha, F =
Floating Point No.,
1 = Integer, X = Skip)
F6.0
1X
A2
1X
A3
1X
A7
1X
F8.0
A15
F8.0
A15
F8.0
A15
Sort Order
(A = Ascending,
D = Descending,
Blank = None)
N/A
A
N/A
D
N/A
N/A
Missing
Data Code
Blank
N/A
Blank
N/A
Blank
N/A
Blank
N/A
-9999
Blank
-9999
Blank
-9999
Blank
47
-------
PART1
INTRODUCTION
Appendix 1.3.6. Generalized Format for the LMMBP Fish Data to be
Analyzed With IDL Programs
Beginning -
Ending
Columns
1 13
14 14
15-27
28 28
29-41
42-42
43-45
46-46
47-49
50-50
51 -52
53-53
54-58
59-59
60 64
Variable Description
Species
Blank Space
Biota Zone
Blank Space
Station Name
Blank Space
No. of Samples in
Composite
Blank Space
Minimum Age
Blank Space
Maximum Age
Blank Space
Minimum Length
Blank Space
Maximum Length
Format
(A = Alpha, F =
Floating Point No.,
1 = Integer, X = Skip)
A13
1X
A13
1X
A13
1X
13
1X
13
1X
12
1X
F5.0
1X
F5.0
Sort Order
(A = Ascending,
D = Descending,
Blank = None)
N/A
N/A
N/A
N/A
N/A
N/A
N/A
Missing
Data Code
Blank
N/A
Blank
N/A
Blank
N/A
Blank
N/A
Blank
N/A
Blank
N/A
.
Blank
N/A
Blank
48
-------
Beginning -
Ending
Columns
65-65
66-87
88-88
89-99
100-100
101 110
111 111
112-119
120-120
121 -128
129-129
130-137
138-138
139-146
147 147
Variable Description
Blank Space
Gender
Blank Space
Sampling Start Date
(mm/dd/yyyy)
Blank Space
Sampling End Date
(mm/dd/yy)
Blank Space
Parameter Units
Blank Space
Biota Part Sampled
Blank Space
Value Parameter 1
Blank Space
Value Parameter 2
Blank Space
1
T
Value Parameter n
Parameter n Flags
Format
(A = Alpha, F =
Floating Point No.,
1 = Integer, X = Skip)
1X
A22
1X
A11
1X
A10
1X
A8
1X
A8
1X
F8.0
1X
F8.0
1X
F8.0
A15
Sort Order
(A = Ascending,
D = Descending,
Blank = None)
N/A
N/A
N/A
N/A
N/A
N/A
N/A
N/A
Missing
Data Code
N/A
Blank
N/A
Blank
N/A
Blank
N/A
Blank
N/A
Blank
N/A
-9999
N/A
-9999
N/A
-9999
Blank
49
-------
PART1
INTRODUCTION
Appendix 1.3.7. Documents From Marcia Kuehl Discussing How Total PCBs
Were Calculated by Focus Group
PCBs total routines
2/28/01
IUAP: reported by PI as pcbtot, is in rdmq output. PI added all congeners except 30, 204,14, 65,166 , and
in two cases 44 due to interference. As PI added prior to rdmq verification, INV data included in totals and
totals not surrogate corrected. Dyncorp redid, using surrogate corrected values for summation and excluding
INV results and #44 as indicated by PI. Values are called pcbtot2 in the rdmq output.
WSAP: PI added prior to rdmq verification and gave to modeler. Not reported to GLNPO and therefore not
in rdmq output. INV data included in totals. Added all congeners except 30, 204, 14, 65, 166. Dyncorp did
totals, using surrogate corrected values for summation and excluding INV results. Some samples have no
pcbtot due to whole sample invalidated from biased Lake Guardian sampling location.
RUAP: PI added and included mention of protocol in ES&T journal article. Not reported to GLNPO and
therefore not in rdmq output. Added all congeners except 30, 204, 14, 65, 166, and 4+10, 6, 8+5, 7+9 due
to interference/contamination. Each sample had its associated field blank (FMB) subtracted from it. No mention
of how to handle negative results included, but I would assume negative values revert become zero. Dyncorp
did totals including INV results and blank subtraction with negative results reverting to zero. Totals were
qualified as field blank corrected (FBC).
LHTP: Modeler added. Not reported to GLNPO and therefore not in rdmq output. INV data not included in
totals. Added all congeners except 30,204,14,65,166. Dyncorp did using these rules and totals were verified
against Faye Blondin's totals.
BALP: Not reported to GLNPO and therefore not in rdmq output. Excluded 14, 65,166,103,30,204 and any
INV flagged results for first run at totals. Subsequent runs may also exclude INT, UNC values and/or
replacement of zeros with a value (my thesis work). Some samples have no pcbtot due to whole sample
invalidated due to extremely low surrogate recoveries. Dyncorp has done using these rules.
NASP: Not reported to GLNPO and therefore not in rdmq output. PI developed routine which excludes
congeners that had small peak heights: 12,13,77,119,129,130,189 and those with interference: 33,49,201
and low recoveries in QC spikes: 16, 19. All trap samples lab blank corrected on a congener by congener
50
-------
basis, negative results become zero. Low level (< 4 ng/g) ponar samples lab blank corrected if total pcbs in
lab blank is > 10% of total pcbs in sample, blank correction done on a congener by congener basis, negative
results become zero. If > 33% of blank corrected results are 0, and/or NAI, pcbtot flagged with LOB. Any lab
blank corrected pcbtot values flagged with (LBC). INV results excluded from any pcbtot calculation. Dyncorp
has done using these rules.
MNPP: As per PI:" We generally summed all congeners, and those below the MDL were assigned a value
of zero for summation purposes. HOWEVER we then applied lots of expert judgement in looking at the sums,
and applied some other criteria (for instance, if a single congener made up more than 10% of the mass of the
sum we rejected it entirely) - and if you simply sum you will definitely get different results than we are
publishing (on a significant number of samples)." PI supplied totals using below MDL values as reported and
considering them zero. I chose to have Dyncorp enter the PI totals that used the below MDL values in the total
as the biggest difference between the two methods was < 7 %.
BSFP: Not reported to GLNPO and therefore not in rdmq output. Added all congeners reported (no 14, 65,
166, 30, 204 reported for BSFP) for each RFS. No INV DATA TO BE INCLUDED.
51
-------
Spreadsheet Highlighting Differences and Similarities in How Total PCBs Were Calculated In Different
LMMBP Focus Groups
FOCUS
IUAP
WSAP
RUAP
BALP
LHTP
NASP
MNPP
.BSFP
INV
Flagged
Congeners
Excluded
Y
Y
Y
Y
Y
Y
Y
Y
Surrogate
Corrected
Y
Y
N
Y
Y
Y
Y
Y
LOB
Flagged
Congeners
Excluded*
N
N
N
N
N
N
N
N
HIB
Flagged
Congeners
Excluded*
N
N
N
N
N
N
N
N
MDL
Flagged
Congeners
Excluded*
Y
Y
Y
Y
Y
Y
Y
Y
Blank
Corrected
N
N
Y,
flagged
FBC*
N
N
Y,
flagged
LCB*,
LOB* if >
33%
results =
0 or NAI*
N
N
Other Conditions
BZ #44 excluded
in two cases
If field blank
subtraction
yielded negative
value, reverted to
0
If lab blank correc-
tion yielded neg.
value, reverted to
0. Small peak
height congeners
excluded: BZ#12,
13,77,119,129,
130, 189. Ex-
cluded congeners
with interferences:
BZ# 33, 49, 201 .
Excluded low
matrix spike re-
covery congeners:
BZ#16, 19.
"Expert judgment"
and any congener
> 1 0% total was
excluded and
retotal done
*Key to Flags: LOB = Low Bias Flag, HIB = High Bias Flag, MDL = Method Detection Limit Flag, FCB = Field
Corrected Blank, LCB = Lab Corrected Blank, NAI = Not Analyzed Due to Interference Flag
52
-------
PART1
INTRODUCTION
Appendix 1.3.8. Uncertainty of Z Values (Slope of Linear Regression Line) for PCBs
in Media Modeled at the LLRS for the LMMBP
Media
Dissolved Water
Particulate Water
Surficial Sediment
Phytoplankton
Zooplankton
Diporeia
Mysis
Alewife <120mm
Alewife > 120mm
Bloater <1 60mm
Bloater >1 60mm
Deepwater Sculpin
Slimy Sculpin
Adult Smelt
Adult Lake Trout
Z - Mean
1.53924569
1.353079101
1.17528269
1.432619553
1 .462624095
1 .3641 858
1.312835
1.439713618
1 .43439061 9
1 .479266265
1.487545716
1 .533308409
1.495781353
1.44344149
1.489208126
Z-sd
0.290911927
0.120677122
0.05661207
0.355361646
0.399167732
0.0432741
0.04129
0.140086181
0.08071193
0.049654336
0.109039931
0.089680915
0.110647788
0.166498179
0.06449037
Number
of
Samples
369
372
132
86
77
42
62
60
70
70
67
74
69
73
246
95%
Confidence
Interval Lower
1 .509305468
1 .340709374
1.16550663
1 .356363486
1.371918262
1 .3506909
1 .302347
1 .403380756
1 .41 5096777
1 .467396609
1 .46090302
1.512520571
1.469140496
1.41625314
1.481079185
95%
Confidence
Interval Upper
1.569185912
1 .365448829
1.18505875
1 .50887562
1 .553329928
1 .3776807
1 .323322
1 .47604648
1 .45368446
1.491135922
1.514188411
1 .554096246
1.52242221
1 .470629841
1 .497337067
53
-------
Coho Salmon
Hatchery
Coho Salmon -
Yearling
Coho Salmon -
Adult
1 .280547696
1.485910089
1 .44826731 7
0.028621549
0.112380374
0.074873035
5
8
54
1 .24501 5047
1.391942816
1 .427797757
1.316080345
1 .579877362
1 .468736876
Z values for the LMMBP Total PCBs Data in Various Media. Z = Y/X, where X = Sum of all modeled
congeners, as measured by Principal Investigator (PI), and Y = PI Calculated Total PCBs.
Notes from David Miller, Statistician at the LLRS:
"I used Histograms with a normal distribution overlayed upon them as well as quantile-quantile plots (Q-Q
Plots) to analyze these data sets for normality.
The assumption of normality appears to be supported for these data sets, although the following observations
should be noted.
Dissolved and particulate water, surficial sediment, phytoplankton, zooplankton, deepwater sculpin, and adult
coho, appear to exhibit some degree of kurtosis. Thus, the confidence intervals for the mean of these data
sets will be conservative.
Bloater >160mm and adult lake trout appear to have a few values at the upper end of the distribution that fall
higher than the rest of the distribution (appear under-represented according to a quantile-quantile plot).
Surficial sediments have some outlying values on the low end of the distribution as examined by the
quantile-quantile plot.
Coho salmon - Hatchery,and coho salmon Yearling have too few data to make a determination."
54
-------
PART1
INTRODUCTION
Chapter 4. Representativeness of the
LMMBP Years Relative to Lake Michigan's
Historic Record
Ronald Rossmann, Kenneth R. Rygwelski, and
Russell G. Kreis, Jr.
U.S. Environmental Protection Agency
Office of Research and Development
National Health and Environmental Effects
Research Laboratory
Mid-Continent Ecology Division
Large Lakes and Rivers Forecasting Research
Branch
Large Lakes Research Station
9311 Groh Road
Grosse lie, Michigan 48138
and
Gregory J. Gerstner, Xiaomi Zhang, and Brent
Burman
Welso Federal Services, LLC
Large Lakes Research Station
9311 Groh Road
Grosse lie, Michigan 48138
1.4.11ntroduction
A major concern related to modeling contaminants in
the lake was the representativeness of the years of
sampling (1994-1995) relative to the historical record.
This was particularly important when using the
models to predict future conditions in the lake. The
models were calibrated using 1994-1995 data
collected during the project. If these data did not
represent something close to average conditions, the
resulting predictions could be biased. Parameters
considered most important to the performance of the
models included ice cover, air temperature, water
temperature, lake water levels, precipitation, tributary
flows, and wave heights. Each of these were
investigated for the representativeness of the 1994-
1995 project data relative to the available historical
data record.
7.4.2 Ice Cover
Ice cover impacts the volatilization, absorption, and
physical mixing of the lake during the winter months.
In locations where there is ice cover, gas exchange
between the water and atmosphere is prevented by
the physical barrier. Physical mixing includes not
only the mixing of the water column but also the
interaction of waves with the lake bottom to
resuspend sediments. Winters having extensive ice
cover yield a more poorly mixed water column, and
a large region of the lake becomes depositional due
to lack of wave resuspension of sediments. Once ice
retreats in the spring, sediments accumulated during
ice cover will be resuspended as a pulse. Ice cover
can cause significant changes in winter circulation
patterns in a large lake (Campbell etal., 1987). The
years of interest that were important were 1982,
1983,1994, and 1995. The hydrodynamic modeling
included three-dimensional lake circulation, surface
flux for atmospheric input, and wind-wave models
(Schwab and Beletsky, 1998). These were calibrated
for the period of 1982-1983 using temperature,
current, water level, and wind-wave measurements.
The calibrated model was applied to 1994-1995 and
validated. There was no ice modeling component for
the version of hydrodynamic model applied. Thus,
ice cover was important for understanding any
potential weakness associated with the
55
-------
hydrodynamic results as well as the dynamics of
exchanges between the water and the atmosphere.
Ice cover data were available from the National
Oceanic and Atmospheric Administration (NOAA)/
Great Lakes Environmental Research Laboratory
(GLERL) (Assel, 2003). This data set is partially
described in Assel etal. (2002). Tabular information
presented in Assel (2003) were summarized in a
manner that seemed appropriate for this discussion
(Table 1.4.1). For the period when ice was recorded
on Lake Michigan, the mean and median daily ice
cover were 16.7% and 14.7%, respectively. Ice
years began with the first ice. For example, 1982
may include December of 1981. Both 1982 and 1994
were greater than the mean and median; whereas,
1983 and 1995 were less than the mean and median.
None of the four years represented an extreme of
mean daily ice cover. The lowest ice cover was
observed in 2002, and the highest was observed in
1977. Results for each winter's maximum daily ice
cover were similar to mean daily ice cover. None of
the four years represented an extreme of maximum
daily ice cover. As before, 1982 and 1994 were
above the mean and median, and 1983 and 1995
were below the mean and median. The maximum
mean occurred in 1977, and the minimum mean
occurred in 2002. For all four years, 1982 and 1994
were above average for number of days ice was
observed, and 1983 and 1984 were slightly below the
average. None of the four years represented a
minimum or maximum extreme. Ice cover is
extremely variable from year-to-year. The impact
upon hydrodynamics as modeled was believed to be
minimal with respect to 1983 and 1995 when ice
cover was quite low. Though high ice cover occurred
during the winters of 1982 and 1994, these periods
were not a part of the hydrodynamic model period.
Using the hydrodynamic model information for
models that are used to predict future conditions
could lead to potential errors. Modeled circulation
patterns could be in error and impact a high bias to
modeled current velocities during the winters of high
ice cover years due to the lack of an ice model within
the hydrodynamics model.
7.4.3 Water and Air Temperatures
Water and air temperature data were retrieved from
the National Data Buoy Center (U.S. Department of
Commerce, 2002). The buoy numbers are 45002
(north buoy) and 45007 (south buoy) (Figure 1.4.1).
Water temperature sensors were located 1 m below
the water surface, and air temperature sensors were
located 4 m above the surface. Water and air
temperature data were available 1979 through 2002
for the north buoy and 1981 through 2002 for the
south buoy.
Water temperature is highly variable from year-to-
year. The data had been stratified in two ways for
presentation. First, monthly mean temperatures
were calculated and plotted for the south (Figure
1.4.2) and north (Figure 1.4.3) buoys. Years of
importance to the hydrodynamic model were
highlighted. It was interesting to note that 1983 and
Figure 1.4.1. Location of the NOAA's buoys in
Lake Michigan.
56
-------
Table 1.4.1. Summary of Lake Michigan Ice Cover Based Upon Assel (2003)
Year
1973
1974
1975
1976
1977
1978
1979
1980
1981
| 1982
| 1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
| 1994
| 1995
1996
1997
1998
1999
2000
2001
2002
Mean
Median
Minimum
Maximum
Mean Daily
Ice Cover
During Ice Period
13.3
16.9
13.9
15.5
46.5
26.6
35.2
18.2
24.6
24.0
8.2
15.6
20.1
25.3
9.1
16.6
13.1
17.5
10.0
8.3
11.0
27.3
7.2
19.4
13.4
6.1
8.7
9.2
13.4
6.0
16.7
14.7
6.0
46.5
Days of
Observed
Ice
104
122
113
119
132
132
132
106
112
135
118
127
119
126
100
104
140
132
120
149
126
134
120
161
156
109
111
103
134
116
124
121
100
161
Maximum
Daily Ice
Cover
33.0
39.4
28.1
29.5
93.1
66.6
92.3
38.6
53.8
60.2 [
23.6 j
43.3
41.3
66.8
19.3
32.7
30.9
32.4
21.5
32.8
32.2
82.7 |
21.6 j
75.0
37.8
15.1
23.0
27.2
29.5
12.4
41.2
32.8
12.4
93.1
57
-------
25
20-
O
a> 15
I
(D
9-10-
5-
[>
c
•
t
south buoy
_»_ data years for Lake W
hydrodynamic modeli
»
i
*
•
li O
chigan
ng
«
t
"]'
rf
•»';
,g
-------
This, in part, explained why sampling missed the
spring 1995 diatom bloom.
The exchange of PCBs between the air and water
were dependent on both water and air temperatures.
Air temperature varied from year-to-year at the south
and north buoys (Figure 1.4.6 and 1.4.7). Because
air temperature drives observed water temperature,
it was not surprising that patterns observed and
conclusions made for water temperature are the
same for air temperature. The cyclic pattern of June
mean water temperatures was also found for the air
temperatures (Figures 1.4.8 and 1.4.9). As additional
data become available, future modeling efforts will
need to address these cyclic patterns and long-term
temperature trends for water and air temperatures.
1.4.4 Lake Water Levels
Lake levels can affect model geometry. If segment
volume deviates significantly from the volumes used
at the time of calibration, model results can be
impacted. On a percentage basis, the impact will be
most noticeable for shallow water segments and
predictions from the hydrodynamic model and
surface water model could be affected. Monthly
mean lake water levels varied between 175.5 and
177.5 m for the period of record (1918-1997). Lake
levels during 1994 and 1995 were near the average
for the period of record (Figure 1.4.10).
1.4.5 Precipitation
Precipitation influences the flux of airborne
contaminants to the lake, impacts tributary loading
rates, and controls water levels. The 1982 and 1983
hydrodynamic years, and the 1994 and 1995 project
years were compared to the previous 50 years of
data (Croley and Hunter, 1994).
1.4.5.1 Annual Comparisons
Precipitation to Lake Michigan for 1982,1983,1994,
1995 were close to the 50-year mean for the lake
(Figure 1.4.11). 1982 and 1983 were slightly above
the mean, and 1994 and 1995 were slightly below the
mean. 1995 total annual precipitation was very close
to the 50-year mean for over-lake precipitation. No
visual trend was apparent in the total annual amounts
of precipitation over the 50-year period.
1.4.5.2 Monthly Comparisons
The monthly mean precipitation for 1982, 1983,
1994, and 1995 were compared to the 50-year mean
for the period of 1949 through 1998 (Figure 1.4.12).
For the years of interest, January, July, November
and December of 1982; May of 1983; and October of
1995 had relative high amounts of precipitation,
exceeding one standard deviation of the 50-year
mean. For the four years of interest, February of
1982; June of 1983; March, May and December of
1994; and June of 1995 had relatively low amounts
of precipitation. This illustrates that, in any one year,
precipitation varies from month-to-month while the
precipitation for the year can be at or near the
average expected.
1.4.6 Tributary Flows
Tributary flows impact the delivery of materials to the
lake, including nutrients and contaminants. During
high flow events triggered by spring snow melt or rain
events, tributary flows increase and materials can be
carried from the watersheds to the tributaries. Within
thetributary, sediments containing contaminants may
resuspend. Thus, the fluxes of solids, nutrients, and
contaminants to the lake have the potential to
increase during high flow events. Tributary flows
were obtained from the United States Geological
Survey (USGS) website (www.usgs.gov). A historical
average and median daily flow were calculated for
each tributary for the period of record, as well as for
the 1994-1995 and 1982-1983 time periods. During
1982 and 1983, tributary flows were approximately
20% greater than the average flows (Figure 1.4.13).
The 1994-1995 time period had relatively ordinary
tributary flows (Figure 1.4.14).
7.4.7 Wave Heights
In Lake Michigan, waves are the driving force for the
resuspension of sediments and their associated
contaminants. As waves move from offshore to
inshore, they begin to interact with the lake bottom.
The energy associated with the waves serves to
resuspend the sediments. Lake Michigan is deep
enough such that it can be divided into three zones
based upon the potential for waves to resuspend
sediments. The zones are non-depositional,
59
-------
25
20:
§15-
0)
f 10:
m
o.
£
'S
0:
-5
south buoy
_«_ data years for Lake Michigan
hydrodynamic modeling
T—i—i—i—|—i—r
1980 1985
1990
1995
2000
2005
Figure 1.4.6. Monthly mean air temperatures in
southern Lake Michigan.
O
25
20:
15:
'10:
ID
t
£
•5 -5:
-10:
north buoy
,. data years for Lake Michigan
hydrodynamic modeling
-15-
1980
1985
1990
1995
2000
2005
25
20-
O
10-
north buoy
June
1980
1985
1990
1995
i I i
2000
2005
Figure 1.4.9. Mean June air temperatures in
northern Lake Michigan.
Lake Michigan and Lake Huron water levels (1918-1997)
data' US Army Corps ol Engineers. Detroil District
—— monthly mean water lovcb (meters - IGLD 1985}
, alt-time, record monthly hjgh and tow water levels
Figure 1.4.10. Record of mean monthly water
levels for Lake Michigan.
Figure 1.4.7. Monthly mean air temperatures in
northern Lake Michigan.
20
16-
O
%
?
55
o.
4>
12-
4-
south buoy
June
1980
1985
1990
1995
2000
2005
Figure 1.4.8. Mean June air temperatures in
southern Lake Michigan.
f\
50 y»at annual mMnj
-total annual precipilation
Figure 1.4.11. Annual precipitation to Lake
Michigan between 1949 and 1998.
60
-------
1200 -
1000 -
600 •
200
Monthly mean precipitation
—*— 1949-1998mean
- -+- - 1949-1998 mean + 1 std.dev.
•- -1949-1998 mean - 1 std.dev
o 1982
—a—1983
o—1994
— o— 1995
JAN FEB MAR APR MAY JUN JUL AUG SEP OCT NOV DEC
Figure 1.4.12. Comparison of 1982,1983,1994, and 1995 monthly mean precipitation to the means for
the period of 1949 through 1998.
bUUU'
5000-
4000'
CD
£ 3000
o
2000
1000'
ni
1
p.
f
Tributary flow
c historical mean
• 1982-83 mean
n 1982 mean
01983 mean
N r-«-n c-«
rn
.lfH.rfl.Ffnj I j
n
(n jn[ ]
Ifli;
.MII.J / rrn
'•s i
2"
ra
'^ 9
-^ TO
Figure 1.4.13. Comparison of tributary flow for hydrodynamic model calibration (1982-1983) to the
historic means.
61
-------
6000"
5000-
4000'
-------
following equation (U.S. Army Coastal Engineering
Research Center, 1973).
'max
where,
2TTL
(1.4.2)
H = wave height, cm
For both 1994 and 1995, mean wave heights at the
northern and southern buoys were similar to historic
mean and median observations (Table 1.4.2).
Maximum wave heights for these years were high at
the northern buoy relative to the historical mean and
median heights for both years; however, they were
less than the historical maximum wave height at both
locations. At the southern buoy, maximum wave
heights were high in 1982,1983, and 1995 relative to
historic mean and median maximum heights. At this
location, the 1994 maximum height was lower than
historic means and medians. For all calibration and
study years, the maximum wave heights were neither
an extreme high or low for the period of observation
at the two locations.
At both buoys, the annual maximum water depths of
wave interaction with sediments were not unusual.
Though at the maximum in the southern basin (78
m), the maximum was achieved for more than one-
half the years of record (Table 1.4.2).
During the calibration and study years, the annual
maximum horizontal component of wave orbital
velocity ranged between 11.2 and 15.0 cm/s (Table
1.4.2). These observations were close to their
historical means. Chambers and Eadie (1981)
hypothesized that thermal bar migration generated
currents of four to 13.4 cm/s which were enough to
resuspend surficial shelf sediment. For fine/medium
sand in southwestern Lake Michigan, a near-bottom
wave orbital velocity of 17.8 cm/s was enough to
initiate resuspension (Lesht, 1989). Similar results
(18 cm/s) were found for silty sand in southeastern
Lake Michigan (Lesht and Hawley 1987). Sediment
resuspension was found in Hamilton Harbor at
bottom current speeds of 4.8 cm/s (Brassard and
Morris, 1997). Thus the annual maximum horizontal
component of wave orbital velocity was sufficient to
at least once, on an annual basis, resuspend fine-
grained (silt and clay) sediment.
Therefore, it appeared that the repetitious use of the
1994 through 1995 Princeton Ocean Model (POM)
results will not introduce bias to the results. Though
1994 and 1995 were not the perfect mean and
median situation, they were not singular extremes of
what has historically occurred within the lake.
1.4.8 Summary
Lake Michigan is acted upon by a number of physical
parameters that impact the physics, chemistry, and
biology of the lake. For a lake the size of Lake
Michigan, changes in these parameters can lead to
significant changes, especially when models are
used in long-term predictions to predict the outcome
of various scenarios. The primary driving forces are
wind, air temperature, and precipitation. These
impact tributary flows, lake levels, waves, water
circulation, water temperature, and ice cover. For the
period of record, these driving forces varied from
year-to-year. The period of 1982 to 1983 was used
to calibrate the hydrodynamic models. Fortunately
for the period of time the models were calibrated,
conditions were not at any extreme. This was also
true for the period of 1994 and 1995 when the
models were applied. However, the impact of ice
cover remains a concern and will have to be dealt
with in the future.
Temperature can impact atrazine modeling. Air
temperature impacts how quickly the lake warms in
any one year. Water temperature impacts the
volatilization of atrazine. There appears to be a four-
year cycle of quicker warming which exists within a
trend of general warming of the lake. The trend of
warming may be part of a longer term undocumented
cycle or may be related to climate change. For future
modeling, these cycles and trends will have to be
considered to improve long-term predictions.
Precipitation will impact both lake levels and tributary
flows. Tributary flows have an impact on the delivery
of contaminants to the lake. Precipitation was within
the normal range for all years of modeling interest,
resulting in lake levels and tributary flows that were
within normal bounds. Changes in lake levels as well
as the response of tributaries to precipitation events
will need to be considered for future modeling used
to predict changes of contaminants within the lake.
63
-------
Table 1.4.2. Descriptive Wave Statistics for POM Calibration Years (1982-1983) and Study Years (1994-
1995) Compared to the Period of Record for NOAA's Buoys in Northern and Southern Lake Michigan
Historical Historical
Description Historical Historical Minimum Maximum
Mean Median (Year) (Year)
1982
1983
1994
1995
Northern Buoy
Annual Maximum Wave 4.3 4.1
Height, m
Wave Height, m 0.7 0.6
Annual Maximum Water
Depth of Wave Interaction 14.1 65
with Bottom Sediments, m
Annual Horizontal
Component of Maximum 4.9 13.4
Orbital Wave Velocity,
cm/s
3.1(1996) 5.9(1991) 4.0
4.5
4.6
0.6 0.9
(numerous) (several)
5.3
0.8 0.9 0.6 0.7
(Mean) (Mean) (Mean) (Mean)
96(1991) 65 78 78 78
18.0(1984) 12.1 11.3 12.7 15.0
Southern Buoy
Annual Maximum Wave 4.5 4.2
Height, m
Wave Height, m 0.6 0.5
Annual Maximum Water
Depth of Wave Interaction 12.7 65
with Bottom Sediments, m
Annual Horizontal
Component of Maximum 4.6 13.6
Orbital Wave Velocity,
cm/s
2.8(1991) 6.2(1998) 4.9
5.3
3.7
78
(numerous)
65
19.2(1998) 13.1
78
15.0
65
11.2
5.2
0.5 0.8 0.8 0.8 0.6 0.5
(several) (numerous) (Mean) (Mean) (Mean) (Mean)
78
14.7
References
Assel, R.A., D.C. Norton, and K.C. Cronk. 2002. A
Great Lakes Digital Ice Cover Data Base for
Winters 1973-2000. National Oceanic and
Atmospheric Administration, Great Lakes
Environmental Research Laboratory, Ann Arbor,
Michigan. NOAA Technical Memorandum ERL
GLERL-121,46pp.
Assel, R.A. 2003. An Electronic Atlas of Great
Lakes Ice Cover. NOAA Great Lakes Ice Atlas.
National Oceanic and Atmospheric
Administration, Great Lakes Environmental
Research Laboratory, Ann Arbor, Michigan.
Brassard, P. and W. Morris. 1997. Resuspension
and Redistribution of Sediments in Hamilton
Harbor. J. Great Lakes Res., 23(1):74-85.
Campbell, J.E., A.M. Clites, and G.M. Green. 1987.
Measurements of Ice Motion in Lake Erie Using
Satellite-Tracked Drifter Buoys. National Oceanic
and Atmospheric Administration, Great Lakes
Environmental Research Laboratory, Ann Arbor,
Michigan. NOAA Technical Memorandum ERL
GLERL-30, 22 pp.
64
-------
Chambers, R.L and B.L Eadie. 1981. Nepheloid
and Suspended Particulate Matter in
Southeastern Lake Michigan. Sedimentology,
28(3):439-447.
Croley, I.E., II and T.S. Hunter. 1994. Great Lakes
Monthly Hydrologic Data. National Oceanic and
Atmospheric Administration, Great Lakes
Environmental Research Laboratory, Ann Arbor,
Michigan. NOAA Technical Memorandum ERL
GLERL-83, 13pp.
Lesht, B.M. and N. Hawley. 1987. Near-Bottom
Currents and Suspended Sediment
Concentration in Southeastern Lake Michigan. J.
Great Lakes Res., 13(3):375-386.
Lesht, B.M. 1989. Climatology of Sediment
Transport on Indiana Shoals, Lake Michigan. J.
Great Lakes Res., 15(3):486-497.
Schwab, D.J. and D. Beletsky. 1998. Lake Michigan
Mass Balance Study: Hydrodynamic Modeling
Project. National Oceanic and Atmospheric
Administration, Great Lakes Environmental
Research Laboratory, Ann Arbor, Michigan.
NOAA Technical Memorandum ERLGLERL-108,
53pp.
Sverdrup, H.U., M.W. Johnson, and R.H. Fleming.
1942. The Oceans: Their Physics, Chemistry,
and General Biology. Prentice-Hall,
Incorporated, New Jersey, 1,087 pp.
U.S. Army Coastal Engineering Research Center.
1973. Shore Protection Manual, Volumes 1, 2,
and 3. U.S. Army Corps of Engineers, U.S. Army
Engineering Waterways Experiment Station,
Vicksburg, Mississippi.
U.S. Department of Commerce. 2002. National
Data Buoys. National Weather Service, National
Oceanic and Atmospheric Administration, Ann
Arbor, Michigan. National Data Buoy Center files
downloaded from www.ndbc.noaa.gov.
65
-------
PART1
INTRODUCTION
Chapter 5. PCBs in the Lake Michigan
Ecosystem
Ronald Rossmann
U.S. Environmental Protection Agency
Office of Research and Development
National Health and Environmental Effects
Research Laboratory
Mid-Continent Ecology Division
Large Lakes and Rivers Forecasting Research
Branch
Large Lakes Research Station
9311 Groh Road
Grosse lie, Michigan 48138
1.5.1 Introduction
Within this chapter, polychlorinated biphenyl (PCB)
data for Lake Michigan are summarized. These data
are then compared to historical data. Summaries of
the 1994-1995 data are taken directly from the
"Results of the Lake Michigan Mass Balance Study:
Polychlorinated Biphenyls and frans-Nonachlor Data
Report" (McCarty et al., 2004) and are referenced
appropriately. Presentation of the data are arranged
by media.
Details of the use of PCBs within the Lake Michigan
basin are difficult to document. However, a list of
significant dates for the basin was compiled from a
variety of online sources. PCBs appear to have been
first purchased for use in the basin in 1948 (Table
1.5.1). Use appears to have been phased out by
1972. Remediation of PCBs at various locations
began in the late 1980s with Sheboygan Harbor
being the earliest completed in 1991. Remediation
continues to this day.
1.5.2 Atmospheric
Atmospheric concentrations measured include vapor
phase, precipitation, particulate, and dry deposition
PCBs. Each of these is plotted for stations within the
basin that are representative of the annual seasons
for 1994 and 1995.
1.5.2.1 Vapor Phase
Median vapor phase concentrations of PCBs are
elevated in the northern end of the lake and
especially offshore of Chicago (Figure 1.5.1). The
elevation in the northern end of the lake is related to
the air station at Beaver Island. The site was
discovered to be impacted by a local source of
contamination and was not representative of that
region of the lake.
Complete data are summarized in Table 1.5.2.
Monthly mean composite concentrations ranged from
110 pg/m3 outside the basin to 2600 pg/m3 at IIT
Chicago (McCarty et al., 2004).
Historic information are sparse. Data sets vary in
temporal and spatial scales. Data sets found include
the period of 1976, 1979, and 1992-2002 (Murphy
and Rzeszutko, 1977; Rice et al., 1982; The
Integrated Atmospheric Deposition Network, 2004).
The Integrated Atmospheric Deposition Network
(IADN) data for 2001 and 2002 were presented in a
graphic by Boughton (2004) and attributed to the
IADN Steering Committee, unpublished 2004. Vapor
66
-------
Table 1.5.1. Significant Dates in the History of PCBs in the Lake Michigan Basin
Date Event
1865 First PCB-like chemical discovered
1881 First PCBs synthesized
1914 Measurable amounts of PCBs found in bird feathers
1927 PCBs first manufactured at Anniston, Alabama
1935 PCBs manufactured at Anniston, Alabama and Sauget, Illinois
1948-1971 Outboard Marine Corporation at Waukegan, Illinois purchased eight million
gallons of hydraulic fluid with PCBs
Mid-1950s to Mid-1960s PCBs loaded to Kalamazoo River from deinking
1950s to 1980s PCBs discharged to Manistique River and Harbor
1954 Appleton Paper Company began using PCBs as PCB-coated carbonless
copy paper
1959-1971 PCBs used by Tecumseh Products Company as a hydraulic fluid was loaded
to Sheboygan River
1959-1972 Outboard Marine Corporation at Waukegan, Illinois used hydraulic fluid with
PCBs for die-casting
1969-1970 Paper company discharges of PCBs to Fox River peaked
1970 PCB production peaked at 85 million pounds and huge contamination noted
at Sauget, Illinois plant
1971 -1972 Appleton Paper Company and NCR Corporation phased out PCB use.
Recycling of carbonless paper had occurred for several decades
1973 U.S. Food and Drug Administration (USFDA) establish 5 ppm PCB tolerance
level in fish
1975 124,000 cans of salmon from Lake Michigan seized because of PCBs
1977 PCB production ends
1984 USFDA lowered PCB tolerance level in fish to 2 ppm
1985 Commercial fishing for carp and other valuable species outlawed on Green
Bay
1991 End Sheboygan River PCB remediation of upper river
1991 U.S. Department of Health and Human Services label PCBs as possible
carcinogen
1992 End Waukegan Harbor PCB remediation
1998 The eight Great Lakes states agreed on a "Great Lakes Protocol for Fish
Consumption Advisories" that lowered the regional standard from the
USFDA commercial standard of 2 ppm down to 0.05 ppm
1997-1998 Milwaukee River PCB remediation
2001 Manistique Harbor PCB remediation completed
2002 Possibly begin Grand Calumet River PCB remediation
67
-------
Naubinway
Manistic
Escanab
Scale
0 km 40 km 80 km
Menominee
Charlevoix
Green Bay
Manitowoc,
Sheboyga
Milwaukee
Racine
I
Waukegan
Chicago^
Mackinaw City
Muskegon
Grand Haven
Saugatuck
South Haven
Benton Harbor
Gary
Michigan City
Figure 1.5.1. Median concentration of vapor phase PCBs in the atmosphere during 1994 and 1995 for
all seasons of both years.
68
-------
Table 1.5.2. Monthly Composite Concentrations of Vapor Phase Total PCBs Measured in Samples
Collected Around Lake Michigan From April 1994 to October 1995
Sampling Station
N Mean (pg/m3) SD (pg/m3)
Shoreline Atmospheric Beaver Island
Stations Chiwaukee Prairie
I IT Chicago
Indiana Dunes
Manitowoc
Muskegon
Sleeping Bear Dunes
South Haven
Out-of-Basin Atmospheric Bondville
Stations Brule River
Eagle Harbor
Over-Water Atmospheric Empire Michigan
Stations GB24M
1
5
6
110
18M
23M
27M
280
310
380
40 M
41
47M
MB19M
11M
19
19
19
19
19
18
15
19
19
19
4
4
4
5
6
4
3
5
4
5
4
3
1
4
2
5
1
2
970
320
2600
680
350
490
380
400
250
110
260
170
940
990
670
1200
810
560
490
360
480
650
290
340
21
410
280
2200
880
230
1900
580
260
410
550
360
150
110
370
77
1600
1500
590
1800
1200
930
600
410
570
430
NA
460
6.4
630
NA
3100
phase PCBs have decreased dramatically since the
very first measurements in 1976 (Figure 1.5.2);
however, the IADN data set vapor phase
concentrations at Sleeping Bear Dunes does not
show a definitive trend between 1992 and 2002
(Figure 1.5.3). For the period of 1992 through 2001,
an examination of the temperature corrected PCB
partial pressure IADN data revealed that partial
pressures were declining with a half-life rate of 8.3 ±
1.5 years (Buehler et al., 2004). The 1994-1995
concentrations measured during the Lake Michigan
Mass Balance Project (LMMBP) are considerably
higher (21 to 2,600 pg/m3) than concentrations
observed at Sleeping Bear Dunes (50 to 110 pg/m3).
Thus it appears that Sleeping Bear Dunes does not
represent the Lake Michigan basin. Concentrations
observed at the Chicago IADN station were greater
than 1,000 pg/m3 for all the years (1993-2000) of
observation (Buehler and Hites, 2002). Recently, an
average concentration of 1,900 pg/m3 PCBs was
reported for June 2001 at Milwaukee, Wisconsin
(Wethington and Hornbuckle, 2005).
1.5.2.2 Precipitation
Median precipitation concentrations of PCBs are
elevated in the southern end of the lake and
69
-------
8000
1970 1975 1980 1985 1990 199520002005
'Year
Figure 1.5.2. Time variation of vapor phase PCBs in Lake Michigan.
250
1990
1995 2000
Year
2005
Figure 1.5.3. Time variation of vapor phase PCBs in Lake Michigan at Sleeping Bear Dunes based on
IADN data.
70
-------
especially offshore of Chicago (Figure 1.5.4).
Concentrations decline from south-to-north.
Complete data are summarized in Table 1.5.3.
•Monthly mean composite concentrations ranged from
290 pg/L outside the basin to 16,000 pg/L at I IT
Chicago (McCarty et al., 2004).
Historic concentrations of RGBs in precipitation were
only available from IADN (2004). The time trend of
the data are plotted in Figure 1.5.5. There is no
apparent trend in the data. Again, the 1994-1995
basin results (360 to 16,000 ng/L) are higher than
those reported by IADN for Sleeping Bear Dunes
(955 to 2,849 ng/L). Sleeping Bear Dunes may
under-represent the PCB concentration in
precipitation. Chicago concentrations ranged
between 3,500 and 8,600 ng/L for the 1996-2000
period of observation (IADN, 2004).
1.5.2.3 Participate
Median atmospheric particulate concentrations of
PCBs are elevated in the southern end of the lake
and especially offshore of Chicago (Figure 1.5.6).
Concentrations decline from south-to-north and then
increase slightly in the northern end of the lake. The
elevation in the northern end of the lake is related to
the air station at Beaver Island. The site was
discovered to be impacted by a local source of
contamination and was not representative of that
region of the lake.
Complete data are summarized in Table 1.5.4.
Monthly mean composite concentrations ranged from
0.37 pg/m3 at an over-water station to 91 pg/m3 at NT
Chicago (McCarty et al., 2004).
Historic particulate concentrations of PCBs were only
available from IADN (2004). The time trend of the
data are plotted in Figure 1.5.7. Again, the 1994-
1995 basin results (0.37 to 91 pg/m3) are higher than
those reported by IADN at Sleeping Bear Dunes (7.8
to 9.2 pg/m3). Sleeping Bear Dunes under-
represents the PCB concentration in particulates.
Samples collected in June 2001 at Milwaukee
averaged 50 pg/m3 (Wethington and Hornbuckle,
2005).
1.5.2.4 Dry Deposition
Some measurements were made of PCB
concentrations in dry deposition (Table 1.5.5).
Highest mean concentration (315,000 ng/m2)
occurred at South Haven and the lowest mean
concentration (1,830 mg/m2) occurred at the Chicago
SWFP crib intake (McCarty et al., 2004). Thus the
mean range was very large and fluxes could be high.
However, the limited number of data points and high
standard deviations precluded any further
interpretation of the data.
1.5.3 Lake Water
Water concentrations measured included dissolved,
particulate, and total PCBs. Results are plotted for
the 1994 through 1995 stations. Only stations
representative of all annual seasons were included in
the figures.
1.5.3.1 Total PCBs
Total PCB concentrations in lake water were elevated
in southern and central Lake Michigan (Figure 1.5.8).
Of note were high concentrations offshore of
Milwaukee and along the eastern shoreline from
South Haven to Manistee. Highest concentrations
were found along the shoreline of the lake.
Concentrations are summarized in Table 1.5.6 and
averaged 0.25 ppb in 1994 and 0.27 ppb in 1995
(McCarty et at., 2004).
Historic data were available for the period of 1976
through 1994 (Chambers and Eadie,1980; Rice etal.,
1982; Anderson et al., 1999, Offenberg and Baker,
2000; Bicksler, 1996; Murphy and Rzeszutko, 1977;
Swackhamer and Armstrong, 1987; Filkins et al.,
1983; Lefkovitz, 1987; Pearson et al., 1996). Since
1976, concentrations have dramatically decreased
(Figure 1.5.9). Expanding the scale and looking only
at the 1986 and later data, concentrations decrease
from 1986 to 1993; however, they appear constant
since 1993 (Figure 1.5.10).
1.5.3.2 Dissolved PCBs
Dissolved PCBs had a pattern similar to that of total
PCBs. Concentrations were highest in the southern
basin of the lake with special note of the Milwaukee,
Michigan City, Saugatuck, and Grand Haven
71
-------
Naubinway
Manistique
Escanaba
Scale
0 km 40 km 80 km
Menominee
Charlevoix
Green Bay
Manitowoc
Sheboygan,
Milwaukee
Racine
Waukegan
'Mackinaw City
Muskegon
Grand Haven
Saugatuck
South Haven
Benton Harbor
Chicago
Gary
Michigan City
Figure 1.5.4. Median concentration of wet (precipitation) PCBs in the atmosphere during 1994 and
1995 for all seasons of both years.
72
-------
Table 1.5.3. Monthly Composite Concentrations of Total PCBs Measured in Precipitation Samples
Collected Around Lake Michigan From April 1994 to October 1995
Sampling Station
Shoreline Atmospheric Beaver Island
Stations Chiwaukee Prairie
NT Chicago
Indiana Dunes
Manitowoc
Muskegon
Sleeping Bear Dunes
South Haven
Out-of-Basin Atmospheric Bondville
Stations Brule River
Eagle Harbor
Over-Water Atmospheric Empire Michigan
Stations GB17
GB24M
1
5
23M
380
3000 n
2400 -
^1800 -
Q.
CD
01200 -
Q_
600 -
0 -
19'
: li
: A
/ \
/ I A
^ ^^
30 1995 2000
N Mean (pg/L)
20 1900
20 1800
17 16000
21 1500
20 2600
20 2600
16 1300
21 3800
21 1700
19 1700
4 290
4 2000
1 2300
1 680
1 750
1 1500
1 360
1 510
2005
SD (pg/L)
2800
1200
28000
1500
4200
4000
880
10000
1100
2900
300
2000
Year
Figure 1.5.5. Time variation of precipitation PCBs in Lake Michigan at Sleeping Bear Dunes based on
IADN data.
73
-------
Naubinway
Manistique
Escanaba.
Scale
••=
0 km 40 km 80 km
Menominee
Charlevoix
Green Bay
Manitowoc
Sheboyga
Milwaukee
Racine
Waukegan
'Mackinaw City
Muskegon
Grand Haven
Saugatuck
South Haven
Benton Harbor
Chicago
Gary
Michigan City
Figure 1.5.6. Median concentration of particulate PCBs in the atmosphere during 1994 and 1995 for
all seasons of both years.
74
-------
Table 1.5.4. Monthly Composite Concentrations of Particulate Phase Total PCBs Measured in Samples
Collected Around Lake Michigan From April 1994 to October 1995
Sampling Station
Shoreline Atmospheric Beaver Island
Stations Chiwaukee Prairie
I IT Chicago
Indiana Dunes
Manitowoc
Muskegon
Sleeping Bear Dunes
South Haven
Out-of-Basin Atmospheric Bondville
Stations Brule River
Eagle Harbor
Over-Water Atmospheric Spatial Composites
Stations Empire Michigan
GB24M
1
5
6
10 -
8 -
CO
£ 6-
D)
Q.
CQ 4 .
0 ^
Q_
2 -
0 -
19!
: \
: /
: /
: "j
90 1995 2000
N Mean (pg/m3)
18 52
19 22
19 91
19 33
19 26
16 24
15 18
18 23
19 25
18 21
4 14
18 19
4 14
2 3.9
1 2.6
3 17
1 0.37
2005
SD (pg/m3)
29
6.1
48
12
22
12
21
12
14
14
4.7
21
4.0
5.2
27
Year
Figure 1.5.7. Time variation of atmospheric particulate PCBs in Lake Michigan at Sleeping Bear Dunes
based on IADN data.
75
-------
Table 1.5.5. Monthly Composite Concentrations of PCBs Measured in Dry Deposition
Sampling Station
Chicago SWFP Crib Intake
Harrison Crib
NT Chicago
Sleeping Bear Dunes
South Haven
N
9
1
13
8
11
Mean (ng/m2)
1830
5400
7060
6120
315,000
SD (ng/m2)
1710
6480
7940
1 ,020,000
NI
Scale
0 km 40 km 80 km
Menominee
Green Bay
Manitowoc
Sheboygan)—^'
Milwaukee
Racine
Waukegan
Chicago1
Manistique
Naubjnway
>^ fCharlevoix
Mackinaw City
Gary
?\Muskegon
Grand Haven
Saugatuck
South Haven
/& / Benton Harbor
'Michigan City
Figure 1.5.8. Distribution of total PCBs (ng/L) in 1994-1995 Lake Michigan water.
76
-------
Table 1.5.6. Concentrations of PCBs in 1994-1995 Lake Michigan Water (ng/L)
Year
Descriptive Statistics
Dissolved PCBs Particulate PCBs Total PCBs
1994
1994
1994
1995
1995
1995
Mean
Standard Deviation
Number of Samples
Mean
Standard Deviation
Number of Samples
0.17
0.11
181
0.21
0.16
142
0.079
0.088
181
0.066
0.060
142
0.25
0.16
181
0.27
0.19
142
1975 1980
1985 1990
Year
1995 2000
Figure 1.5.9. Time variation of total PCBs in Lake Michigan water. Historic data from Chambers and
Eadie (1980), Rice et al. (1982), Anderson et al. (1999), Offenberg and Baker (2000), Bicksler (1996),
Murphy and Rzeszutko (1977), Swackhamer and Armstrong (1987), Filkins et al. (1983), Lefkovitz
(1987), and Pearson et al. (1996).
77
-------
1985
1990
1995
2000
Year
Figure 1.5.10. Time variation of total PCBs in Lake Michigan water since 1986. Historic data from
Anderson etal. (1999), Offenberg and Baker (2000), Bicksler (1996), Lefkovitz (1987), and Pearson et
al. (1996).
locations (Figure 1.5.11). Green Bay appears as a
source of dissolved PCBs to the lake. For the lake
as a whole, concentrations averaged 0.17 ppb with a
standard deviation of 0.11 ppb in 1994 and 0.21 ppb
with a standard deviation of 0.16 ppb in 1995 (Table
1.5.6) (McCarty et al., 2004).
During the summer thermal stratification, dissolved
PCB concentrations in the hypolimnion are notably
high offshore of the region bounded by Racine and
Michigan City and centered on Waukegan (Figure
1.5.12). This region appears to continue to be
impacted by the historical contamination at
Waukegan (Swackhamer and Armstrong, 1988).
Dissolved PCB data were found for the period of
1991 through 1994 (Anderson et al., 1999; Offenberg
and Baker 2000; Pearson et al., 1996). What few
data are available illustrate a downward trend in
concentration (Figure 1.5.13). This is similar to the
trend observed for total PCBs.
1.5.3.3 Particulate PCBs
Particulate PCB concentrations were highest along
the shoreline, notably at Saugatuck and especially at
Milwaukee (Figure 1.5.14). Concentrations averaged
0.079 ppb with a standard deviation of 0.088 ppb in
1994 and 0.066 ppb with a standard deviation of
0.060 ppb in 1995 (Table 1.5.6) (McCarty et al.,
2004).
Particulate PCB data were derived from the
difference between total and dissolved
concentrations for the period of 1991 through 1994
from the data of Anderson et al. (1999), Offenberg
and Baker (2000), and Pearson et al. (1996). The
variation of these and the project data is one of a
decreasing trend (Figure 1.5.15).
1.5.4 Tributaries
Tributary water concentrations measured included
dissolved and particulate forms. Mean particulate
concentrations were higher than mean dissolved
concentrations in the Fox, Grand Calumet, Grand,
Kalamazoo, Pere Marquette, Sheboygan, and St.
Joseph Rivers (Table 1.5.7) (McCarty et al., 2004).
Concentrations of dissolved PCBs were highest in
the Grand Calumet River and lowest in the Pere
Marquette River. Concentrations of particulate PCBs
were highest in the Sheboygan River and lowest in
the Muskegon River. Loads of total PCBs to Lake
Michigan were highest for the Fox River and lowest
for the Manistique River (Figure 1.5.16).
One historic data set for PCBs in Lake Michigan
tributaries was found (Marti and Armstrong, 1990).
These data permitted comparison of the Fox, Grand
Calumet, Grand, Kalamazoo, Manistique,
Menominee, Milwaukee, Muskegon, Pere Marquette,
Sheboygan, and St. Joseph Rivers for dissolved and
particulate PCB concentrations. Between 1980 and
1995, dissolved PCB concentrations decreased at all
78
-------
Manistique
Naubinway
Escanaba
Scale
0 km 40 km 80 km
Me nominee
Charlevoix
Green Bay
Manitowoc
Sheboyga
Milwaukee
Racine
Waukegan
Mackinaw City
Muskegon
Grand Haven
Saugatuck
South Haven
Benton Harbor
Chicago
Gary
Michigan City
Figure 1.5.11. Distribution of dissolved PCBs (ng/L) in 1994-1995 Lake Michigan water.
79
-------
Naubinway
Escanaba.
Scale
0 km 40 km 80 km
Menominee
Green Bay
Manitowoc,
Sheboyga
Milwaukee
Racine
Waukegan
Manistique
Chicago
Charlevoix
Mackinaw City
Muskegon
Grand Haven
Saugatuck
South Haven
Benton Harbor
Gary
Michigan City
Figure 1.5.12. Distribution of dissolved PCBs (ng/L) in 1994-1995 summer hypolimnetic Lake Michigan
water.
80
-------
0.75
0.6 -
•B>0.45 --
DO
O 0.3
a.
0.15 -
0
1990 1991 1992 1993 1994 1995 1996
Year
Figure 1.5.13. Time variation of dissolved PCBs in Lake Michigan water.
Naubinway
Manistiq
Escanaba.
Scale
^•=n
0 km 40 km 80 km
Menominee
Green Bay
Manitowoc.
Sheboyga
Milwaukee
Racine
Waukegan
Chicago^
Charlevoix
'Mackinaw City
Pentwater
Muskegon
Grand Haven
Saugatuck
South Haven
Benton Harbor
Michigan City
Gary
Figure 1.5.14. Distribution of particulate PCBs (ng/L) in 1994-1995 Lake Michigan water.
81
-------
-0.1
1990 1991 1992 1993 1994 1995 1996
Year
Figure 1.5.15. Time variation of particulate PCBs in Lake Michigan water.
Table 1.5.7. Concentrations of PCBs Measured in Tributaries
Fraction
Tributary
N
Mean (ng/L) SD (ng/L)
Dissolved Fox River
Grand Calumet
Grand River
Kalamazoo
Manistique
Menominee
Milwaukee
Muskegon
Pere Marquette
Sheboygan
St. Joseph
Particulate Fox River
Grand Calumet
Grand River
Kalamazoo
Manistique
Menominee
Milwaukee
Muskegon
Pere Marquette
Sheboygan
St. Joseph
39
15
47
38
28
24
38
28
28
36
33
39
15
47
38
28
24
38
28
28
36
33
14
35
0.76
6.9
0.76
1.4
13
0.58
0.43
26
1.0
39
41
1.6
16
0.41
0.52
11
0.25
0.47
55
1.9
7.6
6.5
0.35
2.1
0.39
2.1
4.0
0.40
0.19
8.3
0.53
25
22
0.63
9.6
0.37
0.27
6.2
0.14
0.32
31
0.98
82
-------
0.7-
-50.6-
TO
TJ
jj>0.5-
.f 0.4-
ra
00.3-
-Q
U
Q-0.2-
0.1 -
n -
II..
x o ^^ £ ^ c fl) o c
o o c2 a. c ra 0 « o fcS =
u. Neacv ra a) ^ c a>£ocr
E O-= o O o g E J crtS
.20? •glg^iS'c
5 S5 ss^ssi
Figure 1.5.16. Relative loads of PCBs to Lake Michigan from tributaries.
but one of the tributaries that could be compared
(Table 1.5.8). Particulate PCB concentrations
declined at all locations between 1980 and 1995
(Table 1.5.8).
1.5.5 Sediment
PCB concentrations in Lake Michigan surficial (0.5 to
1.5 cm in thickness) sediments ranged between
0.066 and 220 ng/g for samples collected between
1994 and 1996 (Table 1.5.9). This is similar to the
range reported by Swackhamer and Armstrong
(1988) for samples collected between 1978 and
1980. The mean concentration in 1994-1996 was 47
ng/g. This is within the range of means reported by
Swackhamer and Armstrong (1988) for depositional
(81 ng/g), transitional (26 ng/g), and non-depositional
(7.2 ng/g) regions of the lake as defined by Cahill
(1981). All of these results are considerably higher
than those reported by Frank et al. (1981) for 1975
(Table 1.5.9). Results of the 1975 samples
represented samples that had been freeze-dried.
Freeze-drying can result in the loss of volatile
contaminants from the sediment samples. The
spatial variation of PCBs in the main lake surficial
sediments in 1994-1996 is consistent with that
reported by Frank et al. (1981) and Swackhamer and
Armstrong (1988). Concentrations are elevated in
southeastern Lake Michigan, offshore of Sheboygan,
and offshore of Frankfort (Figure 1.5.17). Of these
three areas, PCB concentrations are highest in
southeastern Lake Michigan.
Currently, LMMBP PCB results are available for six
sediment cores (Figure 1.5.18) (Eadie and Van Hoof,
personal communication). All of these cores illustrate
the decline in PCB concentrations in recent times.
The cores are of varying resoluteness due to the
thickness of the surface mixed layer which results in
varying particle residence times in the mixed layer
(Table 1.5.10). Historic records indicate the first use
of PCBs in the basin began no earlier than 1948.
PCBs occurring in cores prior to 1948 represent
vapor phase PCBs transported to the basin, the
failure of sealed sources within the basin, or physical
processes within the lake associated with the surface
mixed layer of the cores. The surface mixed layer is
a surficial zone of the sediment that is consistently
mixed by physical or biotic processes. This mixing
homogenizes the sediment to a given depth referred
to as the mixed layer. The residence time of a
particle and its associated PCBs varies from core to
core (Bobbins et al., 1999). For the six cores this
residence time varies from 0.0 to 31.5 years. Thus
some cores are highly resolved (LM-94-15) and
some are poorly resolved (LM-95-58). This results in
differences for the apparent time of appearance of
PCBs in the cores.
83
-------
Table 1.5.8. Comparison of PCB Concentrations in Samples Collected From Tributaries in 1994-1995
With Those in Samples Collected From Tributaries in 1980-1983 (Marti and Armstrong, 1990)
1980-1983 1994-1995 1980-1983 1994-1995
Dissolved Mean Dissolved Mean Particulate Mean Particulate Mean
Concentration Concentration Concentration Concentration
Tributary (ng/L) (ng/L) (ng/L) (ng/L)
Fox River
Grand Calumet
Grand River
Kalamazoo
Manistique
Menominee
Milwaukee
Muskegon
Pere Marquette
Sheboygan
St. Joseph
17
24
16
9
6
6
28
4
4
34
7
14
35
0.76
6.9
0.76
1.4
13
0.58
0.43
26
1.0
81
220
41
31
18
9
69
5
10
69
7
39
41
1.6
16
0.41
0.52
11
0.25
0.47
55
1.9
Table 1.5.9. Concentrations of Total PCBs in Lake Michigan Surficial Sediment (ng/g)
Year Number of Standard
Collected Samples Mean Deviation Minimum Maximum Median Source
1994-
1996
113
47
48
0.066
220
29
LMMBP
179 9.7 15.7 Frank et a/., 1981
1978- 60 7.2-81 1.0 201 Swackhamer and
1980 Armstrong, 1988
84
-------
t
N Escanaba,
Scale
^•czi
0 km 40 km 80 km
Menominee
Green Bay
Manitowoc.
Sheboyga,
Milwaukee
Racine
Waukegan
Chicago*
Manistique
Naubinway
Charlevoix
Mackinaw City
Gary
Muskegon
Grand Haven
Saugatuck
South Haven
' Benton Harbor
^Michigan City
Figure 1.5.17. Total PCBs in 1994-1995 Lake Michigan surficial sediments (ng/g).
D)
150 :
50
AB
--- A--
0 T i i—i—i—i—i—i—M-
2000 1980 1960 1940 1920 1900 1880
Year
• LM-94-15 V LM-95-86 A LM-95-61
Figure 1.5.18. Vertical variation of PCBs in dated sediment cores collected for the LMMBP.
85
-------
Table 1.5.10. Physical Parameters Associated With LMMBP Cores (Bobbins etal., 1999; Eadieand Van
Hoof, Personal Communication)
Station Number
Sedimentation
Rate g/cm2/year
Mixed Layer
Residence Time,
Year
Year of Peak PCB
Concentration
Year of PCB Onset
Above Background
LM-94-15
LM-95-58
LM-95-61
LM-95-86
LM-95-103
LM-95-108
0.2235
0.0357
0.1064
0.0312
0.0266
0.051 1
0.0
31.5
7.8
14.9
23.1
12.5
1965, 1972
1969
1968
-1953
-1961
1943
1916
1917
Prior to 191 6
1932
Historic cores include those reported by Swackhamer
and Armstrong (1988), Hermanson et al. (1991),
Golden et al. (1993), and Schneider et al. (2001)
(Figures 1.5.19 to 1.5.22). The core reported by
Swackhamer and Armstrong (1988) was collected in
1980 (Figure 1.5.19). It failed to capture any decline
in PCBs attributable to the ceasing of the
manufacture of PCBs in 1977. For cores collected in
1984, a decline in PCBs was documented in four of
the five cores (Figure 1.5.20). For these cores, the
apparent peak of PCBs occurred after 1965. Cores
collected in 1991 and 1992 had a peak PCB
concentration between the early 1960s and early
1980s (Figure 1.5.21), and the core collected in 1998
had a peak concentration in the mid-1970s (Figure
1.5.22). Peak concentrations in all these cores are
consistent with the LMMBP cores which had a peak
concentration between the early 1960s and early
1970s (Figure 1.5.18).
7.5.6 Biota
Various biota were analyzed for PCBs. These
include phytoplankton, zooplankton, Diporeia, Mysis,
alewife, bloater, deepwater sculpin, smelt, slimy
sculpin, coho, and lake trout (Table 1.5.11). Mean
concentrations ranged from 49 ng/g in phytoplankton
to 3,000 ng/g in lake trout. For alewife and bloater,
the fish were divided into two size classes based on
length. For both of these, PCB concentrations were
higher in the larger fish. Concentrations of PCBs in
members of the lake trout food web increase with
trophic level, with concentrations lowest in the
plankton, higher in benthos, and highest in the forage
fish, illustrating biomagnification of PCBs (Figure
1.5.23). For lake trout, PCB concentrations
increased in a predicable way with fish age until age
10 (Figure 1.5.24). All PCB concentrations in lake
trout exceed those of their prey. For unknown
reasons, age 11 fish and older have PCB
concentrations that vary in no predictable way with
increasing age.
PCB concentrations in lake trout have been declining
since 1975 (Figure 1.5.25). Similarly, concentrations
of PCBs have declined in bloater since 1974 (Figure
1.5.26). These appear to be responding to the
phase-out of PCB use in the basin which occurred at
most locations by the early 1970s (Table 1.5.1).
1.5.7 Summary
Concentrations of PCBs have declined in Lake
Michigan since the phase-out of use by industries
within the basin. Concentrations are highest in
southern Lake Michigan for most media. For air and
water media, concentrations are highest near
shoreline sources.
References
Anderson, D.J., T. B. Bloem, R.K. Blankenbaker, and
T.A. Stanko. 1999. Concentration of
Polychlorinated Biphenyls in the Water Column of
the Laurentian Great Lakes: Spring 1993. J.
Great Lakes Res. 25(1 ):160-170.
86
-------
.O)
^5>
c
100
80 ::
60 :;
40 ::
20 -'-
0
H 1 1-
H 1 1 1 1 1 1-
2000 1980 1960 1940 1920 1900
Year
-•- LM-80-18
Figure 1.5.19. Vertical variation of PCBs in dated sediment cores reported by Swackhamer and
Armstrong (1988).
£3\J -I
onn
ZUU
-i t;n
1 ou
i nn
IUU
KH
ou
n
V
W DT
0
•
D
; * v v
t * ° ° A P-
I ! I 1 i r 1 _ _l . _l J.I I _ 1__J !_„! 1 .. ]. 1 1 1 1
2000 1980 1960 1940 1920 1900 1880
Year
-A- LM-84-SLM-F -m- LM-84-SLM-D -^- LM-84-CLM-M
-V- LM-84-NLM-B -Q- LM-84-NLM-E
Figure 1.5.20. Vertical variation of PCBs in dated sediment cores reported by Hermanson etal. (1991).
87
-------
,o>
^)
200
160
120 '- -•-£•-*
80 --
40
D
P | P ir
0
2000 1980 1960 1940 1920 1900
Year
-m- LM-91-68R-A- LM-91-18 -B- LM-92-47s
Figure 1.5.21. Vertical variation of PCBs in dated sediment cores reported by Golden et al. (1993).
75
60 --
«45
O)
c
30
15 -
0
2000 1980 1960 1940 1920 1900
Year
LM-98-HMS-1
Figure 1.5.22. Vertical variation of PCBs in dated sediment cores reported by Schneider et al. (2001).
88
-------
Table 1.5.11. Mean Concentrations of PCBs Measured in the 1994-1995 Lake Michigan Food Web
(McCarty et al. 2004)
Sample Type
N
Mean (ng/g)
Standard Deviation (ng/g)
Phytoplankton
Zooplankton
Mysis
Diporeia
Smelt
Slimy Sculpin
Deepwater Sculpin
Alewife < 1 20 mm
Alewife > 1 20 mm
Bloater < 1 60 mm
Bloater > 1 60 mm
Coho - Hatchery
Coho - Yearling
Coho - Adult
Lake Trout (All Age Classes)
71
70
53
39
73
69
74
60
70
70
67
5
8
54
246
49
170
250
420
310
430
420
250
580
650
830
120
200
810
3000
38
74
61
100
83
130
200
150
140
180
210
27
90
520
2300
1000
800
3 600
m
o
a.
$
2. 400
200
C
'a.
0
CO
Iff
1 1 3
s ^ ^
5 D 1
o
a
a
E
en
I -S i
0
c
i
C
0
c :
: s
3 •*
Q.
O
S
"5.
c
^
J
2
£
Q
L
•\
>
ff^
E
CO
L.
9)
13
o
m
Figure 1.5.23. PCB concentrations in various members of the lake trout food web during the LMMBP.
89
-------
10000 -
8000-
I? 6000 -
c
CO
O
Q.
•§ 4000 -
2000-
r\
CD K
0
0
O)
•^
O)
0
O)
oo
D)
<
f|,
CD
0
CD
CM
(P
m
^
CO
0
O)
T
1
0
n>
1
1
O)
< <
T
IO
5. *
0
_ TO
2 < i
0
o>
l>
3)
f
Figure 1.5.24. PCB concentrations in various age classes of lake trout during the LMMBP.
E 20
Q.
a
m
o
Q. 10 +
i
1970 1975 1980 1985 1990 1995 2000 2005
Year
Figure 1.5.25. Time variation of PCB concentration in five to six year-old lake trout from Lake
Michigan.
90
-------
6.0
4.8 -
CO
o
Q_
2.4 -•
1.2 -
0.0
1970 1980 1990 2000
Year
Figure 1.5.26. Time variation of PCB concentrations in bloater from Lake Michigan.
Bicksler, J. 1996. PCBs in the Spring-Time Water
Column of the Great Lakes. M.S. Thesis,
University of Minnesota, Minneapolis, Minnesota.
160pp.
Boughton, L. 2004. Human Impact. Presented at
the State of the Lakes Ecosystem Conference.
Toronto, Ontario, Canada, October 6-8, 2004.
Buehler, S.S. and R.A. Hites. 2002. The Great
Lakes' Integrated Atmospheric Deposition
Network. Environ. Sci. Technol., 36(17):354A-
359A.
Buehler, S.S., I. Basu, and R. A. Hites. 2004.
Causes of Variability in Pesticide and PCB
Concentrations in Air Near the Great Lakes.
Environ. Sci. Technol., 38(2):414-422.
Chambers, R.L. and B.J. Eadie. 1980. Nearshore
Chemistry in the Vicinity of the Grand River,
Michigan. National Oceanic and Atmospheric
Administration, Great Lakes Environmental
Research Laboratory, Ann Arbor, Michigan.
NOAA Technical Memorandum ERL GLERL-28,
28pp.
DeVault, D.S., R. Hesselberg, P.W. Rodgers, and
T.J. Feist. 1996. Contaminant Trends in Lake
Trout and Walleye From the Laurentian Great
Lakes. J. Great Lakes Res., 22(4):884-895.
Filkins, J.C., J.M. Townsend, and S.G. Rood. 1983.
Organochlorines in Offshore Waters of the Great
Lakes, 1981. U.S. Environmental Protection
Agency, Office of Research and Development,
ERL-Duluth, Large Lakes Research Station,
Grosse lie, Michigan. 12pp.
Frank, R., R.L. Thomas, H.E. Braun, D.L. Gross, and
T.T. Davies. 1981. Organochlorine Insecticides
and PCB in Surficial Sediments of Lake Michigan
(1975). J. Great Lakes Res., 7(1):42-50.
Golden, K.A., C.S. Wong, J.D. Jeremiason, S.J.
Eisenreich, G. Sanders, J. Hallgren, D.L.
Swackhamer, D.R. Engstrom, and D.T. Long.
1993. Accumulation and Preliminary Inventory of
Organochlorines in Great Lakes Sediments.
Water Sci. Technol., 29(8-9):19-31.
Hermanson, M.H., E.R. Christensen, D.J. Buser, and
L. Chen. 1991. Polychlorinated Biphenyls in
Dated Sediment Cores From Green Bay and
Lake Michigan. J. Great Lakes Res.,
17(1):94-108.
91
-------
Integrated Atmospheric Deposition Network. 2004.
Download from the web site
www.msc.ec.gc.ca/iadri/Data/form/form_e.html.
"The Integrated Atmospheric Deposition Network,
established in 1990, is implemented by the
Canadian Federal (Environment Canada) and
Provincial (Ontario Ministry of the Environment)
Governments and the U.S. Environmental
Protection Agency as Mandated in Annex 15 of
the Great Lakes Water Quality Agreement
(GLWQA)."
Lefkovitz. L.F. 1987. The Particle Mediated
Fractionation of PCBs in Lake Michigan. M.S.
Thesis, University of Wisconsin, Madison,
Wisconsin. 238 pp.
Marti, E.A. and D.E. Armstrong. 1990.
Polychlorinated Biphenyls in Lake Michigan
Tributaries. J. Great Lakes Res., 16(3):396-405.
McCarty, H. B., J. Schofield, K. Miller, R. N. Brent, P.
Van Hoof, and B. Eadie. 2004. Results of the
Lake Michigan Mass Balance Study:
Polychlorinated Biphenyls and frans-Nonachlor
Data Report. U.S. Environmental Protection
Agency, Great Lakes National Program Office,
Chicago, Illinois. EPA/905/R-01/011, 289 pp.
Murphy, T.J. and C.P. Rzesutko. 1977. Precipitation
Inputs of PCBs to Lake Michigan. J. Great Lakes
Res., 3(3-4):305-312.
Offenberg, J.H. and J.E. Baker. 2000. PCBs and
PAHs in Southern Lake Michigan in 1994 and
1995: Urban Atmospheric Influences and
Long-Term Declines. J. Great Lakes Res. 26(2):
196-208.
Pearson, R.F., K.C. Hornbuckle, S.J. Eisenreich, and
D.L. Swackhamer. 1996. PCBs in Lake
Michigan Water Revisited. Environ. Sci.
Technol., 30(5):1429-1436.
Rice, C.P., B. J. Eadie, and K.M. Erstfield. 1982.
Enrichment of PCBs in Lake Michigan Surface
Films. J. Great Lake Res., 8(2):265-270.
Robbins, J.A., N.R. Morehead, R.W. Rood, D.N.
Edgington, and S. Meyer. 1999. Accumulation
and Near-Shore Mixing of Sediments in Lake
Michigan as Determined for the Lake Michigan
Mass Balance Program. Part 1. Cores Collected
between 1994 and 1996 (2 Volumes). Final
Report. U.S. Environmental Protection Agency,
Office of Research and Development, ERL-
Duluth, Large Lakes Research Station, Grosse
lie, Michigan. 274 pp.
Rodgers, P.W. and W.R. Swain. 1983. Analysis of
Polychlorinated Biphenyl (PCB) Loading Trends
in Lake Michigan. J. Great Lakes Res. 9(4):
548-558.
Schneider, A.R., H.M. Stapleton, J. Cornwell, and
J.E. Baker. 2001. Recent Declines in PAH,
PCB, and Toxaphene Levels in the Northern
Great Lakes as Determined from High Resolution
Sediment Cores. Environ. Sci. Technol.,
35(19):3809-3815.
Swackhamer, D.L. and D.E. Armstrong. 1987.
Distribution and Characterization of PCBs in Lake
Michigan Water. J. Great Lakes Res.,
13(1):24-36.
Swackhamer, D.L. and D.E. Armstrong. 1988.
Horizontal and Vertical Distribution of PCBs in
Southern Lake Michigan Sediments and the
Effect of Waukegan Harbor as a Point Source. J.
Great Lakes Res., 14(3):277-290.
Wethington, D.M. And K.C. Hornbuckle. 2005.
Milwaukee, Wl, as a Source of PCBs to Lake
Michigan. Environ. Sci. Technol., 39(1):57-63.
92
-------
PART1
INTRODUCTION
Chapter 6. Congener Pattern Matching of
Data Collected for the Lake Michigan
Mass Balance Project (LMMBP)
David A. Griesmer
Computer Sciences Computer
Large Lakes Research Station
9311 Groh Road
Grosse lie, Michigan 48138
1.6.1 Introduction
As part of the quality assurance (QA) process for the
Lake Michigan Mass Balance Project (LMMBP), a
LMMBP PCB Modeling Peer Review Conference was
held on July 27-28, 2004 at the Crowne Plaza Hotel
in Romulus, Michigan to review the LMMBP
polychlorinated biphenyl (PCB) models developed for
this study. One of the recommendations to come out
of this peer review was to "Investigate congener
patterns in air, water, fish, and sediment. How do
these compare?" (Part 7, Appendix 1). In response
to this question, the peer review response document
states that: "The PCB patterns of multiple media will
be compared to determine similarities and
differences within and among media. This technique
is commonly referred to as PCB fingerprinting or PCB
signature recognition and has had mixed success in
the past. This recommendation has minor
implications to the modeling; however, it is a data
analysis tool and has merit for data presentation and
interpretation purposes. The relative percent of total
PCBs represented by each congener will be
computed and then expressed as a cumulative
frequency plot for comparative purposes. These will
represent data for an entire study period, will be
tested with both mean and median values, and will
be a composite expression of seasonal and spatial
data. In addition, selected evaluation of pattern
recognition using the LMMBP data set can be found
in Kuehl (2002) and McCarty et al. (2004).
Fingerprints will be calculated for sediment, water
column (dissolved and particulate), vapor phase, wet
and dry atmospheric deposition, and age 5-6 year-old
lake trout signatures from the Saugatuck biota site.
Atmospheric signatures will be based on a subset of
all congeners because vapor phase data were
computed by Keri Hornbuckle for the study, and over-
lake concentrations were only calculated for the
congeners that are being modeled at Grosse lie. In
addition, PCB patterns associated with water
discharging from the Kalamazoo River near the
Saugatuck biota site and other selected tributaries
will be compared/contrasted to the lake water".
One of the main objectives of this analysis was to
see if there was a correlation between congener
patterns in the LMMBP biota boxes and the
atmospheric, tributary, and sediment sources of
congener contamination. To accomplish this,
congener pattern matching analysis was expanded
from that originally suggested in the peer review
modelers' response. This included congener pattern
matching for all 11 major tributaries, as well as a
comparison of atmospheric inputs for all 10 surface
segments defined in the LM2-Toxic model. In
addition, ages 5-6 year-old lake trout were evaluated
in Sturgeon Bay, Sheboygan Reef, and Saugatuck
biota boxes (Figure 1.6.1), instead of just in the
Saugatuck biota box. Multiple year classes (age 2,
age 3, and age 9), of lake trout were analyzed for the
Saugatuck biota box, to see if there were differences
in the congener patterns of different aged lake trout.
93
-------
segment-. /'' A/-'1 .i, '
9 " '
segment <*•
10 ;;'
__
x< ../ Sturgeon Bay
/ LMMB-biota1
LM2 surface water
segmentation and
LMMB biota boxes
Sheboygan Reef si?
LMM3-biota2 ife^f
} segment 1
HI
biota survey boxes
segment 2 I """•-..,
Saugatuck .
LMMB-blota3
Figure 1.6.1. LM2 surface water segmentation
and LMMBP biota boxes.
1.6.2 Analytical Approach
This exercise was not meant to be a rigorous
statistical analysis of the data, rather it is strictly an
empirical look at congener patterns in different media
collected for the LMMBP. All available congener
data were included in the analysis. Analysis was
done for most of the media collected for PCBs during
the LMMBP (atmospheric, tributary, lake water,
sediment, and biota). Kuehl (2002) previously had
done a comparison of congener patterns of the biota
media (phytoplankton, zooplankton, Mysis, Diporeia,
bloater chub, slimy and deepwater sculpin, alewife,
rainbow smelt, coho salmon, and lake trout) collected
for the LMMBP. Therefore, with the exception of lake
trout, these biota media were not examined in the
present analysis. For all of the analyses done for the
present study, comparisons were made by plotting
the congener patterns and doing a visual comparison
for obvious similarities or differences.
1.6.3 Methodology
PCB congener analysis for the LMMBP were
performed by a number of different principal
investigators, using different instrumentation and
techniques, which are detailed in the LMMB Methods
Compendium (U.S. Environmental Protection
Agency, 1997). These variation in methods may
have had some impact of the comparison of samples
from different media due to co-elution and congener
detection differences from media-to-media.
All analyzed congeners were used to maximize the
amount of data available. All analyses were done
using Microsoft Excel spreadsheets. Congeners
were ordered in each media by congener number,
with co-eluting congeners ordered by the lowest co-
eluting congener number (see Table 1.6.1) Data
were analyzed by media, where the mean and
median values were calculated for each congener.
Cumulative frequency analysis was then performed
on the means and medians for each congener. The
percent frequency of each congener was calculated
by dividing the means and median of each congener
by the total sum of all the congeners in that media to
give a percent value to each congener. These
percentages were added together to give a
cumulative frequency distribution which totals to one.
These data were graphed, and visual comparisons
were done.
1.6.4 Results
1.6.4.1 Comparison of Modeled Congener
Patterns to All Analyzed Congener Patterns
In an effort to make use of as many congeners as
possible in this analysis, all congeners analyzed in
each media were used rather than just the modeled
94
-------
Table 1.6.1 Comparison of Congeners Available for Analysis in All LMMBP Media
Available Congeners in All Media
Modeled
Congeners
8+5
15+17
16+32
018
026
28+31
033
37+42
044
049
052
56+60
066
70+76
074
77+110
081
92+84
085
087
089
099
101
132+153+105
118
123+149
163+138
146
151
170+190
172+197
180
187+182
195+208
196+203
201
Vapor
Phase
8+5
15+17
16+32
018
026
28+31
033
37+42
044
049
052
56+60
066
70+76
074
77+110
081
92+84
085
087
089
099
101
132+153+105
118
123+149
163+138
146
151
170+190
172+197
180
187+182
195+208
196+203
201
Wet
Deposition
8+5
012
013
15+17
016
018
026
31+28
032
033
037
042
044
049
052
56+60
066
70+76
074
077
081
92+84
085
087
089
099
101
132+153+105
110
118
123+149
163+138
170+190
172
180
187+182
208+195
196
197
201
203
Dry
Deposition
8+5
012
013
15+17
016
108
1026
31+28
032
033
037
042
044
049
052
56+60
066
70+76
074
077
081
92+84
085
087
089
099
101
132+153+105
110
118
123+149
163+138
146
151
170+190
172
180
187+182
108+195
196
197
201
203
Open Lake
Water
4+10
8+5
006
7+9
012
013
014
15+17
016
018
019
021
022
024
025
026
027
31+28
029
032
033
37+42
040
41+71
043
044
045
046
47+48
049
051
052
053
56+60
063
064
066
70+76
074
77+110
081
082
083
92+84
085
087
089
091
095
097
099
100
101
Surficial
Sediment
8+5
006
7+9
012
013
15+17
016
018
019
021
022
24+27
025
026
31+28
029
032
033
37+42
040
41+71
042
043
044
045
046
47+48
049
051
052
053
56+60
063
064
066
70+76
074
77
081
082
083
92+84
085
087
089
091
095
097
099
100
101
105+132+153
107
Tributary
003
4+10
8+5
006
7+9
15+17
16+32
017
018
019
022
24+27
025
026
31+28
033
37+42
040
41+71+64
044
045
046
47+48
049
051
052
053
56+60
063
066
66+95
70+76
074
77+110
082
083
92+84
085
087
089
091
095
097
099
101
105+132+153
118
123+149
128
132+153
135+144
136
137+176
Lake
Trout
022
24+27
31+28
029
033
040
41+71
042
044
47+48
049
052
56+60
063
064
066
70+76
074
77
81+87
082
083
92+84+89
085
091
095
097
099
101
105
107
110
114
118
119
123
126
128
129
131
132+153
134
135+144
137+176
163+138
141
146
149
151
156
157
158
167
95
-------
Table 1.6.1. Comparison of Congeners Available for Analysis in All LMMBP Media (Continued)
Available Congeners in All Media
Modeled Vapor Wet Dry Open Lake
Congeners Phase Deposition Deposition Water
103
105+132+153
107
114+131
118
123+149
147+124
128
129
130
134
135+144
1.36
137+176
163+138
141
146
151
156
157+200
158
167
170+190
202+171
172+197
173
174
175
177
178
180
187+182
183
185
189
191
193
194
208+195
196
198
199
201
203
205
206
207
Total # 209
Congeners:
Surficial
Sediment
110
114+131
118
119
123+149
128
129
130
134
135+144
136
137+176
163+138
141
146
151
156
157+200
158
167
170+190
202+171
172
173
174
175
177
178
180
187+182
183
185
189
191
193
194
208+195
203+196
197
198
199
201
205
206
207
209
Tributary
163+138
141
146
149
151
158
167
170+190
202+171
172
172+197
174
177
178
180
187+182
183
185
193
194
208+195
203+196
198
199
201
206
207
Lake
Trout
170+190
171
172
173
174
175
177
178
180
187+182
183
185
189
191
193
194
195
203+196
197
198
199
200
201
202
205
206
207
208
209
54
54
54
56
127
123
107
98
96
-------
congeners. Because most of the comparisons were
in the same media, this would give the maximum
number of data points. However, there were some
concerns when comparing data from different media.
To see if this approach was feasible, an initial
comparison was done to see if the congener pattern
trends seen using only modeled congeners were
different from the congener patterns seen when "all
available congeners" were analyzed (Table 1.6.1).
This analysis was done for a number of different
media (vapor phase, wet and dry deposition,
dissolved and particulate PCBs in lake water, surficial
sediments, and 5-6 year-old lake trout).
The definition of "all congeners" varied somewhat,
depending on the medium being examined. For
dissolved and particulate water, sediment, tributary,
and lake trout data, all congeners reported were
used in this analysis. This number varied from a low
of 98 congeners in the lake trout analysis to a high of
127 congeners in the open lake water samples. For
the atmospheric samples, Keri Hornbuckle calculated
over-water concentrations of only modeled
congeners in vapor phase, as well as wet and dry
deposition data (Table 1.6.1); therefore, the only
congeners available in this media are the 54 modeled
congeners. While over 100 congeners were
available from Hornbuckle in the atmospheric wet
and dry deposition data sets, only a subset were
used because data needed to be aggregated by
LM2-Toxic modeling segment. These aggregated
data sets contained only 54 congeners of wet
deposition data and 56 congeners of dry deposition
data (Table 1.6.1).
This comparison was limited to the samples collected
in or near the Saugatuck Biota Box, LMMBP-BiotaS,
for lake trout and surficial sediment, or to LMMBP
modeling segment 2 (Figure 1.6.1), which contains
the Saugatuck biota box, for the atmospheric and
lake water samples. In addition, dissolved and
particulate water data for the Kalamazoo River were
included, because this is the tributary which
discharges water closest to the Saugatuck biota box.
Cumulative frequency plots comparing all available
congener data and modeled congener data were
then created. This comparison showed that the
trends in the different media were similar, whether
modeled congeners or all available congeners were
used (Figures 1.6.2-1.6.10). This is not too
surprising because the 54 modeled congeners made
up the bulk of the congener mass in all media
sampled, accounting for an average of 74% of the
mass across all media (77.3% of vapor phase PCB
mass, 63.4% of wet deposition PCB mass, 73% of
dry deposition PCB mass, 80.2% of tributary water
PCB mass, 67.5% of dissolved lake water PCB
mass, 77.2% of particulate lake water PCB mass,
84.7% of surficial sediment PCB mass, and 67.1 % of
lake trout PCB mass). Because these trends were
similar, it was decided that all available congeners
would be used in the data analysis, so that we could
take advantage of the maximum amount of data
points.
1.6.4.2 Comparison of Median to Mean Data
A comparison between mean and median values in
different media was done to see how much results
varied between averaging methods. In vapor phase
samples, dry deposition samples, dissolved and
particulate tributary water samples, there was almost
no difference between median and mean plots
(Figures 1.6.11 1.6.14). In wet deposition samples,
dissolved and particulate water samples, surficial
sediment samples, and 5-6 year-old lake trout
samples, there were some small differences in the
plots (Figures 1.6.15 1.6.19), with median values
being somewhat lower in all instances due to the fact
that some of the congeners have zero median
values. These zero median values were the result
of a large number of congeners with reported values
of zero or near zero. We decided to use mean
values for the rest of the analysis because there
appeared to be little difference between median and
mean plots and to avoid the zero median values.
1.6.4.3 Comparison of Congener Patterns in
Different Media in Segment 2/Saugatuck Biota
Box
When the congener patterns for different media in
segment 2 and the Saugatuck biota box were
compared (Figure 1.6.20), the following general
trends were identified. Vapor phase data appeared
to have a higher percentage of lower chlorinated
congeners than any of the other media. Dissolved
PCB congeners in water from the Kalamazoo River
most nearly matched the congener pattern seen in
the vapor phase, with the congener pattern for
dichloro, tetrachloro - decachloro's most closely
97
-------
1.0
0.9
0.8-
0.7
0.6-
0.5
0.4-
0.3-
0.2
0.1 •
PCB congeners in
segment 2 vapor phase
B» all congeners
— modeled congeners
1 17 "33" 49" 65 81 97" 113 129 145 161 177 193 209
congener
Figure 1.6.2. Cumulative frequency distribution
- PCB congeners in segment 2 vapor phase.
dissolved PCB congeners
in segment 2 water
— all congeners
— modeled congeners
1 17 33 49 65 81 97 113129145161177193209
congener
Figure 1.6.5. Cumulative frequency distribution
- dissolved PCB congeners in segment 2 water.
PCB congeners in
segment 2 dry deposition
'an, all congeners
— modeled congeners
i.o
33 49 65 81 97 113129145161177193209
congener
Figure 1.6.3. Cumulative frequency distribution
- PCB congeners in segment 2 dry deposition.
g 0.9-
o>
g 0.8-
0.6-
o
C 0.
j! 0.3-1
o
c 0.2
0-M
participate PCB congeners
in segment 2 water
~=- all congeners
— modeled congeners
1 17 33 49"65 81 gy'flS 129 145 161 177 193209
congener
Figure 1.6.6. Cumulative frequency distribution
- particulate PCB congeners in segment 2 water.
PCB congeners in
segment 2 wet deposition
all congeners
— modeled congeners
1 17 33 49 65 8l""g'f'm 'l29 i"45"l61 "H~7 193"'209
congener
Figure 1.6.4. Cumulative frequency distribution
- PCB congeners in segment 2 wet deposition.
PCB congeners in
segment 2 surficial sediment
«- all congeners
— modeled congeners
33 49 65 81 97 113 129 145 161 177 193 209
congener
Figure 1.6.7. Cumulative frequency distribution
- PCB congeners in segment 2 surficial
sediment.
98
-------
1.0
0.9-
0.8-
0.7-
0.6-
0.5
0.4
0.3-
0.2
dissolved PCB congeners
in Kalamazoo River water
-»- all congeners
— modeled congeners
~ r
1 17 33 49 65 81 97 113129145161177193209
congener
Figure 1.6.8. Cumulative frequency distribution
- dissolved PCB congeners in Kalamazoo River
water.
PCB congeners in
segment 2 vapor phase
-^i median
r
17 33 49 65 81 97 113129145161 177193209
congener
Figure 1.6.11. Cumulative frequency distribution
- PCB congeners in segment 2 vapor phase.
1.0
0.9
0.7 ^
« 0.5-
g 0.4-
3 0.3-
o
£ 0.2-
particulate PCB congeners
in Kalamazoo River water
-=- all congeners
— modeled congeners
1 17 33 49 65 81 97 113129145161177193209
congener
Figure 1.6.9. Cumulative frequency distribution
- particulate PCB congeners in Kalamazoo River
water.
PCB congeners in
segment 2 dry deposition
*"« median
17 33 49 65 81 97 113 129 145 161 177 193 209
congener
Figure 1.6.12. Cumulative frequency distribution
- PCB congeners in segment 2 dry deposition.
1.0
0,9
0.8
0.7
0.6-
0.5
0.4'
0.3-
0.2-
0.1 •
0-
Age 5-6 Saugatuck Lake Trout
| — all PCB congeners :
| — modeled PCB congeners :
! r
1 17 33 49 65 81 97 113129145161 177193209
congener
Figure 1.6.10. Cumulative frequency distribution
- age 5-6 Saugatuck lake trout.
dissolved PCB congeners in
Kalamazoo River water
=^ median
17 33 49 65 81 97 113129145161177193209
congener
Figure 1.6.13. Cumulative frequency distribution
- dissolved PCB congeners in Kalamazoo River
water.
99
-------
particulate PCB congeners
in Kalamazoo River water
ecu median
— mean
33 49 65 81 97 113129145161177193209
congener
Figure 1.6.14. Cumulative frequency distribution
- particulate PCB congeners in Kalamazoo River
water.
1.0
0.9-
0.8
0.7-
0.6-
0.5-
0.4-
0.3-
0.2
0.1 •
0-
PCB congeners :
in segment 2 wet deposition
-' - median !
— mean
1 17 33 49 65 81 97 113129145161177193209
congener
Figure 1.6.15. Cumulative frequency distribution
- PCB congeners in segment 2 wet deposition.
dissolved PCB congeners
in Saugatuck water
a-n median
— mean
1 17
33 49 65 81 97 113129145161177193209
congener
Figure 1.6.16. Cumulative frequency distribution
- dissolved PCB congeners in Saugatuck water.
particulate PCB congeners '
: in Saugatuck water
i B.«, median
— mean
17 33 49 65 81 97 113129145161177193209
congener
Figure 1.6.17. Cumulative frequency distribution
- particulate PCB congeners in Saugatuck water.
PCB congeners in
surficial sediment
1 17 33 49 65 81 97 113129145161177193209
congener
Figure 1.6.18. Cumulative frequency distribution
- PCB congeners in surficial sediment.
PCB congeners in age 5-6
Saugatuck lake trout
!«= median
1 17 33 49 65 81 97 113129145161177193209
congener
Figure 1.6.19. Cumulative frequency distribution
- PCB congeners in age 5-6 Saugatuck lake trout.
100
-------
All PCB congeners
segment 2 / Saugatuck
dissolved water
•^~ Kalamazoo River dissolved water
• »" Kalamazoo River participate water
participate water
'"""'"" wet deposition
dry deposition
vapor phase
surficial sediment
age 5-6 year lake trout
17 33 49 65 81 97 113 129 145 161 177 193 209
congener
Figure 1.6.20. PCB congeners in segment 2, Saugatuck.
matching the vapor phase data. The pattern seen for
open lake dissolved water samples did not closely
match any of the other media. Open lake dissolved
PCB congeners in water had a lower percentage of
lower chlorinated congeners than either the vapor
phase or dissolved water congener samples from the
Kalamazoo River, but it had a higher percentage of
these congeners than any of the other media.
The Kalamazoo River particulate PCB congeners in
water had a higher percentage of lower congeners
than open lake particulate PCB congeners in water
and closely matched the congener patterns seen in
both wet and dry deposition data for segment 2.
Open lake particulate PCB congeners data most
closely matched the surficial sediment for segment 2.
Age 5-6 year-old lake trout had a much lower
percentage of lower chlorinated congeners than any
of the other media.
1.6.4.4
Data
Comparison of Atmospheric Congener
Atmospheric PCB samples for vapor phase and wet
and dry deposition media were collected from a
number of land-based stations around Lake
Michigan, as well as from shipboard stations (Figure
1.6.21) on several LMMBP surveys (U.S.
101
-------
H+
+
air sampling locations
•H K,
HH
Figure 1.6.21. Air sampling locations.
r LEVELS
7 High resolution 5km X 5km grid
19 sigma layers
1. 2318 surface segments
~> 44,042 water segments
Figure 1.6.22. Lake Michigan high-resolution 5
km x 5 km grid with 19 sigma layers.
Environmental Protection Agency, 1997). These
data were then used by Hornbuckle to generate over-
water concentrations of PCB congeners for all 5 km
surface grid cells used for the high-resolution model
(Green, 2000; Miller etal., 2001) (see Figure 1.6.22).
Hornbuckle only analyzed modeled congeners, thus
the congener set for this media is somewhat smaller
than in other media. These data were then
aggregated for modeling into surface segment cells
for the LM2-Toxic model (Figure 1.6.1). As stated
previously, the Saugatuck biota box (LMMBP-BiotaS)
is located within LM2 segment 2. A congener pattern
comparison was done to see if the congener patterns
102
-------
seen in segment 2 were similar to patterns seen in
other segments. This was indeed the case with
vapor phase and wet deposition data (Figures 1.6.23
and 1.6.24). For dry deposition samples, LM2
segment 1, which is the southwest corner of the lake,
is somewhat lower in tetrachloro and pentachloro
congeners than all of the other segments, which have
a very similar congener pattern (Figure 1.6.25).
PCB congeners
in atmospheric
vapor phase
Segment 1
Segment 2
- - • Segment 3
Segment 4
Segment 5
Segment 6
Segment 7
- - Segment 8
Segment 9
Segment 10
17 33 49 65 81 97 113129145161 177193209
congener
Figure 1.6.23. Cumulative frequency distribution
(mean) - PCB congeners in atmospheric vapor
phase.
1.0 j
0.9-
O.B-
0.6
0.5
0.4
0.3
0.1-
o-lu
Segment 1
Segment 2
- - Segment 3
Segment 4
— Segment 5
Segments
Segment 7
Segment 8
- - - Segment 9
— Segment 10
S~
PCB congeners
in atmospheric
wet deposition
1 17 33 49 65 81 97 113129145161177193209
congener
Figure 1.6.24. Cumulative frequency distribution
(mean) - PCB congeners in atmospheric wet
deposition.
A comparison of vapor phase PCB samples from
segment 2 to wet and dry deposition samples from
this segment (Figure 1.6.26) showed a somewhat
similar congener distribution for wet and dry
•e1'0
g 0.9-
o>
i> 0.8-
I 0.7]
£ 0.6
0.5-
0.4-
0.3
E 0.2-
PCB congeners in
atmospheric dry deposition
•=» segment 1
— all other segments
1 17 33 49 65 81 97 113129145161177193209
congener
Figure 1.6.25. Cumulative frequency distribution
(mean) - PCB congeners in atmospheric dry
deposition.
1.0
0.9
0.8-
0.7-
0.6-
0.5-
0.4-
0.3-
0.2-
0.1 •
PCB congeners in
segment 2 atmospheric data
«"- vapor phase
— wet deposition
dry deposition
1 17 33 49 65 81 97 113129145161177193209
congener
Figure 1.6.26. Cumulative frequency distribution
(mean) - PCB congeners in segment 2
atmospheric data.
deposition samples. The vapor phase congener
pattern was quite different with a much higher
percentage of lower chlorinated congeners.
1.6.4.5 Comparison of Tributary Congener
Patterns
An examination of 11 monitored LMMBP tributaries
show that dissolved water samples clearly had a
higher percentage of lower chlorinated congeners
than do particulate water samples (Figure 1.6.27).
The Fox River dissolved PCBs in water had a higher
percentage of lower congeners than any of the other
dissolved tributary samples. This same trend held
true for Fox River particulate water samples.
103
-------
Manistique
River
Menominee
River,
Sheboygan
River
Milwaukee
River
Kalamazoo River
St. Joseph River
Fox River
Kalamazoo River
Grand Calumet River
Grand River
Sheboygan River
Milwaukee River
St Joseph River
Muskegon River
Menominee River
Manistique River
Pere Marquette
dissolved
^—— participate
- dissolved
—— particulale
dissolved
participate
dissolved
particulate
dissolved
particulate
— — dissolved
particulate
— — dissolved
particulate
— dissolved
______ particulate
dissolved
particulate
— — dissolved
particulate
dissolved
particulate
PCB congeners in
Lake Michigan tributaries
81 97 113 129 145 161 177 193 209
congener
Figure 1.6.27. PCB congeners in Lake Michigan tributaries.
104
-------
When segment 2 data for other media were overlaid
on the tributary data, it appears that vapor phase
data most closely matched dissolved water tributaries
data from the western side of the lake (Figure
1.6.28), dissolved water data from segment 2 most
closely matched dissolved water tributary data from
rivers in the lower peninsula of Michigan (Figure
1.6.29), and particulate water data from segment 2
generally fell in the middle of the particulate tributary
data (Figure 1.6.30).
1.6.4.6 Comparison of Ages 5 and 6 Lake Trout
Congener Patterns in All Biota Boxes
Five and six year-old lake trout were collected and
analyzed from all three LMMBP biota boxes
(Sturgeon Bay: LMMBP-Biotal, Sheboygan Reef:
LMMBP-Biota2, and Saugatuck: LMMBP-BiotaS).
Congener patterns were similar for all three biota
boxes, with Saugatuck having a somewhat higher
percentage of lower congeners, Sheboygan reef
having the lowest percentage of lower congeners,
and Sturgeon Bay being in between (Figure 1.6.31).
1.6.4.7 Comparison of Different Lake Trout Age
Class Congener Patterns in Saugatuck Biota Box
A comparison was done between different age
classes of lake trout in the Saugatuck biota box to
see if the congener patterns were different for
different age classes. Lake trout data for year
classes 2, 3, and 9 were compared to age classes 5
and 6 lake trout (Figure 1.6.32). This comparison
showed that there was very little difference between
the congener patterns of these different age classes.
1.6.5 Conclusions
When all media were looked at, there were very clear
differences in the congener patterns that were
observed.
dissolved PCB congeners in
Lake Michigan water
2 E E 1i segment 2
vapor phase
iiiiiiiiiiiiiiiiiirTlininTiTiiiiiTirmnTiiniMMiiiiiiiiHMi in inn iiiiniiiiiiiini in in MIMMIIHIIIIIIIIIHIII MI HIM m in niiiiiiiiiiiiii n n MIIIIII mi inn HIM mi mini HI
1 17 33 49 65 81 97 113 129 145 161 177 193 209
congener
Figure 1.6.28. Comparison of dissolved PCB congeners in Lake Michigan western tributaries to
segment 2 vapor phase.
105
-------
dissolved PCB congeners in
Lake Michigan water
i segment 2
dissolved water
- - - Pere Marquette
i Muskegon River
aa ass Grand River
Kaiarnazoo River
- St Joseph River
'ere Marquette
:iver
luskegon River
J~~Grand River
Kalamazoo River
St. Joseph River
rillUIIIIIMIlllMllllHinniMIHniMIIIIIIMIllMIIIHnllnTnTnTIIIMIIIMIIIIIIIIIHIHIIirnillirnilHIIIIIIIHIMIIIMIIIIIIIiniHIIIIIIMIIMnuUMIIMIIIIUUMIIIllllllllltlllMiniMI
1 M 33 49 65 81 97 113 129 145 161 177 193 209
congener
Figure 1.6.29. Comparison of dissolved PCB congeners in Lake Michigan eastern tributaries to
segment 2 water.
segment 2
particulate water
Manistique River
Menominee River
Fox River
Sheboygan River
Milwaukee River
Grand Calumet
- — Pere Marquette
...... Muskegon River
Grand River
Kalamazoo River
St Joseph River
particulate PCB congeners in
Lake Michigan tributaries
97 113 129 145 161 177 193 209
congener
Figure 1.6.30. Comparison of particulate PCB congeners in Lake Michigan segment 2 to tributaries.
106
-------
PCB congeners in
age 5 & 6 lake trout
Saugatuck
— Sheboygan Reef
Sturgeon Bay
17 33 49 65 81 97 113129145161177193209
congener
Figure 1.6.31. Cumulative frequency distribution
(mean) - PCB congeners in ages 5 and 6 lake
trout.
1.0
0.9-
0.8
0.7-
0.6-
0.5
0.4
0.3
0.2
0.1
04
PCB congeners in
Saugatuck lake trout
—« age 2 years
— age 3 years
age 5-6 years
— age 9 years
1 17 33 49 65 81 97 113129145161 177193209
congener
Figure 1.6.32. Cumulative frequency distribution
(mean) - PCB congeners in Saugatuck lake trout.
Vapor phase data had the highest percentage of
lower chlorinated congeners of all the media
examined. In all cases, the dissolved fraction of
water samples had a higher percentage of lower
chlorinated congeners than the corresponding
paniculate fraction of water samples. The
Kalamazoo River dissolved and particulate water
samples had a higher percentage of lower
chlorinated congeners than the corresponding open
lake dissolved and particulate water samples. The
Kalamazoo River dissolved water samples most
closely resembled the vapor phase data, while the
Kalamazoo River particulate water data most closely
resembled segment 2 wet and dry deposition data.
This could be due to the possibility that PCBs in the
river are from a newer (non-weathered) source of
contamination than open lake samples.
Open lake particulate samples most closely matched
the surficial sediment data for segment 2. This result
probably is not too surprising considering the close
relationship between sediments and particulates and
the continuous deposition and resuspension of
bottom sediments in the water column.
Atmospheric congener patterns were very uniform
over the entire lake, with the exception of dry
deposition data which had a somewhat different
congener patten in segment 1 than in all of the other
model segments. Wet and dry deposition samples
showed similar congener patterns for segment 2 with
dry deposition samples having a slightly lower
congener distribution pattern than wet deposition
samples. This is the same trend that was seen in
lake water and tributary samples. Vapor phase
samples for segment 2 had a very different congener
pattern with a much higher percentage of lower
chlorinated congeners.
Dissolved tributary water samples had a higher
percentage of lower chlorinated congeners than
particulate water samples. It is interesting that
dissolved water tributary data, with the exception of
the Kalamazoo River, from the western side of Lake
Michigan (Fox, Grand Calumet, Sheboygan,
Milwaukee, Manistique, and Menominee Rivers)
generally had a higher concentration of the lower
chlorinated congeners than tributaries from the lower
peninsula of Michigan (Grand, Muskegon, Pere
Marquette, and St. Joseph Rivers) (Figure 1.6.33).
This same trend generally holds true for the tributary
particulate data (Figure 1.6.34). Overall, the Fox
River had the highest percentage of lower chlorinated
congeners for both dissolved and particulate
samples.
Fish from all three biota boxes had very similar
congener patterns. In addition, fish from different
age classes in the same biota box (Saugatuck) also
had very similar congener patterns. This would seem
to indicate that the same processes are responsible
for determining the congener pattern distribution in
fish. This pattern was also similar to the trend that
Kuehl observed in her analysis of LMMBP biota
samples (Kuehl, 2002), where she reported that
107
-------
CD
c
CD
0)
O
O
DO
O
Q.
"CD
**—
o
E
"5
c
o
"o
CD
1.0-
0.9-
0.8-
0.7-
0.6-
0.5-
0.4-
0.3-
0.2-
0.1 -
o-
dissolved PCB congeners in
Lake Michigan tributaries
west side of Lake Michigan
Manistique River
Menominee River
Fox River
Sheboygan River
Milwaukee River
Grand Calumet
Menominee
River
ere Marquette
River
Sheboygan
River
•''Muskegon River
Grand River
Kalamazoo River
east side of Lake Michigan
— - Pere Marquette
Muskegon River
Grand River
Kalamazoo River
- - - St Joseph River
St. Joseph River
Ti II11 Ml I IN 111 NIIHIIHII11III HI I MM I IHt III III11II11III III 11 til 111 III11III III ml I III! HI II ill III11II11III 111 I III 11II11 III HI
17 33
49
65 81
97 113
congener
129 145 161 177 193 209
Figure 1.6.33. Comparison of dissolved PCB congeners in west side-to-east side of Lake Michigan
tributaries.
1.0'
CD
CD
O)
O
O
DO
O
CL
o
E
0.9-
0.8-
0.7-
0.6-
0.5-
0.4-
c 0.3-
o
fo 0.2-
west side of Lake Michigan
Manistique River
Menominee River
Fox River
Sheboygan River
^^_ Milwaukee River
Grand Calumet
east side of Lake Michigan
- - - Pere Marquette
• ••••' Muskegon River
m EH Grand River
Kalamazoo River
— St Joseph River
particulate PCB congeners in
Lake Michigan tributaries
17 33 49 65
81 97 113 129 145 161 177 193 209
congener
Figure 1.6.34. Comparison of particulate PCB congeners in west side-to-east side of Lake Michigan
tributaries.
108
-------
"With the exception of deepwater sculpin samples
collected in Biota Box #3 near Saugatuck, all LMMBP
biota sample congener patterns were homogeneous",
and "A relatively consistent pattern of the selected
PCB congeners was measured in all of the biota
species or classifications." Also, while the age class
5-6 year-old lake trout congeners pattern for biota
box number 3 most nearly matched the congener
patterns of surficial sediment and particulate water
samples from modeling segment 2, this was not a
very close match.
From this data investigation, it was clear that it was
impossible to relate PCB contamination of lake trout
to a tributary source based on congener pattern
matching. This is due to the fact that tributary
congener patterns for Lake Michigan, regardless of
source, had congener patterns which were
significantly different from all lake trout congener
patterns. In addition, congener patterns seen in lake
trout from different biota boxes had very similar
congener patterns, which would indicate that there
are no spatial differences in lake trout congener
patterns.
References
Green, M.L 2000. Geographic Information System
Based Modeling of Semi-Volatile Organic
Compounds Temporal and Spatial Variability.
Ph.D. Thesis, University of New York, Buffalo,
New York. 250 pp.
Kuehl, M. 2002. Polychlorinated Biphenyl (PCB)
Congener Patterns in Lake Michigan Mass
Balance Study Biota. M.S. Thesis, University of
Wisconsin, Green Bay, Wisconsin. 120 pp.
McCarty, H.D., J. Schofield, K. Miller, R.N. Brent, P.
Van Hoff, and B. Eadie. 2004. Results of the
Lake Michigan Mass Balance Study:
Polychlorinated Biphenyls and frans-Nonachlor
Data Report. U.S. Environmental Protection
Agency, Great Lakes National Program Office,
Chicago, Illinois. EPA/905-R-01/011, 289 pp.
Miller, S.M., M.L. Green, J.V. DePinto, and K.C.
Hornbuckle. 2001. Results from the Lake
Michigan Mass Balance Study: Concentrations
and Fluxes of Atmospheric Polychlorinated
Biphenyls and frans-Nonachlor. Environ. Sci.
Techno!., 35(2):278-285.
U.S. Environmental Protection Agency. 1997. Lake
Michigan Mass Balance Study (LMMBP) Methods
Compendium, Volume 2: Organic and Mercury
Sample Analysis Techniques. U.S.
Environmental Protection Agency, Great Lakes
National Program Office, Chicago, Illinois.
EPA/905/R-97/012b, 532 pp.
109
-------
PART1
INTRODUCTION
Chapter 7. Hindcasting and Forecasting
Functions for PCBs in the Lake Michigan
Ecosystem
Ronald Rossmann
U.S. Environmental Protection Agency
Office of Research and Development
National Health and Environmental Effects
Research Laboratory
Mid-Continent Ecology Division
Large Lakes and Rivers Forecasting Research
Branch
Large Lakes Research Station
9311 Groh Road
Grosse lie, Michigan 48138
1.7.1 Introduction
The challenge for both hindcasting and forecasting is
the development of functions that represent the
change of polychlorinated biphenyl (PCB) loads to
the lake with time. To accomplish this, quality
historical databases are required that bracket both
sides of the peak load to the lake. The use of PCB
production information are of limited usefulness.
They do not represent the historical use of PCBs in
the Lake Michigan basin. The early use of PCBs was
primarily in sealed containers as capacitors and
transformers. PCBs were not released except at
points of manufacture and locations where capacitors
and transformers were manufactured. The number
of these sites was limited. None were within the
Lake Michigan basin. The open use of PCBs within
the basin were primarily for the manufacture of
carbonless paper and as a hydraulic fluid for die-
casting. The first recorded use of PCBs for
manufacture of carbonless paper was in 1954 in the
Fox River Valley at Green Bay, Wisconsin (Table
1.7.1). In the Fox River Valley, its use for this
purpose peaked in 1969-1970. Its use for production
of the paper as well as recycling of the paper was
phased out in 1971 -1972. Other locations associated
with production of this carbonless paper or its
recycling included the Manistique and Kalamazoo
Rivers. The use of hydraulic fluids with PCBs began
in 1948 and ended in 1971 at Waukegan, Illinois.
Another location where these hydraulic fluids were
used was Sheboygan, Wisconsin. It appears that
these two uses are most responsible for the loads of
PCBs to Lake Michigan. These uses are poorly
quantified. Therefore, historical observations of
PCBs in various media were used to construct load
functions for both forecasts and hindcasts.
1.7.2 Forecast Functions
Functions developed for the purpose of forecasting
PCBs in various media were developed from the
1994-1995 Lake Michigan Mass Balance Project
(LMMBP) data and data from various published and
unpublished sources. The data were used to
develop functions that describe the apparent loading
trends for PCB loads from the atmosphere and
tributaries. Data were extremely limited. Historical
data are of unknown quality due to changes in
methodology. Specifically, those data generated
using packed columns rather than capillary columns
for the gas chromatography are suspect. Packed
columns did not provide a good separation of PCB
peaks. Older results were expressed as Aroclors
rather than congeners. These weaknesses can lead
to a high bias. However, these data were all that
110
-------
Table 1.7.1. Significant Dates in the History of PCBs in the Lake Michigan Basin
Date
Event
1865
1881
1914
1927
1935
1948-1971
1954
Mid-1950s to Mid-1960s
1950s to 1980s
1959-1972
1959-1971
1969-1970
1970
1971-1972
1973
1975
1977
1984
1985
1991
1991
1992
1998
1997-1998
2001
2002
First PCB-like chemical discovered
First PCBs synthesized
Measurable amounts of PCBs found in bird feathers
PCBs first manufactured at Anniston, Alabama
PCBs manufactured at Anniston, Alabama and Sauget, Illinois
Outboard Marine Corporation at Waukegan, Illinois purchased eight million gallons of
hydraulic fluid with PCBs
Appleton Paper Company began using PCBs as PCB-coated carbonless copy paper
PCBs loaded to Kalamazoo River from deinking
PCBs discharged to Manistique River and Harbor
Outboard Marine Corporation at Waukegan, Illinois used hydraulic fluid with PCBs for
die-casting
PCBs used by Tecumseh Products Company as a hydraulic fluid was loaded to
Sheboygan River
Paper company discharges of PCBs to Fox River peaked
PCB production peaked at 85 million pounds and huge contamination noted at Sauget,
Illinois plant
Appleton Paper Company and NCR Corporation phased out PCB use. Recycling of
carbonless paper had occurred for several decades
U.S. Food and Drug Administration (USFDA) establish 5 ppm PCB tolerance level in fish
124,000 cans of salmon from Lake Michigan seized because of PCBs
PCB production ends
USFDA lowered PCB tolerance level in fish to 2 ppm
Commercial fishing for carp and other valuable species outlawed on Green Bay
End Sheboygan River PCB remediation of upper river
U.S. Department of Health and Human Services label PCBs as possible carcinogen
End Waukegan Harbor PCB remediation
The eight Great Lakes states agreed on a "Great Lakes Protocol for Fish Consumption
Advisories" that lowered the regional standard from the USFDA commercial
standard of 2 ppm down to 0.05 ppm
Milwaukee River PCB remediation
Manistique Harbor PCB remediation completed
Possibly begin Grand Calumet River PCB remediation
were available at the time of model development and
execution.
1.7.2.1 Tributary Loads
Several data sets were used to develop a first-order
exponential decay function for tributary loads. The
derivation of the PCB loading attenuation rate half-life
for tributaries is straightforward. It utilized 1994 and
1995 LMMBP tributary data (McCarty ef a/., 2004)
along with data for the Fox River for 1989-1990
(Velleux and Endicott, 1994) and for Lake Michigan
tributaries for 1982 (Marti and Armstrong, 1990).
The loading data for the Fox River, Lake Michigan
tributaries, and 1994-1995 were fit using Equation
1.7.1.
= C2e
-kt
(1.7.1)
where:
C, = load at time t1
C7 = load at time t,
k = attenuation rate
t = interval of time between t, and t2
111
-------
The equation was solved for k. For the Fox River
and Lake Michigan tributaries, k was calculated to be
0.053/year and 0.054/year, respectively. These
yielded the half-lives of 12-13 years. The k of
0.054/year was used.
Recent reports of half-lives for individual tributaries
range from 6.1 years (k = 0.114/year) in the
Kalamazoo River (Blasland, Bouck and Lee, Inc.,
2000) to 6.8 years (k = 0.102/year) at the Fox River
DePere Dam and 9.0 years (k = 0.077/year) at the
river mouth (Limno Tech, Inc., 2002). Thus, the
choice of a k equal to 0.054 is probably conservative
for the purposes of a forecast; that is, tributary
loadings could be biased high, leading to later dates
when lake trout will cease to have consumption
advisories.
1.7.2.2 Atmospheric Loads
For the period of 1992 to 1997, Simcik et al. (2000)
reported a half-life of 6.9 ± 3.5 for precipitation. The
half-life for atmospheric vapor phase PCBs was
taken as reported by Schneider et al. (2001) based
upon the work of Hillery et al. (1997). Hillery et al.
(1997) based their work on the 1992-1995
International Atmospheric Deposition Network (I ADN)
data. At that time, the reported half-life was six
years. This is one of the bounding half-lives used for
all forecasts. More recently published half-lives
include those of Simcik et al. (1999) who reported
half-lives of 2.7 years and 3.0 years over water and
land, respectively. Their results were based on
1992-1997 IADN data. For the period of 1992
through 2000 at Sleeping Bear Dunes, Buehler et al.
(2002) reported half-lives of 3.1 ± 0.7 years for 1992
through 1995,4.9 ± 0.9 years for 1992 through 1997,
and 20.0 ± 8.6 years for 1992 through 2000. For the
period of 1992 through 2001, an examination of the
temperature-corrected PCB partial pressure IADN
data revealed that partial pressures were declining at
Sleeping Bear Dunes with a half-life rate of 8.3 ± 1.5
years (Buehler et al., 2004). Because of the
uncertainty concerning the rate of decline and the
apparent increase in half-life with the addition of
more recent data, bounding half-lives of 6 and 20
years were used for the purposes of the forecasts.
The uncertainty in the half-life of atmospheric vapor
phase PCBs suggests that a half-life of 13 years for
tributaries is within reason.
1.7.3 Hindcast Functions
Development of a hindcast load function was more
problematic. The only data that exist for years prior
to the year of peak load are for preserved museum
forage fish specimens (Neidermyer and Mickey,
1976) and lake trout (DeVault et al., 1996). Though
the forage fish data can not be used for a hindcast
because they are preserved specimens, they do
provide valuable information about when PCBs first
appeared in Lake Michigan fish. The fourhorn
sculpin data are of most interest. PCBs were not
detected in the fish collected in 1949 but were found
in fish collected in 1951. For rainbow smelt, PCBs
were not detected in 1942 but were detected and
measured in 1960. This is consistent with the first
known reported purchase of hydraulic fluids with
PCBs in 1948 for use at Waukegan, Illinois. Thus, it
appears contamination of the lake with PCBs did not
begin until after 1948. The lake trout annual data
only go back to 1972 (DeVault et al., 1996). The
peak in the lake trout occurred in 1974-1975; hence,
not enough data for establishing a function for the
onset of contamination.
The only way currently available to reconstruct the
load function of PCBs for the lake was to utilize
information available from dated sediment cores. The
number of sediment cores for which data are
available are limited (Figure 1.7.1). Cores SLMD,
SLMF, CLMM, NLMB, and NLME are from the work
of Hermanson et al. (1991). The 18S core is from
Swackhamer and Armstrong (1988), the HMS1 core
is from Schneider et al. (2001), and cores 18G, 47s,
and 68k are from Golden et al. (1993). The LMMBP
cores are from Stations 15, 61, and 86 (Van Hoof
and Eadie, personal communication). All the cores
are of varying quality. Core quality is dependent
upon sedimentation rate, depth of surficial sediment
mixing by physical and biological processes, location
with respect to sources, and thickness of the core
interval samples. Of 13 cores, core 15 is of the
highest quality. Sedimentation rate (0.2235
g/cm2/year) is one of the highest for Lake Michigan
(core is highly resolved with the surficial 1 cm
representing 1.2 years of deposition), its mixed layer
is less than the 1 cm interval sampled, and it is
located in an area of the lake that is very responsive
to loadings. The location of core 15 is in the region
impacted by an annual spring plume of suspended
112
-------
A
N Escanaba
Scale
__ _ ••=
0 km 40 km 80 km
Menominee
Manistique
Naubinway
Charlevoix
Green Bay
Manitowoc
Milwaukee
Racine
Waukegan
Mackinaw City
Sheboygan.
Chicago
Gary
Muskegon
Grand Haven
Saugatuck
SLMFV /south Haven
Benton Harbor
Michigan City
Figure 1.7.1. Locations of dated cores analyzed for RGBs by Hermanson etal. (1981), Swackhamer and
Armstrong (1988), Schneider etal. (2001), Golden etal. (1993), and Van Hoof (personal communication)
for the LMMBP.
113
-------
participate matter that has concentrations four to 10
times that of the lake (Eadie et al., 1996). Because
core 15 is highly resolved, relatively undisturbed by
post-depositional mixing processes (mixed layer less
than 1 cm), and in a region responsive to an annual
transport event in the lake, it was chosen for the
development of a total PCB loading function for the
lake.
The distribution of RGBs within the core with time can
be broken into two linear functions; one prior to peak
concentration and one after peak concentration
(Figure 1.7.2). The pre-peak function was based on
the period of 1947 to 1965, and the post-peak
function was based on the period 1972 to 1994. The
two functions crossed one another in 1967, indicating
this should be the peak load year. The exact location
of the peak was problematic because there are two
peaks in the observed data. The year of peak
loading was constrained by the date of the peak
concentration of PCBs in lake trout. This peak
occurred between 1974 and 1975 for five to six year-
old fish (DeVault era/., 1996). Thus, peak exposure
could have occurred as early as 1968 and as late as
1970. This was used as guidance when selecting the
year ranges from which the linear functions were
300
chosen. The peak concentration for the intersection
of the two functions needed to be near the range of
1968 to 1970. In addition, the simple linear
regression model for the onset of contamination had
to intersect the x-axis around 1 949 when PCBs were
first noted in forage fish. The function derived for the
onset of PCB loading to the lake is:
CPCB = [(-26316.85) + (13.509334 * yr)]
(R2 = 0.865) (1.7.2)
where:
CpCB =• concentration of PCBs in ng/g
yr = calendar year.
The function derived for the decline of PCB load to
the lake is:
= [(15647.899) + (-7.823438 * yr)]
(R2 = 0.897) (1.7.3)
0
1940 1950 1960 1970 1980 1990 2000
Year
- Measured _B_ 1947-! 965 Linear Fuction
- 1972-1994 Linear Function
Figure 1.7.2. Fit of concentration functions to observed data for core 15.
114
-------
The two function lines crossed in roughly 1967. This
was considered acceptable and assumed to be the
peak load year.
Finally, results from these concentration functions
were converted to annual loads. This was done by
using mass sedimentation rate information
(Rossmann and Edgington, 2000; Rossmann, 2002).
The flux of PCBs to the lake can be described as:
(1.7.4)
where
PCB
- flux of PCBs in ng/cm2/year
MSP - mass sedimentation rate in g/cm2/year.
This calculated flux was then corrected for physical
processes that redistribute sediment within the lake.
Sediment focusing is the process by which sediments
are moved from one location to another. The
process includes sediment resuspension and
transport by currents until deposition at another
location. The materials that are preferentially
resuspended are the fine-grained fraction of
sediments. This fraction of the sediments has 210Pb,
137Cs, PCBs, and other contaminants associated with
it. Thus sediments at the new deposition site
become enriched in these while those at the original
resuspension site may become depleted in these.
The focusing factor applied was that for 210Pb
(Rossmann and Edgington, 2000). The 210Pb
focusing factor is defined as:
(1.7.5)
where:
FF,
210
2WPb ~
'Pb focusing factor
Activity of 210Pb stored in the core for
time period year
Activity of 210Pb supported by radium in
the sediment
Decay corrected activity of 210Pb
deposited for time period year.
The 210Pb flux from the atmosphere is constant, and
the amount stored is calculated from core
measurements. The focusing factor corrected PCB
flux (FFFPCB) in ng/cm2/year is calculated using:
FFF
PCB
*FF,
210Pt>
(1.7.6)
This flux is then converted to a load to the lake using
the depositional and transitional areas of the lake.
The depositional area is defined as the area of the
lake with water depths greater than 100 m, the
transitional area is defined as the areas of the lake
with water depths between 40 and 100 m, and the
non-depositional area is defined as the area of the
lake with water depths less than 40 m (Figure 1.7.3).
Together (excluding Green Bay), the transitional and
depositional areas of the lake represent 68.6% of the
lake's total area of 58,016 km2 (Table 1.7.2).
Table 1.7.2. Sedimentary Zones of Lake Michigan
Region of the
Lake
Non-Depositional
Transitional
Depositional
Green Bay
Percent of the
Lake's Area
24.7
32.8
35.8
6.7
Area, km2
14,334
19,032
20,756
3,894
Total
100.0
58,016
Flux is converted to load using the equation:
LL
-PCB
*(1010cm2//f/772)*>\
(1.7.7)
TD
where:
LPCB = Total load to lake depositional and
transitional areas in kg/year
ATD = Total depositional and transitional area of
the lake in km2.
115
-------
Naubinway
Okm
Manistique
A
N
Escanaba
Scale
•
40km 80km
.^S
Menominee X J f'
Green Bay
Manitowoc
Sheboygan
Milwaukee
Racine
Waukegan
Chicago
Michigan City
Gary
Mackinaw City
Legend
Depositional
Zone
Transitional
Zone
Non-Depositional
Zone
Figure 1.7.3. Lake Michigan non-depositional (0-40 m), transitional (40-100 m), and depositional (> 100
m) zones based on water depth and the depth of wind-wave interaction with sediments.
116
-------
The remaining terms are unit conversion terms.
Loads in the post-peak function were normalized to
the 1989 load (2,731 kg/year) because the best linear
fit line went through 1989. The result of this function
in 1967 of 8,129 kg/year was used to normalize the
onset function so that its result in 1967 was also
8,129 kg/year. The resulting load function had a
PCB load onset in 1949, a peak load of 8,129 kg/year
in 1967, and a 1994 load of 1,504 kg/year (Figure
1.7.4). The 1994 total load was 1.9 times higher than
the measured load of 786 kg/year. This was
considered within reason because at least one
additional loading source was identified since the
project. The source is Milwaukee, Wisconsin with an
estimated load of 130 kg/year (Wethington and
Hornbuckle, 2005). In addition, atmospheric loads
did not include the coarse particulate fraction which
would contribute an additional load. Franz et al.
(1998) estimated a particle dry deposition flux of
1,100 kg/year which was considerably higher than
previously reported dry deposition fluxes of 16 to 170
kg/year. For the LMMBP, the average particulate
PCB flux reported was 120 ng/m2/month (Miller et al.,
2001). This converts to roughly 85 kg/year for the
whole lake. Thus the coarse particulate load could
be as high as 1,000 kg/year. Therefore, using a load
function that results in a 1994 load of 1,504 kg/year
seems quite plausible given the uncertainties in the
dry deposition fluxes. Finally, it is suspected that
other loads, similar to the Milwaukee load,
were undetected due to the lack of sampling stations
in all metropolitan areas.
1.7.4 Estimated PCB Storage
To provide an estimate of the PCBs stored in Lake
Michigan sediment, the three LMMBP core PCB
results were manipulated in the same way as
described above. This included the application of
210Pb focusing factors to each core so that storage in
the entire lake could be estimated. Cores 15, 61,
and 86 yielded a lake-wide storage of 209,239,
64,533, and 41,192 kg, respectively. The mean of
these is 104,988 kg. This is higher than the 75,000
kg reported by Golden et al. (1993). As seen from
the three LMMBP cores, storage results are highly
variable and location dependent. Application of the
above procedure used for estimating storage in the
lake's sediments to cores reported in the literature
(Golden et al., 1993; Hermanson et al., 1991;
Schneider et al., 2001), the mean storage in the main
lake is estimated to be 46,466 kg. This is
comparable to more recent work by Brian J. Eadie
and Patricia Van Hoof (National Oceanic and
Atmospheric Administration (NOAA), Great Lakes
Environmental Research Laboratory (GLERL), Ann
Arbor, Michigan, personal communication) based on
the LMMBP data which has yielded an estimate of
40,700 kg for the main lake and 60,000 kg for the
main lake plus Green Bay.
10000
^
D)
0 ji 1*1*1 f I i i i i I i i i i I i i i i I i i i i I i i i i
1940 1950 1960 1970 1980 1990 2000
Year
Core 15 Load —e— Load Function
Figure 1.7.4. Comparison of load function to 210Pb focusing factor corrected core 15 loads.
117
-------
References
Blasland, Bouck, and Lee, Incorporated. 2000.
Allied Paper, Incorporated/Portage
Creek/Kalamazoo River Superf und Site Remedial
Investigation/Feasibility Study. Remedial
Investigation Report Phase I. A 10/30/2000
draft for state and federal review.
Buehler, S.S., I. Basu, and R.A. Hites. 2002. Gas-
Phase Polychlorinated Biphenyl and
HexachlorocyclohexaneConcentrationsNearthe
Great Lakes: A Historical Perspective. Environ.
Sci. Technol., 36(23):5051-5056.
Buehler, S.S. and R.A. Hites. 2002. The Great
Lakes' Integrated Atmospheric Deposition
Network. Environ. Sci. Technol., 36(17):354A-
359A.
Buehler, S.S., I. Basu, and R.A. Hites. 2004.
Causes of Variability in Pesticide and PCB
Concentrations in Air Near the Great Lakes.
Environ. Sci. Technol, 38(2):414-422.
DeVault, D.S., R. Hesselberg, P.W. Rodgers, and
T.J. Feist. 1996. Contaminant Trends in Lake
Trout and Walleye From the Laurentian Great
Lakes. J. Great Lakes Res., 22(4):884-895.
Eadie, B.J., D.J. Schwab, R.A. Assel, N. Hawley,
M.B. Lansing, G.S. Miller, N.R. Morehead, and
J.A. Robbins. 1996. Development of Recurrent
Coastal Plume in Lake Michigan Observed for the
First Time. EOS, Transaction, American
Geophysical Union, 77:337-338.
Franz, T.P., S.J. Eisenreich, and T.M. Holsen. 1998.
Dry Deposition of Particulate Polychlorinated
Biphenylsand Polycyclic Aromatic Hydrocarbons
to Lake Michigan. Environ. Sci. Technol.,
32(23):3681-3688.
Golden, K.A., C.S. Wong, J.D. Jeremiason, S.J.
Eisenreich, G. Sanders, J. Hallgren, D.L.
Swackhamer, D.R. Engstrom, and D.T. Long.
1993. Accumulation and Preliminary Inventory of
Organochlorines in Great Lakes Sediments.
Water Sci. Technol., 29(8-9):19-31.
Hermanson, M.H., E.R. Christensen, D.J. Buser, and
L.Chen. 1991. Polychlorinated Biphenyls in
Dated Sediment Cores From Green Bay and
Lake Michigan. J. Great Lakes Res., 17(1):94-
108.
Hillery, B.L., I. Basu, C.W. Sweet, and R.A. Hites.
1997. Temporal and Spatial Trends in a Long-
Term Study of Gas-Phase PCB Concentrations
Near the Great Lakes. Environ. Sci. Technol.,
Hillery, B.L., M.F. Simcik, I. Basu, R.M. Hoff, W.M.J.
Strachan, D. Burniston, C.H. Chan, K.A. Brice,
C.W. Sweet, and R.A. Hites. 1998. Atmospheric
Deposition of Toxic Pollutants to the Great Lakes
as Measured by the Integrated Atmospheric
Deposition Network. Environ. Sci. Technol.,
32(1 5):221 6-2221.
Limno Tech, Incorporated. 2002. Recent Data
Collection Efforts and Trends in Data. A
presentation at the Fox River Group Comment
Meeting.
Marti, E.A. and D.E. Armstrong. 1990.
Polychlorinated Biphenyls in Lake Michigan
Tributaries. J. Great Lakes Res., 1 6(3):396-405.
McCarty, H.B.; J. Schofield, K. Miller, R.N. Brent, P.
Van Hoof, and B. Eadie. 2004. Results of the
Lake Michigan Mass Balance Study:
Polychlorinated Biphenyls and frans-Nonachlor
Data Report. U.S. Environmental Protection
Agency, Great Lakes National Program Office,
Chicago, Illinois. EPA-905/R/01 -01 1 , 289 pp.
Miller, S.M., M.L. Green, J.V. DePinto, and K.C.
Hornbuckle. 2001. Results From the Lake
Michigan Mass Balance Study: Concentrations
and Fluxes of Atmospheric Polychlorinated
Biphenyls and frans-Nonachlor. Environ. Sci.
Technol., 35(2):278-285.
Neidermyer, W.J. and J.J. Hickey. 1976.
Chronology of Organochlorine Compounds in
Lake Michigan Fish, 1929-1966. Pest. Monit. J.,
10(3):92-95.
118
-------
Rossmann, R. and D.N. Edgington. 2000. Mercury
in 1987-1990 Green Bay, Lake Michigan Surficial
Sediments. J. Great Lakes Res., 26(3):323-339.
Rossmann, R. 2002. Lake Michigan 1994-1996
Surficial Sediment Mercury. J. Great Lakes Res.,
28(1):65-76.
Schneider, A.R., H.M. Stapleton, J. Cornwell, and
J.E. Baker. 2001. Recent Declines in PAH,
PCB, and Toxaphene Levels in the Northern
Great Lakes as Determined From High
Resolution Sediment Cores. Environ. Sci.
Technol., 35(19):3809-3815.
Simcik, M.F., I. Basu, C.W. Sweet, and R.A. Hites.
1999. Temperature Dependence and Temporal
Trends of Polychlorinated Biphenyl Congeners in
the Great Lakes Atmosphere. Environ. Sci.
Technol., 33(12):1991 -1995.
Swackhamer, D.L. and D.E. Armstrong. 1988.
Horizontal and Vertical Distribution of PCBs in
Southern Lake Michigan Sediments and the
Effect of Waukegan Harbor as a Point Source. J.
Great Lakes Res., 14(3):277-290.
Velleux, M.L. and D. Endicott. 1994. Development
of a Mass Balance Model for Estimating PCB
Export From the Lower Fox River to Green Bay.
J. Great Lakes Res., 20(2):416-434.
Wethington, D.M. and K.C. Hornbuckle. 2005.
Milwaukee, Wl as a Source of Atmospheric PCBs
to Lake Michigan. Environ. Sci. Technol.,
39(1):57-63.
119
-------
PART 2
LM3-EUTRO
James J. Pauer and Katie W. Taunt
Welso Federal Services, LLC
and
Wilson Melendez
Computer Sciences Corporation
Large Lakes Research Station
9311 Groh Road
Grosse lie, Michigan 48138
Chapter 1. Conclusions (Executive
Summary)
LM3-Eutro was developed in conjunction with several
other mathematical models as part of the Lake
Michigan Mass Balance Project (LMMBP). These
models work together to determine contaminant
concentrations in Lake Michigan fish predators under
present and future conditions.
LM3-Eutro was based on the CE-QUAL-ICM model
transport framework (Cerco and Cole, 1995) and
used state-of-the-science eutrophication kinetics to
simulate the interactions between plankton and
nutrients. LM3-Eutro is a high-resolution framework
containing 44,042 water column segments. The
model is driven by the Princeton Ocean
hydrodynamics Model (POM) (Schwab and Beletsky,
1998). A sediment model is under development.
Until developed, LM3-Eutro includes user-defined
fluxes to simulate sediment-water interactions. The
model has 17 state variables, including a single
zooplankton class, two phytoplankton classes, and
several particulate and dissolved nutrient (including
carbon) states.
LM3-Eutro has several advantages over historical
Great Lakes models:
• A high-resol ution segmentation framework (44,042
cells and 19 sigma layers), enabling a better
description of areas such as nearshore and
offshore zones, bays, river confluences, and the
thermocline.
• Use of POM to simulate water movement is a
significant improvement over historical models
which traditionally used tracers, chloride, and
temperature to estimate diffusive and advective
flows.
• The model is carbon-based, which is an
improvement over chlorophyll a due to high
variability of this pigment in phytoplankton.
• The expansion of nutrient variables to include
dissolved, labile particulate, and refractory
particulate forms allows for more realistic
description of phytoplankton-nutrient interactions.
120
-------
• Important improvements were made to the light
calculation by using a three-hour rather than 24-
hour (one day) average estimate of solar radiation.
The 24-hour average approach has been criticized
by some scientists.
Another advantage of this modeling effort was the
large supporting data set. These data were used to
establish atmospheric and tributary loads, estimate
initial conditions, perform model calibration and
confirmation and, to a lesser extent, assist in
estimating a number of kinetic coefficients. Most of
the data were collected during eight sampling cruises
in 1994-1995 (U.S. Environmental Protection
Agency, 1997). Limited data were also collected in
1998 and 2000. The data went through rigorous
quality assurance (QA) and quality control (QC)
procedures (Richardson et a/., 2004). In general,
most of the emphasis was placed on the main lake
as relatively little field data were collected from Green
Bay. Supplemental data were gathered for loads
such as shoreline erosion and internal sediment
fluxes (Monteith and Sonzogni, 1976; Hall and
Robertson, 1998). Most of the kinetic model
coefficients were derived from the literature and
historical Great Lakes models (e.g., Thomann and
DiToro, 1975; Ambrose et at., 1993).
The model was calibrated on the high-resolution
(44,042 cells) Level 3 framework as well as the 41
segment Level 2 framework. The Level 2 calibration
enabled us to visually observe known spatial and
temporal trends such as the spring diatom bloom and
phytoplankton concentration gradients between the
epilimnion and hypolimnion. The Level 3 calibration
was performed on a whole-lake basis. Model output
was compared to field data for different calibration
runs using simple statistical parameters such as
slope and squares of the correlation coefficient. The
1994-1995 LMMBP field data were used to calibrate
the model. The final calibration was chosen based
on the best Level 3 calibration, but Level 2 output
was visually inspected to ensure that expected
phytoplankton and nutrients trends were reflected.
Overall, the calibrated model fits the data well. We
were especially satisfied with how well the model was
able to mimic the particulate organic carbon (POC)
field data because providing carbon production for
use in LM2-Toxic was the most important objective of
this modeling effort. The phytoplankton fit was not as
good, but could be partly explained by the uncertainty
in using an in s/fufluorometer (Seabird Instrument) to
estimate phytoplankton concentrations and the large
natural variation in phytoplankton communities
(Clesceri et at., 1998).
Model confirmation was performed by comparing the
model to limited total phosphorus data for 1998 and
2000 and to a historical model, MICH1, which was
developed and calibrated in the 1970s and modified
more recently. All comparisons were done on a
whole-lake basis, and LM3-Eutro fits the 1998 and
2000 data well. LM3-Eutro and MICH1 compared
surprisingly well, especially given the fact that they
are based on very different frameworks, kinetics, and
segmentation. Compared to field data and LM3-
Eutro predicted, MICH1 underpredicted both total
phosphorus concentrations. This was probably due
to the fact that MICH1 does not have any phosphorus
internal sediment recycle. Lower phosphorus values
also caused MICH1 to underpredict chlorophyll a
concentrations in the lake.
Several model forecast scenarios were performed,
and long-term total phosphorus, POC, and
phytoplankton predictions were observed. One
scenario utilized alternating 1994 and 1995 tributary
and atmospheric phosphorus loads for 30 years. The
autochthonous solid (primary production carbon)
output from this model was used in the contaminant
fate and transport polychlorinated biphenyl (PCB)
model. The model predicted a steady-state total
phosphorus concentration of 4.3 ug/L, a steady-state
POC concentration of 0.2 mg/L, and an epilimnetic
spring chlorophyll a peak of 2.36 ug/L. Steady-state
was reached within 28 years. Several load reduction
scenarios were performed, and total phosphorus,
POC, and chlorophyll a concentrations were
predicted. The Great Lakes Water Quality
Agreement (GLWQA) target total phosphorus load of
5,600 MT was revisited and the impact of increasing
the load to this level was predicted for total
phosphorus and chlorophyll a concentrations in the
lake (International Joint Commission, 1978). Under
the GLWQA loading levels, total phosphorus
concentrations were predicted to reach 7.5 ug/L, and
spring epilimnetic chlorophyll a peaked at 4.0 ug/L.
The model was used to estimate the total
phosphorus loading required to reach the
International Joint Commission's (IJC) total
phosphorus concentration guideline of 7 ug/L (Great
Lakes Research Advisory Board, 1978). The model
121
-------
predicted that an annual load of 5,020 maximum total
(or a Total Maximum Daily Load (TMDL) of 14
maximum total/day) would result in the 7 ug/L steady-
state total phosphorus concentration and a spring
maximum epilimnetic chlorophyll a concentration of
3.7 ug/L.
References
Ambrose, R.B., Jr., T.A. Wool, and J.L. Martin.
1993. The Water Quality Analysis Simulation
Program, WASPS; Part A: Model Documentation.
U.S. Environmental Protection Agency,
Environmental Research Laboratory, Athens,
Georgia. 202 pp.
Cerco, C. and T. Cole. 1995. User's Guide to the
CE-QUAL-ICMThree-DimensionalEutrophication
Model. U.S. Army Corps of Engineers, U.S.
Army Engineer Waterways Experiment Station,
Vicksburg, Mississippi. Technical Report EL-95-
15,2,420pp.
Clesceri, L.S., A.E. Greenberg, and A.D. Eaton
(Eds.). 1998. Standard Methods for the
Examination of Water and Waste Water, 20th
Edition. American Public Health Association,
American Water Works Association, and Water
Environment Federation, Hanover, Maryland.
1,205pp.
Great Lakes Research Advisory Board. 1978.
Annual Report to the International Joint
Commission. International Joint Commission,
Windsor, Ontario, Canada. 44 pp.
Hall, D. and D. Robertson. 1998. Estimation of
Contaminant Loading from Monitored and
Unmonitored Tributaries to Lake Michigan for the
USEPA Lake Michigan Mass Balance Study.
Quality Systems and Implementation Plan.
Submitted October 23,1998. U.S. Environmental
Protection Agency, Great Lakes National
Program Office, Chicago, Illinois. 19 pp.
International Joint Commission. 1978. Great Lakes
Water Quality Agreement of 1978, with Annexes
and Terms of Reference, Between the United
States and Canada. Signed at Ottawa,
November 22, 1978. International Joint
Commission, Windsor, Ontario, Canada. 60 pp.
Monteith, T.J. andW.C. Sonzogni. 1976. U.S. Great
Lakes Shoreline Erosion Loadings. Great Lakes
Basin Commission, Ann Arbor, Michigan. 223
PP-
Richardson, W.L., D.D. Endicott, R.G. Kreis, Jr., and
K.R. Rygwelski (Eds.). 2004. The Lake Michigan
Mass Balance Project Quality Assurance Plan for
Mathematical Modeling. Prepared by the
Modeling Workgroup. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
EPA/600/R-04/018, 233 pp.
Schwab, D.J. and D. Beletsky. 1998. Lake Michigan
Mass Balance Study: Hydrodynamic Modeling
Project. National Oceanic and Atmospheric
Administration, Great Lakes Environmental
Research Laboratory, Ann Arbor, Michigan.
NOAATechnical Memorandum ERLGLERL-108,
53pp.
Thomann, R.V., D.M. DiToro, R.P. Winfield, and D.J.
O'Connor. 1975. Mathematical Modeling of
Phytoplankton in Lake Ontario, Part 1 Model
Development and Verification. U.S.
Environmental Protection Agency, Office of
Research and Development, ERL-Corvallis,
Large Lakes Research Station, Grosse lie,
Michigan. EPA/660/3-75/005, 177 pp.
U.S. Environmental Protection Agency. 1997. Lake
Michigan Mass Balance Study (LMMB) Methods
Compendium, Volume 1: Sample Collection
Techniques. U.S. Environmental Protection
Agency, Great Lakes National Program Office,
Chicago, Illinois. EPA/905/R-97/012a, 1,440pp.
122
-------
PART 2
LM3-EUTRO
Chapter 2. Recommendations
LM3-Eutro captured nutrient and phytoplankton
trends in Lake Michigan and fit the project field data
relatively well. This built confidence in how well the
model will be able to describe the system and predict
total phosphorus, phytoplankton, and particulate
organic carbon (POC) concentrations under different
loading scenarios. However, LM3-Eutro has a
number of limitations and there are several
improvements that can be made to improve the
accuracy and predictive capability of the model.
• Presently the framework does not include a
sediment sub-model and instead uses user-
defined fluxes. A sediment model is
recommended to describe nutrient interactions
between the sediment and water column. A
sediment submodel, coupled with the present
water column model, would provide an integrated
framework that conserves mass in both the water
and sediments.
• The initial conditions were estimated on the Level
2 segmentation scheme. Because the model is
sensitive to the initial conditions, especially over
the first few years of predictions, it would be
preferable to calculate initial condition on the high-
resolution Level 3 segmentation scheme.
• Few laboratory and field measurements were
performed to estimate kinetic coefficients. Limited
laboratory production experiments were
conducted for use in the model. Measurements
estimating Lake Michigan specific coefficients,
especially the phytoplankton growth coefficients,
would improve the reliability of the model.
• Additional field and laboratory data would have
benefitted the construction, calibration, and
confirmation of the LM3-Eutro model.
• Although the lake was sampled eight times during
the 1994-1995 period as part of the Lake Michigan
Mass Balance Project (LMMBP), additional field
measurements would have allowed this high-
resolution model to be better constrained.
• Additional sampling during the spring
phytoplankton bloom period would improve our
understanding of Lake Michigan phytoplankton
dynamics and thus assist in construction and
calibration of the model.
• Laboratory chlorophyll a measurements, in
addition to Seabird fluorescence estimates, should
be used for future phytoplankton estimates.
• No zooplankton concentration estimates below the
thermocline were available for this study.
Hypolimnetic or whole water column zooplankton
tows would improve the accuracy of the model.
• Limited soluble reactive phosphorus (SRP) data
were collected for this study and most of the
results were below the detection limit of 1 ug/L. A
more complete SRP data set, using an analytical
method with a lower detection limit, would be
helpful.
123
-------
Access to an independent data set(s) to confirm
the model would improve the credibility of the
model.
Relatively few samples were taken in Green Bay
during this project. Examination of the Green Bay
Mass Balance Project (GBMBP) data set suggests
that these data are already "out-of-date" due to
improvements in conditions in the bay. A more
complete recent data set for Green Bay would
improve the model's ability to describe the bay
and, to a lesser extent, benefit the overall
mathematical framework.
The hydrodynamics for the project year were
incomplete. A constant overall lake temperature
was assumed for the first three months of 1994 in
the Princeton Ocean Model (POM) hydrodynamics
calculation. No hydrodynamics exist beyond
December 21, 1995. A complete two-year
hydrodynamics data set would be useful during
model calibration and forecast scenario
simulations.
Limited sensitivity analyses (not included in this
document) have been performed to date. No
uncertainty analyses have been performed.
Sensitivity and uncertainty analyses will identify
the effect of the many processes and coefficients
on the model and indicate how accurate the model
predictions are. However, both of these
procedures are a major undertaking in a high-
resolution model like LM3-Eutro due to the large
number of segments and time required to
complete a single simulation. Commonly used
methods such as Monte Carlo analysis are almost
impossible to perform due to the time and disk
space requirements for a single simulation (and
hundreds, if not thousands, of simulations will
probably be necessary). In addition, these
methods provide only an estimate of parameter
uncertainty and do not address issues such as
structure or scenario uncertainty.
A longer record of hydrodynamics data for Lake
Michigan (e.g. 1983-1995) would be useful in
running forecast scenarios. This would allow
determination of "typical" hydrodynamics data for
use in longer model simulations.
Updated phosphorus loading data (and to a lesser
extent nitrogen and silica) would allow for a more
accurate estimation of future loading trends used
in longer model scenarios.
124
-------
PART 2
LM3-EUTRO
Chapter 3. Model Description
The LM3-Eutro model offers the current "state-of-the-
science" in modeling and is capable of providing the
type of spatial information lacking in the MICH1
model (Rodgers and Salisbury, 1981a,b) and many
other historical Great Lakes frameworks (Di Toro and
Connolly, 1980; Thomann and Di Toro, 1975). The
model is based on the standard eutrophication
kinetics used in the WASP family of models
(Thomann and Di Toro, 1975; Ambrose et a/., 1993)
as well as the U.S. Army Corps of Engineers' QUAL
models (Cerco and Cole, 1995). Important
improvements over earlier Great Lakes models
include the high-resolution segmentation and the use
of a sophisticated hydrodynamics model (Princeton
Ocean Model [POM], Schwab and Beletsky, 1998) to
drive the lake's hydrodynamics. Earlier models used
tracers such as chloride or temperature to calibrate
water movement and account for the transport within
a system. These approaches frequently introduced
large uncertainties which are, for the most part,
avoided with the implementation of the POM.
2.3.1 Transport Scheme for Lake Michigan
Considerable attention had been paid to correctly
simulate water column transport in the Lake Michigan
Mass Balance Project (LMMBP). The correct
implementation of hydrodynamics flow and dispersion
and the simulation of concentration gradients had
been identified as key elements of water quality
analysis for the Lake Michigan System, and they
were given particular emphasis throughout the model
development.
The computational transport scheme for the LMMBP
consisted of three linked submodels in which the
output of one submodel was used as input for
another submodel. The models consisted of a
hydrodynamics model that simulated three-
dimensional velocity and temperature fields in the
lake, a wave model, and a particle transport model.
The hydrodynamics model was based on the POM
which was adapted to Lake Michigan by David
Schwab (Schwab and Beletsky, 1998). This model
simulated currents, dispersion coefficients, and water
temperature over a 5 km grid. The grid was three-
dimensional and consisted of 2,318 horizontal cells
and 19 vertical layers that resulted in a total of
44,042 water column segments. This 5 km grid was
also used in, LM3-Eutro. Tributary inflows, the
Chicago River olitflow, and the Straits of Mackinac
were incorporated into POM by David Schwab as
part of the hydrodynamics simulation. The POM
output consisted of water temperature, horizontal and
vertical dispersion, and horizontal and vertical
currents for each segment in the water column. This
output was used as input for LM3-Eutro. The 1994-
1995 POM simulation assumed a constant uniform
water temperature of 2°C for the period January 1 to
March 31, 1994, while no hydrodynamics data were
available after December 21,1995. In order to obtain
hydrodynamics data for the complete 1994-1995
period (used in model calibration and long-term
simulations), the first three months of 1994 were
replaced with January to March 1995 data (including
temperatures), while the corresponding 1994 data
was used for the last 10 days in 1995.
125
-------
The transport model was fairly complex and was
incorporated within LM3-Eutro itself. This transport
model was based on the ULTIMATE QUICKEST
transport scheme, originally developed by Leonard
(1991) and subsequently augmented for use with
variable grid sizes by Chapman et al. (1997).
ULTIMATE QUICKEST was also modified by
Chapman et al. (1997) to incorporate particle settling
velocities into the vertical transport calculation,
resulting in a more realistic simulation of settling in
Lake Michigan. The resulting transport algorithm has
been coded in Fortran and applied to the
Chesapeake Bay (CE-QUAL-ICM) model (Cerco and
Cole, 1994, 1995). A linkage between POM and
LM3-Eutro was developed by Chapman et al. (1997).
The linkage was essentially a mapping of POM cell
numbers with ULTIMATE QUICKEST flow face
numbers and the relationship between horizontal and
vertical components. The LM3-Eutro model
calculation performed numerical integration of
spatially varying particle concentrations using
quadratic interpolation of the concentration to infer its
value at flow faces. It also performed analytic
integration over space and time to account for
changes in the concentration at the cell wall during
each time step. Further details of the dimensional
derivation of ULTIMATE QUICKEST transport
method can be found in Settles et al. (2002).
2.3.2 Sediments
The sediments are leaky sinks of nutrients and
carbon in Lake Michigan. Phytoplankton and
particulate detrital matter containing carbon, nitrogen,
phosphorus, and silica settle to the lake bed and are
recycled back to the water column via resuspension,
diagenesis, and diffusion. The ultimate goal of the
eutrophication modeling effort was to develop a
coupled water column and sediment transport
framework. As a short-term approach, the model
code was modified to incorporate user-specified
sediment fluxes. Although the framework has the
flexibility to specify fluxes for any of the state
variables, we only used fluxes of the dissolved
nutrients (soluble reactive phosphorus [SRP],
ammonia [NHJ, dissolved silica [DSi]) and dissolved
organic carbon (DOC). These fluxes, in effect, are
loads that are evenly distributed over the bottom
sediments. These loads were input into the cells of
the lowest water column layer, with each cell
receiving exactly the same load value. The loads
were, thus, independent of time and space. It is well
documented that the majority of nutrient mass is
recycled within the lake on an annual basis (Meyers
and Eadie 1993). Using this knowledge, nutrient
sediment fluxes were calculated. These values
compared favorably to limited published nutrient
fluxes (Quigley and Bobbins, 1986; Conley, et al.,
1988).
2.3.3 Formulation of Eutrophication
Equations
Two important features of eutrophication models
were the multiple interactions among nutrients,
plankton, and sediments and the complexity of the
transformation reactions describing the conversions
between dissolved and particulate phases. The
model simulated two phytoplankton classes, diatoms
and "non-diatoms," a single herbivorous zooplankton
class, and several nutrient state variables (Table
2.3.1). In a modeling framework, each interaction
was described as a mathematical equation and the
challenge was to define a relatively simple
expression to approximate complex biochemical
processes. Most of the equations formulated and
used here were based on the WASP family of models
(Thomann and Di Toro, 1975; Ambrose era/., 1993)
and the CE-QUAL-ICM model (Cerco and Cole,
1995).
Table 2.3.1. Nutrient State Variables
Nutrient
Phosphorus
Nitrogen
Silica
Carbon
Dissolved
Species
Soluble reactive,
dissolved organic
Ammonia, nitrate,
dissolved organic
Biogenic silica
Dissolved carbon
Particulate Organic
Species
Labile, refractory
Labile, refractory
Unavailable silica
Labile, refractory
The complete set of mathematical equations used in
this model can be found in Appendix 2.3.1. Here we
provide a brief explanation of the changes made in
formulating the equations describing algal light
126
-------
dependence, as this was a significant improvement
over previous eutrophication modeling approaches.
The general equation for expressing net
phytoplankton production is given below.
Net production = gross production - mortality
dp i\f if \ o if 7 (231)
where
P = phytoplankton concentration (mass/volume)
f = time
kg = phytoplankton growth rate (time"1)
kd = phytoplankton mortality rate (time"1)
kg; = predation rate (time"1)
Z = zooplankton concentration (mass/volume)
The growth rate can be written as:
k9-kgma}(f(N)f(T)f(l) (2.3.2)
where
*gmax = optimum growth rate (time"1)
f(N) = nutrient growth dependency
1(1) = light growth dependency
f(T) = temperature growth dependency
A number of equations had been proposed to
describe the effect of light intensity on phytoplankton
production. Steele's equation (Steele, 1962) is one
of the most commonly used expressions, while a light
saturation equation (similar to the Monod equation)
is also frequently used (Di Toro et al., 1971). We
described light dependency in this model according
to Steele's equation:
1(1) = j- exp
s
—+ 1
/„
(2.3.3)
where
1(1) = light limitation (fraction between 0 and 1)
/ = solar light intensity (energy/time/area)
ls = saturating light intensity (energy/time/area)
The Beer-Lambert equation was used to estimate the
light penetration in the water:
/z = /0exp(-frez)
where
(2.3.4)
lz = the light intensity at depth z
(energy/time/area)
/0 = the surface light intensity (energy/time/area)
ke = light extinction coefficient (1/length)
z = depth (length)
Substituting this equation into the previous equation
yields:
„ /Oexp(-/cz)
exp
/0 exp (- ke z)
I,
+ 1
(2.3.5)
This equation calculates the light limitation at an
instantaneous time and at a specific depth.
However, for models like ours, light limitation must be
estimated in a certain cell (with a given depth range)
and over a time period (the time step). Thus, we
needed to integrate this equation over time and
depth. Di Toro et al. (1971) formulated an equation
assuming a constant light intensity over the
photoperiod. They integrated Steele's equation over
a 24-hour period and the total depth of a segment.
127
-------
(2.3.6)
where
where
fd = the photoperiod
la = average light intensity over the
photopheriod (energy/time/area)
This approach is still commonly used, although it has
been criticized for losing the power to represent
midday surface inhibition (Di Toro et a/., 1971;
Kremer and Nixon, 1 978). LM3-Eutro had the luxury
of performing variable time averaging from hourly to
1 2-hour averages and it allowed observation of the
differences in time steps. However, if one wanted to
estimate the light limitation for less than a day and
the average light intensity of that period is known,
one can solve Steele's equation as follows (note: it
is only integrated over depth, but not over time):
z2
z1
exp
- /. exp (- k. z)
L
dz
(2.3.7)
The solution is almost the same as before, without
the fraction of daylight in the equation.
- a,) -exp(-c,0),
The average light intensity (la) can be calculated as
follows:
'.=
J 'o (0
(2.3.9)
where
t
= measured incident solar radiation
(energy/time/area)
= time
and can, thus, be approximated by
1 "
a ~ ~n ,-=1 ° (2.3.10)
where
n = number of discrete time intervals at which I0
is measured
The ability to estimate light limitation on a three-hour
basis (the time interval used in LM3-Eutro) rather
than an average daily basis allowed a more accurate
portrayal of the environment in which phytoplankton
grow. The frequency of light measurements in the
LMMBP allowed an important model improvement.
References
Ambrose, R.B., Jr., T.A. Wool, and J.L. Martin.
1993. The Water Quality Analysis Simulation
Program, WASPS: Part A: Model Documentation.
U.S. Environmental Protection Agency,
Environmental Research Laboratory, Athens,
Georgia. 202 pp.
Cerco, C. and T. Cole. 1994. Three-Dimensional
Eutrophication Model of Chesapeake Bay. U.S.
Army Corps of Engineers, U.S. Army Engineer
Waterways Experiment Station, Vicksburg,
Mississippi. Technical Report EL-94-4, 658 pp.
Cerco, C. and T. Cole. 1995. User's Guide to the
CE-QUAL-ICM Three-Dimensional Eutrophication
Model. U.S. Army Corps of Engineers, U.S.
Army Engineer Waterways Experiment Station,
Vicksburg, Mississippi. Technical Report EL-95-
15,2,420pp.
128
-------
Chapman, R.S., T.M. Cole, and T.K. Gerald. 1997.
Development of Hydrodynamic/Water Quality
(POM-IPXMT) Linkage for the Lake Michigan
Mass Balance Project. Final Report. U.S.
Environmental Protection Agency, Office of
Research and Development, ERL-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
63pp.
Conley, D.J., M.A. Quigley, and C.L. Schelske.
1988. Silica and Phosphorus Flux From
Sediments: Importance of Internal Recycling in
Lake Michigan. Canadian J. Fish. Aquat. Sci.,
45(6): 1030-1035.
Di Toro, D.M., D.J. O'Connor, and R.V. Thomann.
1971. A Dynamic Model of the Phytoplankton
Population in the Sacramento-San Joaquin Delta.
Adv. Chem., 106:131-180.
Di Toro, D.M. and J.P. Connolly. 1980.
Mathematical Models of Water Quality in Large
Lakes. Part 2: Lake Erie. U.S. Environmental
Protection Agency, Office of Research and
Development, ERL-Duluth, Large Lakes
Research Station, Grosse lie, Michigan.
EPA/600/3-80/065, 97pp.
Kremer, J. and S. Nixon. 1978. A Coastal Marine
Ecosystem Simulation and Analysis. Springer
Verlag, New York, New York. 210 pp.
Leonard, B. 1991. The ULTIMATE Conservative
Difference Scheme Applied to Unsteady One-
Dimensional Advection. Comp. Methods Appl.
Mechan. Engin., 88(1 ):17-74.
Meyers, P.A. and B.J. Eadie. 1993. Sources,
Degradation and Recycling of Organic Matter
Associated with Sinking Particles in Lake
Michigan. Org. Geochem., 20:47-56.
Quigley, M.A. and J.A. Robbins. 1986. Phosphorus
Release Processes in Nearshore Southern Lake
Michigan. Canadian J. Fish. Aquat. Sci.,
43(6):1201-1207.
Rodgers, P.W. and D. Salisbury. 1981 a. Modeling
of Water Quality in Lake Michigan and the Effect
of the Anomalous Ice Cover of 1976-1977. Great
Lakes Environmental Planning Study, Great
Lakes Basin Commission, Ann Arbor, Michigan.
Contribution Number 44, 53 pp.
Rodgers, P.W. and D. Salisbury. 1981b. Water
Quality Modeling of Lake Michigan and
Consideration of the Anomalous Ice Cover of
1976-1977. J. Great Lakes Res., 7(4):467-480.
Schwab, D.J. and D. Beletsky. 1998. Lake Michigan
Mass Balance Study: Hydrodynamic Modeling
Project. National Oceanic and Atmospheric
Administration, Great Lakes Environmental
Research Laboratory, Ann Arbor, Michigan.
NOAATechnical Memorandum ERLGLERL-108,
53pp.
Settles, M., W. Melendez, and J. Pauer. 2002. LM3:
The Lake Michigan Mass Balance Model.
Internal Report. U.S. Environmental Protection
Agency, Office of Research and Development,
National Health and Environmental Effects
Research Laboratory, MED-Duluth, Large Lakes
Research Station, Grosse lie, Michigan. 203 pp.
Steele, J.H. 1962. Environmental Control of
Photosynthesis in the Sea. Limnol. Oceanogr.,
7:137-150.
Thomann, R.V., D.M. Di Toro, R.P. Winfield, and D.J.
O'Connor. 1975. Mathematical Modeling of
Phytoplankton in Lake Ontario, Part 1 Model
Development and Verification. U.S.
Environmental Protection Agency, Office of
Research and Development, ERL-Corvallis,
Large Lakes Research Station, Grosse lie,
Michigan. EPA/660/3-75/005, 177 pp.
129
-------
PART 2
LM3-EUTRO
Appendix 2.3.1. Development of LM3-
Eutro Equations
Important characteristics of eutrophication modeling
are the many interactions among nutrients, plankton,
and sediments and the transformation reactions
describing the conversions between dissolved and
paniculate phases. In a modeling framework, each
interaction is described as a mathematical equation,
and the challenge is to define a relatively simple
expression to approximate more complex
biochemical processes. The model equations used
in LMS-Eutro are described here.
A2.3.1.1 Phytoplankton Growth
Although several phytoplankton groups are present
in Lake Michigan, the lake is dominated by diatoms
and flagellates. The major differences between
these classes are silica dependence by the diatoms,
settling rates, carbon content, and growth rates at
different times of the year. It has been speculated
that diatoms grow faster than "non-diatoms" and that
they grow better at cold temperatures because their
blooms are usually observed during the spring in the
Great Lakes. The kinetic equations used in this
model are based on the WASP family of models
(Thomann and Di Toro, 1975; Di Toro and Connolly,
1980; Rodgers and Salisbury, 1981 a, b) and CE-
QUAL-ICM, developed by the U.S. Army Corps of
Engineers (Cerco and Cole, 1993). These equations
do not include settling rates. Settling was included
as part of the model transport in the LM3 model
framework. The basic phytoplankton growth
equation can be written as:
Net Production = Gross Production - Mortality
^ = (ka-kd)P-kgzZ (A2.3.1.1)
g
where
P - phytoplanktonconcentration(mass/volume)
t = time
kg = phytoplankton growth rate (1/time)
kd = phytoplankton mortality rate (1/time)
= predation rate (1/time)
= zooplankton concentration (mass/volume)
The growth rate can be written as:
__ ) (A2.3.1.2)
where
kgmax = optimum growth rate (1/time)
f(N) = nutrient growth dependency
f(l) - light growth dependency
f(T) - temperature growth dependency
For the nutrient growth dependency, we used the
standard Monod equation, but treated diatoms
slightly differently than non-diatoms. Assuming that
Kg
Z
kg = kgmax f(N) f(T)
130
-------
a fraction of the dissolved organic phosphorus (OOP)
was readily available for algal uptake, available
phosphorus, Pav, was defined as follows:
Pav =
f
DOP
(A2.3.1.3)
where
f,
OOP
= fraction of available OOP
DOP= dissolved organic phosphorus
concentration (mass/volume)
For the non-diatoms, the Liebig's law of minimum
applied with no silica dependency.
f(N) = min
NH4 + NO3
av
(A2.3.1.4)
where
NH3 = ammonia concentration (mass/volume)
NO3 = nitrate concentration (mass/volume)
ksat-N - half-saturation coefficient for nitrogen
uptake (mass/volume)
ksat-p = half-saturation coefficient for phosphorus
uptake (mass/volume)
The diatoms were described using the product of the
silica limitation and the minimum of nitrogen and
phosphorus:
f(N) =
Si
mm
NO
(A2.3.1.5)
where
*..
half-saturation coefficient for silica
uptake
The temperature dependency was expressed using
an equation analogous to the Arrhenius temperature
correction. Thus:
f(T) =
where
exp
' M
(A2.3.1.6)
temperature effect below optimum
temperature (°C)
temperature effect
temperature (°C)
above optimum
TM - optimum temperature for phytoplankton
growth (°C)
7 = temperature (°C)
A number of equations had been proposed to
describe the effect of light intensity on phytoplankton
production. Steele's equation (Steele, 1962) is one
of the most commonly used expressions, while a light
saturation equation (similar to the Monod equation)
is also frequently used (Di Toro et ai, 1971). In this
model, light dependency is described according to
Steele's equation.
= — exp
s
where
(A2.3.1.7)
1(1) = light limitation (fraction between 0 and 1)
/ = incident solar light intensity (energy/
time/area)
/s = saturating light intensity (energy/time/area)
The Beer-Lambert equation was used to estimate the
light penetration in the water:
/z = /0 exp (- ke z)
where
(A2.3.1.8)
lz = the light intensity at depth z (energy/
time/area)
131
-------
I0 = the surface light intensity (energy/time/
area)
ke = light extinction coefficient (1/length)
z = depth (length)
Substituting Equation (A2.3.1.8) into Equation
(A2.3.1.7):
1(1
/Oexp(-/rez)
exp
/o exp (- ke z) + 1
L
(A2.3.1.9)
This equation calculated the light limitation at an
instantaneous time and at a specific depth.
However, the need to estimate the light limitation at
a certain cell (with a given depth range) and over a
time period (the time step) was desired. Thus, it was
necessary to integrate this equation over time and
depth. Di Toro et al. (1971) formulated an equation
assuming a constant light intensity over photoperiod.
They integrated Steele's equation (Equation
A2.3.1.9) over a 24-hour period and the total depth of
a segment.
Az
- at) - exp (- a0)]
where
"0 = exp (-*,
a, = y- exp(- kgz2)
(A2.3.1.10)
(A2.3.1.11)
(A2.3.1.12)
and
fd = the photoperiod (time)
la = average light intensity over the photoperiod
(energy/time/area)
This approach is still very commonly used, although
it has been criticized for losing the power to represent
midday surface inhibition (Di Toro et al., 1971;
Kremer and Nixon, 1978). In our model, we had the
luxury of performing variable time averaging from
hourly to 12-hour averages and observing the
difference.
However, if one wanted to estimate the light limitation
for less than a day and the average light intensity of
that period was known, one can solve Steele's
equation as follows (note: it is only integrated over
depth, but not over time):
z1
exp
- la exp (- kg z)
+ I
dz
(A2.3.1.13)
The solution was almost the same as Equation
(A2.3.1.10), without the fraction of daylight in the
equation.
(A2.3.1.14)
The average light intensity (la) here can be calculated
as follows:
>a =
f /o (0 dt
J*
(A2.3.1.15)
where
/„ = measured incident solar radiation
(energy/time/area)
t = time
and can, thus, be approximated by:
la = - I I0(tn) (A2.3.1.16)
where
n = number of discrete time intervals at which I0
is measured.
132
-------
A2.3.1.2 Zooplankton Kinetics
Zooplankton predation is important in regulating the
phytoplankton densities in Lake Michigan, especially
during the stratified summer months (Scavia et at.,
1988). Lake Michigan zooplankton are dominated by
herbivorous species, with copepods making up the
majority of the total biomass for most of the year.
However, cladocerans exhibit significant peaks in the
late summer and fall. Due to the limited zooplankton
data reported for the lake and the lack of kinetic
laboratory and field studies for Lake Michigan
zooplankton populations, we avoided complex
zooplankton equations (e.g., Bowie et a/., 1985), and
chose a relatively simple formulation. The equations
describing herbivorous zooplankton growth were
based on formulations from the literature (Bowie et
a/., 1985; Di Toro and Connolly, 1980; Di Toro and
Matystik, 1980; Thomann and Mueller, 1987).
Carnivorous zooplankton were not directly simulated
here, but were represented in a herbivorous
zooplankton mortality term.
The following equation was used:
ii i \ -* I A1? ^ 1 17\
/p If If \ f ^/^^.O. \ , \ I I
where
Z = zooplankton concentration (mass/volume)
t = time
kgz = growth rate (1/time)
kdz = mortality (1/time)
e = growth efficiency
t-
where
P r.T-T.
(A2.3.1.18)
= maximum growth rate (1/time)
P = diatom and greens concentration
(mass/volume)
ks = half-saturation coefficient (mass/volume)
6 = temperature correction factor
Tu = reference temperature (°C)
The maximum growth rate is a term that lumps the
filtration and assimilation rates into a single term.
The mortality term lumps respiration, excretion, and
higher predation in a single term. We can, thus, write
the overall equation:
dZ
dt
ek.
gzmax
*dz
(A2.3.1.19)
A2.3.1.3 Carbon Interactions
Several carbon interactions were described in the
model, including phytoplankton and zooplankton
carbon, carbon loads from tributaries, shoreline
erosion, and detrital carbon from plankton. The
carbon state variables in this model were diatom,
non-diatom, and zooplankton carbon; labile detrital
carbon; refractory detrital carbon; and dissolved
organic carbon (DOC). Carbon dioxide (CO2) was
not simulated, although a mineralization reaction was
included. Diatom and non-diatom carbon were
simulated, as described in the previous section.
Labile detrital carbon referred to the organic detrital
carbon from the phytoplankton species which breaks
down, as the name implies, relatively rapidly. In
contrast, the refractory detrital carbon is the
combination of the fraction of the plankton breaking
down, as well as other forms of organic carbon in the
system, e.g., carbon from tributaries, the sediments,
etc. These forms of carbon break down slowly, but
are not totally refractory. In the equation, we specify
the fractions of labile and refractory carbon. We
assumed that phytoplankton utilizes CO2 as the
carbon source during photosynthesis and releases
carbon as dissolved (CO2 and DOC) and paniculate
(refractory organic carbon [ROC] and labile organic
carbon [LOG]) forms.
Phytoplankton Mortality and Decay
Phytoplankton respiration and non-predatory
mortality were grouped together in the model as a
"mortality" term. The release of carbon as C02 from
these processes was split into different fractions of
DOC and POC.
133
-------
dDOC
dt
*-p
(A2.3.1.20)
simulate any higher predation such as carnivorous
zooplankton). We assumed that the detrital
zooplankton carbon consisted of dissolved, labile,
particulate, and refractory particulate fractions.
where
fcdm = fraction of DOC from mortality
dROC _ f .
— -j: -- 'cm Ka
where
fcrm = fraction of ROC from mortality
dLOC
dDOC
dt
where
(A2.3.1.26)
(A2.3.1.21) fcdz = fraction of DOC from zooplankton mortality
f
=
clm
(A2.3.1.22)
where
fclm = fraction of LOG from mortality
However, phytoplankton carbon was also converted
to detrital and DOC through predation (messy
feeding) and zooplankton-imposed mortality.
dDOC = f n _ \ k z
JA cdp * ' gz
or
where
fcdp = fraction of DOC from predation
(A2.3.1.23)
where
crp
(A2.3.1.24)
crp
= fraction of ROC from predation
dLOC
dt
where
(A2.3.1.25)
clp
- fraction of LOG from predation
Zooplankton Mortality and Decay
The zooplankton mortality term included respiration,
non-predatory mortality, and predation (we did not
dLOC
dt
= f
clz
(A2.3.1.27)
where
fc,z = fraction of LOG from zooplankton mortality
dROC , . -,
dt
where
(A2.3.1.28)
fcr2 = fraction of ROC from zooplankton mortality
Particulate fractions (both labile and refractory) were
hydrolyzed to DOC, while DOC mineralized to C02.
Since we did not explicitly model bacteria in this
model, their breakdown of carbon was modeled by
including a dependency on the phytoplankton, which
acted as a surrogate of the heterotrophic bacterial
activity in the lake. We also calculated a temperature
limitation to the hydrolysis and mineralization. The
equations can be written as follows:
(A2.3.1.29)
(A2.3.1.30)
Tfhdr = exp(TkMr(T-Trhdr)]
where
Tfmnl = temperature correction for mineralization
(°C)
Tfhdr = temperature correction for hydrolysis (°C)
Tkmn, = mineralization temperature coefficient (°C"1)
Tkhdr = hydrolysis temperature coefficient (°C"1)
134
-------
7rmn/ =
Trhrir =
optimum temperature correction for
mineralization (°C)
optimum temperature correction for
hydrolysis (°C)
dDOC
dt
dROC
dt
dLOC
dt
where
= Tfmn, (kdc
(A2.3. 1 .31 )
(A2.3.1.32)
kdc = DOC minimum mineralization rate (1/time)
kdcp = DOC mineralization relating to
phytoplankton (volume/mass/time)
krc = ROC minimum hydrolysis rate (1/time)
krcp = ROC hydrolysis relating to phytoplankton
(volume/mass/time)
klc = LOG minimum hydrolysis rate (time"1)
kicp = LOG hydrolysis relating to phytoplankton
(volume/mass/time)
Examining the last two equations, we calculated that
the gain in DOC equaled the sum of the loss of ROC
and LOC.
A2.3.1.4 Phosphorus
In our model, phosphorus existed as one of four
species (in addition to being tied up in the
phytoplankton). Note that all four forms were in the
same oxidation state, thus, no oxidation reactions
occurred. The forms were soluble reactive
phosphorus (SRP), DOP, and two forms of
particulate organic phosphorus (POP) - a labile
(LOP) and a refractory (ROP) form. SRP and a small
fraction of the DOP were taken up by the
phytoplankton during production (photosynthesis). It'
was released due to mortality and predation.
Particulate phosphorus was hydrolyzed to DOP and
DOP to SRP.
Phosphorus Uptake by Phytoplankton
Soluble Reactive Phosphorus Uptake:
dSRP
dt
where
= - /•„
SRP
9 SRP +
(A2.3.1.34)
DOP
rpc = the P:C ratio
(A2.3.1.33) Dissolve Organic Phosphorus Uptake:
dDOP
dt
(A2.3.1.35)
where
pc
= the P:C ratio
An interesting concept, common in many
phytoplankton models, is the way in which the
nutrients, including phosphorus, are accounted for
within the phytoplankton. The model kept track of the
carbon and used a constant carbon:nutrient
relationship to make these determinations.
Phytoplankton in the water column were hydrolyzed
and mineralized to all four phosphorus forms. During
algal metabolic/mortality processes, phytoplankton-
phosphorus was converted to particulate and
dissolved organic forms as well as directly to SRP.
dSRP
dt
dLOP
r f k
rpc 'dop Kd
-r f k P
rpc 'lop Kd r
r f
rpc
-------
where
fsrp = fraction SRP from metabolic processes
fdop = fraction DOP from metabolic processes
flop = fraction of LOP from metabolic processes
frop = fraction of ROP from metabolic processes
During the phytoplankton predation, zooplankton
assimilated only a fraction of the phytoplankton and
the remainder of the detrital phytoplankton was
released directly to the water. This process is
commonly referred to as "messy feeding." The
phosphorus was released in both the dissolved and
particulate forms.
dSRP _
dt
dDOP
dt
dROP
dt
dLOP
dt
d -
rpc kgz
•pc "0z
(A2.3.1.40)
(A2.3.1.41)
(A2.3.1.42)
where
fpip = fraction of SRP from predation
fpdp = fraction of DOP from predation
fplp - fraction of LOP from predation
•prp
- fraction of ROP from predation
The model also included equations to describe
zooplankton mortality. Phosphorus was released to
the water column in both the dissolved and
particulate forms.
_ f k
'piz rpc Kdz
f r k Z
'pdz 'pc Kdz *-
(A2.3.1.45)
dLOP _
dt
dROP
dt
where
pc
(A2.3.1.46)
(A2.3.1.47)
fPiz = fraction of SRP from zooplankton mortality
fpdz = fraction of DOP from zooplankton mortality
fplz = fraction of LOP from zooplankton mortality
•prz
- fraction of ROP from zooplankton mortality
Particulate phosphorus was hydrolyzed to DOP and
DOP was mineralized back to SRP as follows:
dDOP
dt
dLOP
dt
"'sat-pt
k*»PSRP+k.
'sat-ptj
sat-pt
^sat-pt)
(A2.3.1.43) dRQp
dt
-k
(A2.3.1.48)
(A2.3.1.49)
(A2.3.1.50)
nsat-pt)
where
kdp = DOP mineralization coefficient (1/time)
klp = LOP hydrolysis coefficient (1/time)
Vp
= ROP hydrolysis coefficient (1/time)
kdpa = DOP mineralization coefficient algal
dependence (volume/mass/time)
klpa = LOP hydrolysis coefficient algal
dependence (volume/mass/time)
(A2.3.1.44) k,
rpa = ROP hydrolysis coefficient algal
dependence (volume/mass/time)
ksat-Pt = mean saturation coefficient of algal classes
for SRP
136
-------
A2.3.1.5 Nitrogen
The transformation of nitrogen was similar to
phosphorus, although nitrogen existed in more than
one oxidation state. Algal nitrogen was released as
two forms of particulate organic nitrogen (labile
[LON], refractory [RON]), dissolved organic nitrogen
(DON), and ammonia (NH4). Particulate forms were
hydrolyzed to DON. DON was further mineralized to
NH4 and NH4 is oxidized to nitrate (NO3).
In our model, we assumed that phytoplankton had no
preference between NH4 and NO3 as a nitrogen
source.
dDIN
dt
~ rnc kg P
(A2.3.1.51)
where
DIN = NH4 + NO3
rnc - N:C ration
Because we assumed no preference, then
NH4
dt
dN03
dt
DIN
NO,
(A2.3.1.52)
DIN
(A2.3.1.53)
Similar to phosphorus, nitrogen bound to
phytoplankton can be released as particulate organic,
dissolved organic and NH4 forms.
dNH4
_
= rnc 'din
_ f
ft ~ rnc 'don Kd ^
(A2.3.1.54)
(A2.3.1.55)
where
fdin = fraction NH4 from metabolic processes
'don
'Ion
= fraction DON from metabolic processes
= fraction LON from metabolic processes
fron = fraction RON from metabolic processes
As described for carbon and phosphorus, the
nitrogen balance was affected by the zooplankton
through "messy feeding" and zooplankton mortality.
dt
dDON
dt
dLON
dt
dRON
dt
nc
nc
(A2.3.1.58)
(A2.3.1.59)
(A2.3.1.60)
nrp'
(A2.3.1.61)
where
fnip = fraction of NH4 from predation
fndp = fraction of DON from predation
fnlp = fraction of LON from predation
'nrp
= fraction of RON from predation
The release of nitrogen during zooplankton mortality
can be expressed similarly to the phosphorus.
Qf/VH4
dt
rnc kdz
(A2.3.1.62)
dLON
_
~ rnc 'ion Kd
(A2.3.1.56)
dDON
rnc kd.
(A2.3.1.63)
r f
'nc 'ran
(A2.3.1.57)
dt
nc
(A2.3.1.64)
137
-------
dRON
dt
(A2.3.1.65)
where
fniz = fraction of NH4 from zooplankton mortality
'ndz
nlz
- fraction of DON from zooplankton mortality
= fraction of LON from zooplankton mortality
= fraction of RON from zooplankton mortality
A2.3.1.6 Silica
The behavior of silica was similar to that of
phosphorus and nitrogen. Two silica species,
biogenic silica (SU) and available silica (SA), were
simulated in the lake. Dissolved silica was utilized by
phytoplankton, while both dissolved and biogenic
silica were released via phytoplankton mortality,
predation upon phytoplankton by zooplankton, and
zooplankton mortality. The major difference from the
other nutrients was that only diatoms had a silica
dependency.
The diatom silica consumption can be written as
follows:
dt
where
k P
P
(A2.3.1.66)
rsc = Si:C ratio
Note that in all the silica equations, the variable
phosphorus refers only to the diatom concentration.
Like the other nutrients, silica was released via
diatom mortality.
dSU
dt
(A2.3.1.67)
Both classes of silica could be released via
zooplankton predation.
dSA
,
— r r If
sap 'sc ngz
dSU
dt
where
= - f r k
sup 'sc ngz
(A2.3.1.69)
fsap = fraction of SA from predation
fsup = fraction of SU from predation
We assumed that no silica accumulated within the
zooplankton so there were no terms for silica release
from zooplankton mortality.
References
Bowie, G.L., W.B. Mills, D.B. Porcella, C.L
Campbell, J.R. Pagenkopf, G.L. Rupp, K.M.
Johnson, P.W.H. Chan, S.A. Gherini, and C.E.
Chamberlin. 1985. Rates, Constants and Kinetic
Formulations in Surface Water Quality Modeling,
2nd Edition. U.S. Environmental Protection
Agency, Environmental Research Laboratory,
Athens, Georgia. EPA/600/3-85/040, 455 pp.
Cerco, C.F. and T. Cole. 1993. Three-Dimensional
Eutrophication Model of Chesapeake Bay. J.
Environ. Engin., 119(6):1 006-1 025.
Di Toro, D.M., D.J. O'Connor, and R.V. Thomann.
1971. A Dynamic Model of the Phytoplankton
Population in the Sacramento-San Joaquin Delta.
Adv. Chem., 106:131-180.
Di
Di
Toro, D.M. and J.P. Connolly. 1980.
Mathematical Models of Water Quality in Large
Lakes. Part 2: Lake Erie. U.S. Environmental
Protection Agency, Office of Research and
Development, ERL-Duluth, Large Lakes
Research Station, Grosse lie, Michigan.
EPA/600/3-80/065, 97 pp.
Toro, D.M. and W.F. Matystik, Jr. 1980.
Mathematical Models of Water Quality in Large
Lakes. Part 1 : Lake Huron and Saginaw Bay.
U.S. Environmental Protection Agency, Office of
Research and Development, ERL-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
EPA/600/3-80/056, 180 pp.
(A2.3.1.68)
138
-------
Kremer, J. and S. Nixon. 1978. A Coastal Marine
Ecosystem Simulation and Analysis. Springer-
Verlag, New York, New York. 210 pp.
Rodgers, P.W. and D. Salisbury. 1981 a. Modeling
of Water Quality in Lake Michigan and the Effect
of the Anomalous Ice Cover of 1976-1977. Great
Lakes Environmental Planning Study, Great
Lakes Basin Commission, Ann Arbor, Michigan.
Contribution Number 44, 53 pp.
Rodgers, P.W. and D. Salisbury. 1981b. Water
Quality Modeling of Lake Michigan and
Consideration of the Anomalous Ice Cover of
1976-1977. J. Great Lakes Res., 7(4):467-480.
Scavia, D., G.A. Lang, and J.F. Kitchell. 1988.
Dynamics of Lake Michigan Plankton: A Model
Evaluation of Nutrient Loading, Competition, and
Predation. Canadian J. Fish. Aquat. Sci.,
45(1 ):165-177.
Steele, J.H. 1962. Environmental Control of
Photosynthesis in the Sea. Limnol. Oceanogr.,
7:137-150.
Thomann, R.V., D.M. Di Toro, R.P. Winfield, and D.J.
O'Connor. 1975. Mathematical Modeling of
Phytoplankton in Lake Ontario, Part 1 Model
Development and Verification. U.S.
Environmental Protection Agency, Office of
Research and Development, ERL-Corvallis,
Large Lakes Research Station, Grosse Me,
Michigan. EPA/660/3-75/005, 177 pp.
Thomann, R.V. and J.A. Mueller. 1987. Principles of
Water Quality Modeling and Control. Harper and
Row Publishers, New York, New York. 644 pp.
139
-------
PART 2
LM3-EUTRO
Chapter 4. Model Input and Field Data
2.4.1 Loading and Sediment-Water
Interactions
2.4.1.1 Atmospheric Loads
Measurements were made at eight locations around
Lake Michigan (Miller et a/., 2000; U.S.
Environmental Protection Agency, 1997) and loads
were calculated for total phosphorus, total Kjeldahl
nitrogen (TKN), and nitrate (NO3). Monthly total
loads were available for March 1994 through October
1995 (Table 2.4.1). Table 2.4.1 shows the
phosphorus loads for this period. In order to obtain
a complete two-year record (necessary for model
calibration and forecast simulations), January and
February of 1994 were assumed to be the same as
January and February of 1995, while November and
December of 1995 were assumed to be the same as
November and December of 1994. The total
phosphorus loads were split between labile organic
phosphorus (LOP) (67% of total phosphorus) and
soluble reactive phosphorus (SRP) (33% of total
phosphorus). All other forms were assumed to be
insignificant. We assumed that the TKN atmospheric
loading is split evenly between labile organic nitrogen
(LON) and refractory organic nitrogen (RON) forms.
2.4.1.2 Tributary Loads
Loads from 11 monitored tributaries were calculated
using the stratified Beale ratio estimator model (Hall
and Robertson, 1998). Loads from 18 unmonitored
tributaries (two of which represented portions of
monitored tributaries) were also estimated based on
results from monitored watersheds and individual
watershed and flow attributes (Hall and Robertson,
Part 7, Appendix 2). Monitored tributaries were
sampled at sites as far downstream as possible to
provide the most accurate load estimates.
Composite samples were prepared from two depths
at three points along a cross-sectional transect of the
river. Most samples were taken during high flow
Table 2.4.1. 1994-1995 Monthly Atmospheric
Total Phosphorus Loads
1994 Atmospheric Total Phosphorus Loads
(kg/month)
March
April
May
June
July
August
September
October
November
December
17552
34665
27465
14429
38184
38303
29908
17334
32418
7937
1995 Atmospheric Total Phosphorus Loads
(kg/month)
January
February
March
April
May
June
July
August
September
October
22748
8430
17552
34665
27465
14429
26457
39458
18893
43254
140
-------
periods (Hall and Robertson, 1998). Loads provided
in the original Great Lakes National Program Office
(GLNPO) data set included chlorophyll a, dissolved
organic carbon (DOC), participate organic carbon
(POC), total phosphorus, SRP, TKN, ammonia-N
(NH4), NO3, and dissolved silica (DSi). Daily loads
were provided in units of kg/d for the period of
January 1,1994 to December 1, 1995. Table 2.4.2
provides a summary of the total phosphorus loads for
the 11 monitored tributaries.
Loads provided for each tributary were used to
calculate additional parameters of interest.
Chlorophyll a loads were converted to phytoplankton
carbon by assuming a 40:1 carbon-to-chlorophyll
ratio. This ratio was chosen to maintain consistency
within the model (see Section 2.4.2.2). This carbon
value was then converted into diatom carbon and
non-diatom carbon by assuming that tributary
phytoplankton populations were 75% diatom and
25% non-diatom (Allan, 1995). Labile paniculate
organic carbon (LOG) and refractory particulate
organic carbon (ROC) were estimated by subtracting
total algal carbon from POC and multiplying by 0.55
for LOG and 0.45 for ROC. Algal phosphorus was
estimated by assuming a phosphorus:carbon ratio of
0.01 (algal carbon multiplied by .01). Organic
phosphorus was taken to be total phosphorus minus
the sum of algal phosphorus and SRP- From the
estimate of organic phosphorus, dissolved organic
phosphorus (OOP) was assumed to be 10% and
LOP and refractory organic phosphorus (ROP) were
both assumed to be 45%. Algal nitrogen was
estimated using a nitrogen:carbon ratio of 0.2 (algal
carbon multiplied by 0.2). Organic nitrogen was
calculated as TKN minus the sum of algal nitrogen
and NH4. As in the case of phosphorus, dissolved
organic nitrogen (DON) was represented by 10% of
organic nitrogen, while labile organic nitrogen (LON)
and refractory organic nitrogen (RON) were each
represented as 45% of organic nitrogen.
2.4.1.3 Shoreline Erosion
Shoreline erosion, mainly along the western shore,
contributes significantly to the solids concentration in
Lake Michigan. The shoreline erosion estimates
were based on the long-term, county-level estimates
of Monteith and Sonzogni (1976). David Schwab,
National Oceanic and Atmospheric Administration
(NOAA), Great Lakes Environmental Research
Laboratory (GLERL), Ann Arbor, Michigan, used
these estimates to calculate erosion loads of coarse-
Table 2.4.2. Tributary Total Phosphorus Loads (kg/year)
River
Total
1994
1995
2676481
2315700
Two-Year Average
Menominee
Fox
Sheboygan
Milwaukee
Calumet
St. Joseph
Kalamazoo
Grand
Muskegon
Pere Marquette
Manistique
Total Monitored Tributaries
Total Unmonitored Tributaries
83753
562865
28424
33731
44710
275772
176318
663972
62490
34937
25966
1992937
683544
127281
595991
21703
31320
39782
264341
137918
351250
43497
26828
25367
1665276
650424
105517
579428
25063
32525
42246
270057
157118
50761 1
52993
30882
25667
1829107
666984
2496091
141
-------
and fine-grained particles (personal communication).
Organic carbon makes up a very small fraction of the
bluff material (Monteith and Sonzogni, 1976) so we
used a carbon fraction of 0.5% for the fine-grain
material in estimating the POC erosion loads to the
lake. We assume this POC is in the refractory form.
2.4.1.4 Sediment
A brief description of the sediment component of the
model was previously provided in the model
description section (Part 2, Chapter 3). A summary
of the phosphorus fluxes, settled masses, and
reported literature values can be found in Table
2.4.3.
Table 2.4.3. Sediment Masses, Fluxes, and Loads
Mass Mass Literature
State Settled Recycled Comparison
Variable (kg/year) (kg/year) (kg/year)
Phosphorus 7x106 4x106 3.6-12 x106*
1.1 x106"
* Quigley and Bobbins, 1986.
**Conley et al., 1988.
2.4.2 Field Data
Large amounts of data were collected between April
1994 and October 1995 during eight sampling
cruises (Table 2.4.4). Sampling stations were
scattered throughout the lake (Figure 2.4.1). The
data sets included lake nutrient concentrations;
physical measurements such as solar radiation and
temperature; and biological data related to
phytoplankton, zooplankton, and fish communities in
the lake. These data have been used to describe the
current state-of-the-lake and to gain a better
understanding of the lake as a whole and the
processes affecting it. They were also useful in
model calibration. Many of the samples collected
were analyzed in situ or on the ship immediately
following collection, while others were carefully
preserved and sent out for analysis by several
laboratories around the country. Detailed
descriptions of sampling techniques and sample
analyses used can be found in the Lake Michigan
Table 2.4.4. The LMMBP Sampling Cruises
Cruise
Number
Cruise 1
Cruise 2
Cruise 3
Cruise 4
Cruise 5
Cruise 6
Cruise 7
Cruise 8
Start Date
April 24, 1994
June 17, 1994
Augusta, 1994
October 14, 1994
January 16, 1995
March 23, 1995
Augusts, 1995
September 16, 1995
End Date
May 11, 1994
June 26, 1994
August 26, 1994
November 7, 1994
January 25, 1995
April 18, 1995
August 16, 1995
October 13, 1995
Mass Balance Project (LMMBP) Methods
Compendium (U.S. Environmental Protection
Agency, 1997).
All LMMBP data were subjected to rigorous water
quality assurance (QA) procedures (U.S.
Environmental Protection Agency, 2004). Once
available to the modelers, data were examined for
completeness and content. Any data which seemed
suspect (unusually high or low values as compared
to historical data, missing values or codes, etc.) were
resubmitted to GLNPO for additional examination.
Data which appeared reasonable and complete were
subjected to a standardized data assessment
protocol by individual modelers. This data
assessment provided basic statistical information
about the data (mean, minimum, maximum, median,
standard deviation), identified outliers, and evaluated
sample normality. In some cases, data averaging
and grouping were necessary before adequate
assessment could be performed. Once the data
assessment was completed by the modeling team,
data were imported into the modeling database and
kept unchanged for the rest of the data evaluation,
model development, and model validation. Summary
statistics for the nutrient data are shown in Table
2.4.5. Due to the limited number of Green Bay
samples, the table includes only the open lake
statistics.
2.4.2.1 Open Lake Nutrient and Carbon Data
2.4.2.1.1 Total Phosphorus
Total phosphorus represented the sum of all
phosphorus species in the sample, including the
142
-------
Muskegon River
Cterand River
Sample Type
water survey stations
Q\ biota survey stations
D tributary monitoring
stations
Sheboygan
River
Milwauke
River
•.vat'id Kalamazoo River
. Joseph River
Chicago
Grand Calumet
/ Harbo?
Figure 2.4.1. The LMMBP sampling locations.
143
-------
Table 2.4.5. The LMMBP Open Lake Nutrient Data Summary Statistics
Nitrate
SRP
Total Phosphorus
DSi
TKN
NH4
POC
DOC
Number
of
Samples
847
504
846
847
845
505
363
364
Minimum
0.01
0.0
1.8
0.038
0.01
0.0
36.15
1.05
Maximum
0.45
6.70
28.7
2.11
0.43
0.30
989.26
2.9
Median
0.28
0.60
4.6
0.52
0.15
0.01
201.7
1.54
Mean
0.27
0.71
5.0
0.54
0.16
0.02
203.0
1.55
Standard
Deviation
0.064
0.69
2.1
0.29
0.07
0.03
92.0
0.19
Outliers*
None
19
31
10
10
13
6
9
Note: POC, SRP, and total phosphorus are expressed in ug/L, all others are in mg/L.
*Larger than twice the standard deviation.
phosphorus dissolved in the water; phosphorus
sorbed to particles such as iron, calcium, and
magnesium; and the phosphorus contained within the
phytoplankton, zooplankton, and detrital particles.
Large spatial and temporal changes were not
expected in the lake over the project years (1994-
1995) because of the conservative nature of total
phosphorus. Changes were typically limited to
particulate settling and incoming and outgoing loads.
It appeared that the LMMBP data verified this,
suggesting small, if any, changes in concentration
and no apparent nearshore/offshore or north/south
trends in the lake. Seasonal trends were observed,
with complete mixing early in the year and slightly
higher total phosphorus in surface waters during the
spring/early summer bloom. Total phosphorus was
lower in the surface waters later in the summer,
possibly due to settling out of the algal phosphorus.
Higher concentrations were observed at the bottom.
On a few occasions, unexpectedly high total
phosphorus values were observed. These may
result from local inputs (tributary) or natural or
sediment disturbances during sampling.
2.4.2.1.2 Dissolved Phosphorus
Dissolved phosphorus was defined as the
concentration of phosphorus found in a sample after
filtration through a membrane filter. SRP, the
preferred form of phosphorus used by algae, was
some portion of dissolved phosphorus. A general
seasonal concentration trend in dissolved
phosphorus was observed on a lake-wide basis.
Early in spring, the concentration was relatively low,
with values just above 2 ug/L. There was a slight
increase in early summer, followed by a decrease in
summer to a level frequently below detection limits.
Dissolved phosphorus increased in the fall and the
pattern repeated for the following year.
2.4.2.1.3 Soluble Reactive Phosphorus
SRP is one of the most important nutrients because
it is widely considered to be the driving force for algal
primary productivity in Lake Michigan (Tarapchak
and Nalewajko, 1987). There has been considerable
discussion in the literature about the meaning,
measurement, and role of SRP, but most agree that
SRP levels can be used to predict algal growth
(Tarapchak and Nalewajko, 1987). SRP was not
analyzed for samples collected during the first three
cruises and the majority of the data from the other
five cruises fell below the detection limit of 1 ug/L
(U.S. Environmental Protection Agency, 1997). The
remarkably good correlation between SRP and
dissolved phosphorus, which was previously
discussed, was useful in making estimates of SRP
for the first two cruises. However, because of the
lack of actual SRP data, we did not speculate about
trends in the lake. This weakness in this important
data set made the analysis and subsequent modeling
exercise difficult.
144
-------
2.4.2.1.4 Nitrate
Nitrate analysis methods actually measure the sum
of N03 and nitrite (NO2). Nitrite values were
assumed to be low enough to be considered
negligible. No obvious nitrate spatial trends were
observed (nearshore versus offshore or northern
basin versus southern basin), but it appeared that the
concentration, on average, was slightly lower in the
summer than in the winter. This was probably due to
the uptake of dissolved nitrogen during phytoplankton
production in the summer.
2.4.2.1.5 Ammonia
Ammonia is the most reduced nitrogen form and is,
therefore, the most available for algal uptake. It
occurred in the lake at very low concentrations, often
below the detection limit of 20 ug N/L. Though the
data set was incomplete because no samples for
ammonia analysis were taken during the first three
sampling cruises, no obvious spatial or temporal
trends were observed.
2.4.2.1.6 Total Kjeldahl Nitrogen
TKN is a measure of all of the reduced nitrogen
present in the water, including organic nitrogen
(particulate and dissolved) and ammonia. No spatial
or temporal trends were observed for the TKN values
but noticeably higher values were observed
throughout the lake during the August 1994 cruise.
2.4.2.1.7 Dissolved Silica
Only the dissolved form of silica (DSi) was measured
in this study. A reasonably good representation of
the open lake concentration could be constructed.
Silica concentrations followed a distinct pattern, with
highest observed values occurring uniformly in the
lake early in the year. The silica was depleted during
the spring and summer by diatom consumption in the
epilimnion, while silica increased in the lake's
hypolimnion during this part of the year, mainly due
to diatom settling and detrital silica (Laird et at.,
1988). Toward the end of the summer and early fall,
silica, in the strongly stratified epilimnion, decreased
to approximately 0.2 mg/L, while it was greater than
1 mg/L in the hypolimnion. This seasonal trend was
observed for both project years. No obvious
differences could be observed between the Michigan
or Wisconsin shores or between the southern and
northern parts of the lake. Epilimnion values tended
to be higher during the summer in shallow nearshore
sites than in deeper open lake sites.
2.4.2.1.8 Dissolved Organic Carbon
DOC remained remarkably constant in Lake
Michigan over the two-year period. We observed few
spatial or temporal trends in the lake, although
significantly higher and lower concentrations were
observed at individual stations in the lake.
2.4.2.1.9 Particulate Organic Carbon
As expected, there was a large variation in POC
concentrations in the lake. Typical concentrations for
the open lake ranged from 100 to 300 ug/L. In Lake
Michigan, POC consisted mainly of phytoplankton
carbon, detrital carbon, and to a lesser extent,
zooplankton carbon. High POC was strongly related
to the timing and locations of phytoplankton blooms.
In general, POC was higher in the euphotic zone
during the warmer summer months. Early in the
spring, POC was higher in the nearshore, probably
due to higher temperatures which resulted in early
spring phytoplankton production.
2.4.2.1.10 Green Bay Nutrient Data
While Lake Michigan is classified as an oligotrophic
system, Green Bay is eutrophic and has drastically
different properties than the Lake Michigan proper.
Green Bay exhibited much higher concentrations of
nutrients and large phytoplankton and zooplankton
populations (and, thus, higher carbon
concentrations). Most state variables had a
concentration gradient, with highest levels (several
times higher than the open lake) close to the Fox
River mouth and lowest concentrations close to the
confluence with the lake. This gradient was
especially prominent for phosphorus, phytoplankton,
and carbon. During the LMMBP (1994-1995), Green
Bay was only sampled in two locations (Figure 2.4.1).
This lack of data complicated the estimation of many
state variables. Available historical data and scaling
of open lake data were used to estimate
concentrations in many instances (Bierman et a/.,
1992; DeStasio and Richman, 1998; Sager and
Richman, 1991).
145
-------
2.4.2.2 Plankton
2.4.2.2.1 Phytoplankton
Data were collected during eight cruises between
April 1994 and October 1995. Samples represented
composites of 1, 5,10, and 20 m sub-samples. Data
were communicated by group densities and
biovolumes (diatoms, "all else" (primarily flagellates),
greens, non-nitrogen fixers, and nitrogen-fixers) and
species densities and biovolumes. Sampling stations
were distributed through the lake (Figure 2.4.1).
Sampled phytoplankton populations differed in overall
density and biovolume in 1994 and 1995. Diatoms
and "all else" occurred in higher numbers in 1994
than 1995, while greens and blue-greens occurred in
similar numbers during both years. This density
difference was reflected in the 1994 and 1995
biovolume data. Overall, phytoplankton biovolume
was much higher in 1994 as a result of higher diatom
and "all else" biovolume. This finding could be the
result of sampling which was not evenly divided
across the calendar year, with 1994 being spring-
weighted and 1995 being fall-weighted.
Blue-green algae (non-nitrogen fixers and nitrogen-
fixers categories) dominated the samples in the total
number of cells present. Blue-greens were the
dominant cell type present in all months. Peak
densities of blue-greens occurred in August-October
1994 and August-September 1995. Peak densities
of diatoms were observed in May-June 1994 and
April-August 1995. "All else" category phytoplankton
peaked in number in May-June 1994, again in
October-November 1994, and then remained stable
throughout the 1995 sampling months.
Diatoms dominated phytoplankton biovolume in April-
June 1994 and again in January-August 1995. "All
else" phytoplankton dominated total biovolume in
October 1994, November 1994, and September
1995. Diatoms and "all else" contributed similarly to
total phytoplankton biovolume in August 1994 and
October 1995. Green algae and blue-green algae
contributed slightly more to total biovolume in
August-October 1994 and August-October 1995 but
never contributed more than approximately 20% and
15%, respectively. In general, diatoms and "all else"
composed >75% of the total phytoplankton
biovolume every month, while the blue-greens
contributed approximately 6% of the phytoplankton
biomass.
Average sizes for each phytoplankton category
further supported the biovolume data. Diatoms
averaged 898.6 um3/cell, "all else" 574.3 urrvYcell,
greens 374.4 um3/cell, non-nitrogen fixers 12.2
um3/cell and nitrogen-fixers 167.7 um3/cell. Because
total carbon content was expressed as a function of
cell biovolume and diatoms and "all else" dominated
the total biovolume of the epilimnetic waters, it was
safe to assume that the major phytoplankton carbon
source would be the diatom and "all else"
phytoplankton categories. The blue-greens, although
high in numbers, made up an insignificant
percentage of phytoplankton carbon mass.
2.4.2.2.2 Chlorophyll a
Chlorophyll a data were provided by GLNPO for the
1994-1995 LMMBP field season. Data were
collected using a Seabird fluorometer and calibrated
to extracted chlorophyll a data. Due to laboratory
error, extracted chlorophyll a data from all cruises
except Cruise 8 (September-October 1995) were
declared invalid. Thus, Seabird data for the 1994-
1995 sampling season were calibrated with fall 1995
and 1997 extracted chlorophyll a data (Goldsmith,
1999). This was accepted as the best alternative,
and the chlorophyll a profiles generated from the
calibrated data generally agreed with trends and
overall concentration levels expected for the lake.
Raw, station-specific chlorophyll a data files
contained information such as station code and
location, date, time, depth of measurement,
chlorophyll a (mg/L), and percent transmissivity.
Most chlorophyll a depth profile measurements were
taken in 0.1-0.5 m increments, although occasionally
only 1 m increments were provided. Some level of
"cleaning" was required for all files. All data with
depth measurements less than or equal to zero
meters were discarded in the analysis, as were data
reporting a measurement of -0.18. Both of these
values were utilized as data flags by GLNPO. In
addition, chlorophyll a data near the surface or
bottom were frequently reported as a long-series of
identical measurements. The chlorophyll a profile
used in the analysis included the last of these
"repeats" and its coordinating depth if the repeats
occurred at surface depths, or the first of the
146
-------
"repeats" and coordinating depth if occurring at lake
bottom. These repeating values were likely the result
of equipment limitations and sampling error (hitting
bottom, etc.) and could not be deemed reliable.
2.4.2.2.3 Phytoplankton Carbon
The eutrophication model required phytoplankton to
be expressed as carbon and divided into diatom and
non-diatom classes. Multiple data transformations
were necessary to satisfy these requirements. The
determination of which approach should be used to
estimate phytoplankton carbon was a complicated
first step. The LMMBP data set included
phytoplankton biovolume data from 0-20 m integrated
samples as well as chlorophyll a depth profiles.
Biovolume data could be converted to phytoplankton
carbon using equations published by Strathmann
(1967) and Rocha and Duncan (1985). While this
approach is generally accepted in the scientific
community, some researchers question whether it is
possible to avoid propagating error using this method
(Sicko-Goad etal., 1984). In calculating biovolume,
organism dimensions are measured and then
multiplied to yield cubic volume. Any measurement
error is, thus, cubed and then further compounded by
inclusion of the erroneous value in the volume-to-
carbon equation. The microscopic nature of
phytoplankton makes some degree of measurement
error inevitable. Another issue was the presence of
vacuoles and thick walls in some phytoplankton
species. These would be included in a microscopic
measurement of an organism as biovolume but
contribute relatively little to the carbon content of the
organisms (Sicko-Goad et al., 1984). In addition to
methodological difficulties, the limitation of
phytoplankton biovolume data to integrated samples
from the top 20 m of the water column made it
difficult to estimate phytoplankton carbon for discrete
depths and deeper waters using these data.
Another method of estimating phytoplankton carbon
is converting chlorophyll a using a carbon-to-
chlorophyll a ratio. This approach also has
shortcomings. Carbon-to-chlorophyll ratios may vary
with species and light and nutrient conditions. Some
researchers have found that the variation was
greatest under nutrient limitation, a common
occurrence in Lake Michigan (Riemann etal., 1989).
The chlorophyll a calibration difficulties encountered
during the LMMBP, and discussed earlier, further
complicated the issue, as the chlorophyll a data set
was not as reliable as desired. The LMMBP
chlorophyll a data set, however, was quite thorough
and any error contained within it as a result of actual
measurement or calibration was probably consistent
across the entire data set. The chlorophyll a data
set also lent itself to comparison with the large
volume of historical data from Lake Michigan, as well
as measurements taken in Green Bay as part of the
Green Bay Mass Balance Project (GBMBP) modeling
effort (Bierman etal., 1992).
A cruise-by-cruse comparison of biovolume and
chlorophyll a derived carbon data for the entire lake
was made (Figure 2.4.2). Chlorophyll a values from
the top 20 m of the water column were averaged and
converted to carbon using several commonly cited
carbon-to-chlorophyll a ratios (35:1, 40:1, and 50:1)
(Riemann et al., 1989; Montagnes et al., 1994;
Cloern et al., 1995). Visual analysis of the results
presented in Figure 2.4.2 suggested that a 40:1
carbon-to-chlorophyll ratio provided the best fit with
biovolume carbon data over all eight sampling
cruises. It was our belief that this chlorophyll a
carbon estimation approach provided the greatest
consistency among integrated 0-20 m samples,
deeper water samples, and Green Bay estimates,
and it provided the best fit to biovolume carbon
estimation methods.
The40:1 carbon-to-chlorophyll relationship was used
to generate carbon values for model fitting exercises.
Chlorophyll a values for each station and cruise were
converted to carbon at each depth along the depth
profile, and then separate average carbon values
were calculated for the 0-10 m and 11 -20 m intervals.
Diatom/non-diatom proportions were taken from the
corresponding 0-20 m phytoplankton biovolume data
and used to divide the total average carbon value
into diatom and non-diatom categories. Estimates of
phytoplankton carbon deeper in the water column
were also calculated from chlorophyll a data. Set
depths of 25 m and 40 m were chosen, and the total
depth between 50 m and the bottom for each station
was split into thirds, and the midpoint of each third
was used for carbon estimation. When total depth
was less than 65 m, a few set depths were used
instead (50 m, 60 m, etc.). Occasionally, an
additional depth was added to allow better
representation of the deep chlorophyll layer.
Phytoplankton carbon at these depths was estimated
147
-------
160
D)
C
o
.a
CO
O
Phytoplankton carbon estimates
r- C predicted from chlorophyll-a data (C:Chl 50 :1)
C predicted from chlorophyll-a data (C:Chl 40 :1)
- * - C predicted from chlorophyll-a data (C:Chl 35 :1)
-•— C calculated from biovolume data
0.35
0.30
6
o c predicted from chlorophyll-a data (C:Chl 40 :1)
• C calculated from biovolume data
0.25
0.20
c
o
.0
l_
CD
O
0.15
9>
4 5
Cruise #
6
Figure 2.4.2. Lake-wide phytoplankton carbon calculated from biovolume data and carbon-to-
chlorophyll a ratios for the eight LMMBP cruises.
148
-------
by again assuming a carbon:chlorophyll ratio of 40:1.
Total carbon was then split into diatom and non-
diatom carbon using station- and cruise-specific 0-20
m diatom proportions. Biovolume data were not
available for many stations and some station/cruise
combinations and cruise average diatom proportions
were used in these instances.
2.4.2.2.4 Zooplankton
Zooplankton were collected with plankton net tows
from 20 m to the surface. At stations that were less
than 20 m in depth, the zooplankton tow was done
from 1 m above the bottom to the surface. Data
were communicated by group densities and
biovolumes (Bythotrephes, carnivores, detrivores,
Dreissena veligers, herbivores, and Mysis) and
species densities and biovolumes. Sampling stations
were distributed throughout the lake.
Zooplankton species level analyses revealed that
several species had high average abundances.
Conochilus unicornis, Polyarthra vulgaris, Polyarthra
major, Dreissena veliger, copepod nauplii, Keratella
cochlearis, Synchaeta, Kellicottia longispina, and
Diaptomus copepodites all occurred at average
densities greater than 5,000/m3. Of these organisms,
only Dreissena, copepod nauplii, and Diaptomus are
not rotifers. Other species were found in virtually all
of the samples and were well distributed throughout
the lake, regardless of season. These included
copepod nauplii, Cyclops copepodites, Diaptomus
copepodites, Diaptomus minutus, Keratella
cochlearis, Synchaeta, Cyclops bicuspidatus,
Diaptomus ashlandi, Kellicottia longispina, and
Polyarthra vulgaris.
Zooplankton were further divided into groups by class
(rotifer, copepod, and cladocera). As suggested by
the species level data, rotifers dominated the overall
zooplankton abundance, although copepods were
more important in early spring and winter (April-June
1994, January-May 1995). Rotifer abundance
peaked in July-August 1994 and July-August 1995
while copepods seemed to peak in August of both
years. Cladocerans experienced a brief but
significant peak in number in mid-August and
September of each year. Total zooplankton
abundance peaked at 400,000 organisms/m3 in 1994
and 700,000 organisms/m3 in 1995. This annual
difference may be the result of differential
reproductive success between years or the timing of
sampling, as discussed for the phytoplankton data.
Copepods overwhelmingly dominated zooplankton
biomass throughout most of the year, with peaks
from August to mid-October 1994 and August to
September 1995. Cladocerans did experience
seasonal peaks, however, in which they accounted
for most of the zooplankton biomass present in the
lake over a very short period of time. These
cladoceran biomass peaks coincided with peaks in
the Daphnia galeata population at the sampling
stations. The peaks occurred in mid-August 1994
and early-August 1995, with a smaller peak in
October 1995. Rotifer biomass was always quite low
despite peaks in abundance and generally high
numbers. Overall, total zooplankton biomass peaked
in the 225,000 to 275,000 mg/m3 range in August
1994 and August 1995.
In order to better understand the impact of
zooplankton on phytoplankton populations, trends of
carnivorous species versus herbivorous species were
examined. Detrivores and Dreissena did account for
10-30% of the total zooplankton abundance during
several sampling months, but no biomass data were
available for these organisms. Carnivores accounted
for less than 25% of the total zooplankton biomass
and abundance during all months sampled.
Carnivore abundance and biomass, in fact, were
relatively static, with only small peaks in each
observed in August of 1994 (around 46,000 mg/m3
and 12,000 organisms/m3) and 1995 (around 48,000
mg/m3 and 4,000 organisms/m3). Herbivore
abundance increased from April 1994 to August 1994
and then began to decline. The same abundance
peak was observed in 1995, but with a slight
resurgence in October of 1995. Herbivore biomass
increased from April 1994 through August 1994 and
slowly declined through December 1994. Herbivore
biomass began to increase again in April 1995 and
followed a similar pattern to that of 1994 over the rest
of 1995. Herbivore abundance peaked at
approximately 210,000 organisms/m3 in 1994 and
275,000 organisms/m3 in 1995, while biomass
peaked at around 225,000 mg/m3 both years.
Differences in carnivore and herbivore abundance
and biomass among sampling stations were analyzed
using a one-way ANOVA. No significant differences
were found among stations for herbivore abundance
149
-------
or biomass, but significant differences did exist
among stations for carnivore abundance and
biomass. Statistical differences in carnivore biomass
arose primarily from a difference between Stations
47M and MB19M. Carnivore abundance differed
between many stations and Station GB24M. This
was not unexpected since GB24M is located in
Green Bay rather than the open lake.
2.4.2.2.5 Zooplankton Carbon
LMMBP zooplankton data were provided as dry
weight biomasses (mg/m3). Data corresponding to
herbivorous species were extracted from the data set
for further analysis. Herbivorous species were
selected because their grazing activities directly
impacted phytoplankton and were, thus, important to
the eutrophication model. Herbivore data were
converted to units of g/L and then converted to
carbon by assuming that carbon accounted for 50%
of the dry weight (Baudouin and Ravera, 1972;
Hessen, 1990; Andersen and Hessen, 1991).
Carbon data were incorporated into the model with
accompanying station and date information. No
zooplankton carbon values were estimated for
segments below 20 m due to the lack of applicable
LMMBP or historical data.
2.4.3 Initial Conditions
The model simulation started in January 1994, but no
field data were available until late April 2004.
Seasonal changes of the state variables were much
larger than changes (increases or decreases) over a
one-year period. We, therefore, based our initial
conditions for the nutrients, carbon, and plankton on
January 1995 (LMMBP Cruise 5) field data. The
carbon estimates were derived from the LMMBP
chlorophyll a data for the 41 segments in the LM2
model. Level 2 segmentation is detailed in Figure
2.4.3. A 40:1 carbon:chlorophyll a ratio was
assumed and used throughout. Diatom/non-diatom
proportions were taken from the 0-20 m
phytoplankton biovolume data wherever possible,
and the same diatom/non-diatom proportions were
maintained throughout the water column. When
insufficient phytoplankton biovolume data existed (as
was the case for many segments in Cruise 5), a
cruise average value (52% diatoms) was used.
When no chlorophyll a profiles were available for a
given segment for Cruise 5, values from neighboring
segments were used. In general, if no values were
available for segments 4, 5, and 6, the average of
total phytoplankton carbon for segments 1, 2, and 3
were used (diatom/non-diatom carbon proportions
were assigned later based on segment specifics). If
values were available for segment 6 but 4 and 5 were
missing, segment 6 values were assigned to
segment 5 and segment 3 values were assigned to
segment 4. These estimated surface segment
values were then mirrored throughout the water
column.
Inadequate Green Bay data existed to follow the
previously described approach for assigning initial
conditions. After review of the LMMBP chlorophyll
a profiles available for Green Bay stations, January
chlorophyll a was estimated to be 1 ug/L for
segments 7 and 8, 2 ug/L for segment 9, and 3 pg/L
for segment 10. Using Green Bay specific
diatom/non-diatom proportions estimated from the
literature and a 40:1 carbon-to-chlorophyll ratio,
these values were converted to diatom and non-
diatom carbon initial conditions (Sager and Richman,
1991; DeStasio and Richman, 1998). Values for
deeper segments mirrored the surface values.
Carbon data for zooplankton collected from waters 0-
20 m depth in January 1995 (Cruise 5), March 1995,
and April 1994 and 1995 (Cruises 1 and 6) were
examined in order to estimate initial conditions for the
lake. No zooplankton samples existed for several
surface segments within the lake and most estimates
for these segments follow from estimated values of
neighboring segments. Carbon values varied with
segment, but, generally, the same value was
assigned to the 0-10 m, 10-20 m, and 20-30 m
segments within each surface segment sector, and a
value of 150% of this 0-10 m carbon value was
assigned to the 30-50 m depth segment. The bottom
segment (50 m maximum depth) was assigned a
carbon value equal to the 0-10 m carbon value.
Many of the non-biological field measured variables
did not directly relate to the state variables used in
the model. As a result, assumptions were made and
calculations were performed to determine the
appropriate initial conditions for the modeled state
variables. Table 2.4.6 lists the field measurements
and modeled state variables for the nutrients and
carbon. Specific assumptions and calculations used
in estimating these model state variables from field
150
-------
3
13
22
31
38
10
10
10
20
212
1
11
20
29
36
10
10
10
20
103
2
12
21
30
37
10
10
10
20
113
5
15
24
33
40
10
10
10
20
44
water column
segment
numbers
11
20
29
36
10 average
10 depth of
10 segment
20
103
Figure 2.4.3. Level 2 model segmentation for LM3-Eutro.
151
-------
Table 2.4.6. Relationship of Field Measurements
and Model State Variables
Variable
Field
Measurements
Model State
Variable
Phosphorus
Nitrogen
Silica
Carbon
Total Phosphorus
Dissolved
Phosphorus
Soluble Reactive
(SRP)
Total Kjeldahl
(TKN)
Ammonium (NH4)
Nitrate (NO3)
Dissolved (DSi)
Particulate Organic
(POC)
Dissolved Organic
(DOC)
Labile Organic (LOP)
Refractory Organic
(ROP)
Soluble Reactive
(SRP)
Dissolved Organic
(OOP)
Labile Organic (LON)
Refractory (RON)
Dissolved Organic
(DON)
Ammonium (NH4)
Nitrate (NO3)
Dissolved (DSi)
Biogenic (BSi)
Labile Organic (LOG)
Refractory Organic
(ROC)
Dissolved Organic
(DOC)
measurements can be found in Appendix 2.4.1. It
was assumed that the particulate forms for carbon,
phosphorus, and nitrogen were split evenly between
the labile and refractory forms. It was also assumed
that the DON was insignificant.
2.4.4 Parameter Estimation
One of the most challenging tasks in the model
development process was the estimation of the
different model coefficients. A limitation of this
project was the lack of field and laboratory
experiments to determine values for the many
coefficients. Some physical data were available for
model coefficient estimation, and these instances are
detailed below. In addition, the use of primary
productivity experiments to assist with the estimation
of production-related coefficients will be discussed.
Values for all other parameters were obtained
initially from the literature, with further refinement via
calibration.
2.4.4.1 Physical Measurements
2.4.4.1.1 Secchi Disk
Secchi disk measurements were performed during
the eight sampling cruises in 1994-1995 to obtain an
estimate of water clarity. Cruise averages for all
available stations were calculated and are shown in
Figure 2.4.4. Secchi disk values were used in an
empirical equation (Thomann and Mueller, 1987) to
estimate the light extinction coefficients used in the
eutrophication model.
2.4.4.1.2 Solar Radiation and Temperature
Primary productivity was strongly affected by both
available light (solar radiation) and temperature. As
part of the output of the hydrodynamics model
(Princeton Ocean Model [POM]) used to generate
Lake Michigan hydrodynamic parameters, lake-wide
short wave solar radiation and temperature data were
generated (Schwab and Beletsky, 1998). Solar
radiation was one of the forcing functions driving the
phytoplankton growth. In the model, it was referred
to as incident solar light intensity (I0).
2.4.4.2 Primary Production Estimates
The rates at which phytoplankton grow and utilize
available nutrients are among the most important and
complex processes in any eutrophication model.
Primary productivity laboratory experiments were
conducted as part of the LMMBP. However, due to
the difficulty in converting laboratory production rates
into reasonable in situ primary production
information, the model production rates were
generated using coefficients gleaned from published
literature and the model calibration process (Table
2.4.7). The laboratory primary production
experiments were used to verify the overall
production rates in the model (Figure 2.5.2).
Laboratory productivity data were provided by
GLNPO for the 1994-1995 project field season. The
14C incubation productivity determination method was
utilized. This method calls for the inoculation of
water sub-samples with 13C radiotracer followed by
incubation at varying light intensities for two to four
hours. Sub-samples were filtered and radioactivity
152
-------
0
1-
2-
3-
Average Lake Michigan Secchi Disk Depth 1994-1995
Error bars represent 95% confidence intervals
Figure 2.4.4. Lake-wide Secchi depths for the eight LMMBP cruises.
Table 2.4.7. Important LM3 Model Coefficients
Parameter
CCHLD
CCHLG
KHPD
KHPG
KHSD
PMD
PMG
TMD
TMG
Unit
No Unit
No Unit
M9/L
ug/L
mg/L
1/day
1/day
Value
40
40
0.5
0.5
0.03
2.5
2.1
20
20
Literature
Values
10-1002'3'4
10-1001-4
0.5-1.02'3'4
0.5-1.01'4
0.03 - 0.062'3'4
0.58 - 8.02'3'4
0.58 - 8.01 4
2Q2.3.4
2Q1-4
Description
Carbon:chlorophy!l ratio (diatoms)
Carbon :chlorophyll ratio (non-diatoms)
Phosphorus half-saturation coefficients for diatoms
Phosphorus half-saturation coefficients for non-
diatoms
Si half-saturation coefficient for diatoms
Diatom growth coefficient
Non-diatom growth coefficient
Optimum diatom growth temperature
Optimum non-diatom growth temperature
1Rodgers and Salisbury, 1981
2Di Toro and Connolly, 1980
3Bowieefa/., 1985
"Thomann and Di Toro, 1975
153
-------
of algal cells was measured (U.S. Environmental
Protection Agency, 1997). Measured radioactivity
should be proportional to the amount of carbon fixed
by the algae. Other variables used in calculating the
final productivity estimate include light intensity,
length of incubation, temperature, and basic
information about carbon and chlorophyll levels in the
water samples. Variables reported included station
code, date, sample depth, temperature, sample
identification number, productivity results (mg C/L/h),
total incubation time and incubation light level
(mE/m2/s). Each station was sampled several times
from April 1994 to October 1995, and 12 sub-
samples were incubated (at different light intensities)
for each station/date/depth combination. Discrete
and integrated samples were collected, and efforts
were made to include hypolimnetic samples during
stratification.
Most of the analysis effort was devoted to determine
how productivity changes with light, temperature,
phytoplankton carbon, chlorophyll a, etc., and to
compare these changes with the output of the model
equation. Data appeared to follow typical irradiance
versus production curves, with production increasing
with increasing light levels and then reaching a
plateau. Limited light ranges, however, prevented
determination of the presence/absence or degree of
light inhibition. For purposes of further analysis of
laboratory versus model productivity predictions,
optimum light levels were designated. For each set of
experiments, optimum light was taken to be that light
at which maximum production (mgC/L/h) was
reported.
There was some degree of uncertainty associated
with all estimates of phytoplankton production
derived from incubation experiments. It is well-
known that results from short experiments (< 6
hours) are frequently higher than those estimated
from longer experiments (24 hours). It is generally
believed that short-term 14C incubations measure
something between gross and net production
(Fahnenstiel and Scavia, 1987). This is a factor
which must be considered when comparing
laboratory data to predictions from model equations.
References
Allan, J.D. 1995. Stream Ecology: Structure and
Function of Running Waters. Chapman and Hall,
London, England. 104pp.
Andersen, T. and D.O. Hessen. 1991. Carbon,
Nitrogen, and Phosphorus Content of Freshwater
Zooplankton. Limnol. Oceanogr., 36(4):807-814.
Badouin, M.F. and O. Ravera. 1972. Weight, Size
and Chemical Composition of Some Freshwater
Zooplankters: Daphnia hyalina (Leydig). Limnol.
Oceanogr., 17(4):645-649.
Bierman, V.J., Jr., J.V. DePinto, T.C. Young, P.W.
Rodgers, S.C. Martin, and R. Raghunathan.
1992. Development and Validation of an
Integrated Exposure Model for Toxic Chemicals
in Green Bay, Lake Michigan. Final Report. U.S.
Environmental Protection Agency, Office of
Research and Development, ERL-Duluth, Large
Lakes Research Station, Grosse He, Michigan.
381 pp.
Bowie, G.L, W.B. Mills, D.B. Porcella, C.L
Campbell, J.R. Pagenkopf, G.L. Rupp, K.M.
Johnson, P.W.H. Chan, S.A. Gherini, and C.E.
Chamberlin. 1985. Rates, Constants and Kinetic
Formulations in Surface Water Quality Modeling,
2nd Edition. U.S. Environmental Protection
Agency, Environmental Research Laboratory,
Athens, Georgia. EPA/600/3-85/040, 455 pp.
Cloern, J.E., C. Grenz, and L. Vidergar-Lucas. 1995.
An Empirical Model of the Phytoplankton
ChlorophylkCarbon Ratio - The Conversion
Factor Between Productivity and Growth Rate.
Limnol. Oceanogr., 40(7):1313-1321.
Conley, D.J., M.A. Quigley, and C.L. Schelske.
1988. Silica and Phosphorus Flux From
Sediments: Importance of Internal Recycling in
Lake Michigan. Canadian J. Fish. Aquat. Sci.,
45(6): 1030-1035.
154
-------
DeStasio, B.T., Jr. and S. Richman. 1998.
Phytoplankton Spatial and Temporal Distributions
in Green Bay, Michigan, Prior to Colonization by
the Zebra Mussel (Dreissena polymorpha). J.
Great Lakes Res., 24(3):620-628.
Fahnenstiel, G.L andD. Scavia. 1987. Dynamics of
Lake Michigan Phytoplankton: Primary
Production and Growth. Canadian J. Fish. Aquat.
Sci., 44(3):499-508.
Goldsmith, J.C. 1999. Calibration of In Vivo
Fluorometer Response Measurements With
Known Amounts of Extracted Chlorophyll a.
Internal report and presentation. U.S.
Environmental Protection Agency, Great Lakes
National Program Office, Chicago, Illinois. April
29,1999.
Hall, D. and D. Robertson. 1998. Estimation of
Contaminant Loading from Monitored and
Unmonitored Tributaries to Lake Michigan for the
USEPA Lake Michigan Mass Balance Study.
Quality Systems and Implementation Plan.
Submitted October 23,1998. U.S. Environmental
Protection Agency, Great Lakes National
Program Office, Chicago, Illinois. 19 pp.
Hessen, D.O. 1990. Carbon, Nitrogen and
Phosphorus Status in Daphnia at Varying Food
Conditions. J. Plankton Res., 12(6): 1239-1249.
Laird, G.A., D. Scavia, G.L. Fahnenstiel, LA. Strong,
and G.A. Lang. 1988. Dynamics of Lake
Michigan Phytoplankton: Relationship to Nitrogen
and Silica Fluxes. Canadian J. Fish. Aquat. Sci.,
45(8): 1459-1466.
Miller, S.M., C.W. Sweet, J.V. DePinto, and K.C.
Hornbuckle. 2000. Atrazine and Nutrients in
Precipitation: Results from the Lake Michigan
Mass Balance Study. Environ. Sci. Technol.,
34(1):55-61.
Montagnes, D.J.S., J.A. Berges, P.J. Harrison, and
F.J.R. Taylor. 1994. Estimating Carbon,
Nitrogen, Protein, and Chlorophyll a From
Volume in Marine Phytoplankton. Limnol.
Oceanogr., 39(5): 1044-1060.
Monteith.T.J.andW.C.Sonzogni. 1976. U.S. Great
Lakes Shoreline Erosion Loadings. Great Lakes
Basin Commission, Ann Arbor, Michigan. 223
pp.
Quigley, M.A. and J.A. Robbins. 1986. Phosphorus
Release Processes in Nearshore Southern Lake
Michigan. Canadian J. Fish. Aquat. Sci.,
43(6): 1201-1207.
Richardson, W.L., D.D. Endicott, R.G. Kreis, Jr., and
K.R. Rygwelski (Eds.). 2004. The Lake Michigan
Mass Balance Project Quality Assurance Plan for
Mathematical Modeling. Prepared by the
Modeling Workgroup. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
EPA/600/R-04/018, 233 pp.
Riemann, B., P. Simonsen, and L. Stensgaard.
1989. The Carbon and Chlorophyll Content of
Phytoplankton From Various Nutrient Regimes.
J. Plankton Res., 11(5):1037-1045.
Rocha, O. and A. Duncan. 1985. The Relationship
Between Cell Carbon and Cell Volume in
Freshwater Algal Species Used in Zooplanktonic
Studies. J. Plankton Res., 7(2):279-294.
Rodgers, P.W. and D. Salisbury. 1981. Modeling of
" Water Quality in Lake Michigan and the Effect of
the Anomalous Ice Cover of 1976-1977. Great
Lakes Environmental Planning Study, Great
Lakes Basin Commission, Ann Arbor, Michigan.
Contribution Number 44, 53 pp.
Sager, P.E. and S. Richman. 1991. Functional
Interactions of Phytoplankton and Zooplankton
Along the Trophic Gradient in Green Bay, Lake
Michigan. Canadian J. Fish. Aquat. Sci.,
48(1):116-122.
Schwab, D.J. and D. Beletsky. 1998. Lake Michigan
Mass Balance Study: Hydrodynamic Modeling
Project. National Oceanic and Atmospheric
Administration, Great Lakes Environmental
Research Laboratory, Ann Arbor, Michigan.
NOAA Technical Memorandum ERL GLERL-
108, 53pp.
155
-------
Sicko-Goad, L.M., C.L. Schelske, and E.F. Stoermer.
1984. Estimation of Intracellular Carbon and
Silica Content of Diatoms From Natural
Assemblages Using Morphometric Techniques.
Limnol. Oceanogr., 29(6):1170-1178.
Strathmann, R.R. 1967. Estimating the Organic
Carbon Content of Phytoplankton From Cell
Volume or Plasma Volume. Limnol. Oceanogr.,
12:411-418.
Tarapchak, S.J. and C. Nalewajko. 1987. A Review:
Phosphorus-Plankton Dynamics and Phosphorus
Cycling in Aquatic Systems. National Oceanic
and Atmospheric Administration, Great Lakes
Environmental Research Laboratory, Ann Arbor,
Michigan. NOAA Technical Memorandum ERL
GLERL-60, 57 pp.
Thomann, R.V., D.M. Di Toro, R.P. Winfield, and D.J.
O'Connor. 1975. Mathematical Modeling of
Phytoplankton in Lake Ontario, Part 1 - Model
Development and Verification. U.S.
Environmental Protection Agency, Office of
Research and Development, ERL-Corvallis,
Large Lakes Research Station, Grosse lie,
Michigan. EPA/660/3-75/005, 177 pp.
Thomann, R.V. and J.A.Mueller. 1987. Principles of
Water Quality Modeling and Control. Harper and
Row Publishers, New York, New York. 644 pp.
U.S. Environmental Protection Agency. 1997. Lake
Michigan Mass Balance Study (LMMB) Methods
Compendium, Volume 1: Sample Collection
Techniques. U.S. Environmental Protection
Agency, Great Lakes National Program Office,
Chicago, Illinois. EPA/905/R-97/012a, 1,440pp.
156
-------
PART 2
LM3-EUTRO
Appendix 2.4.1. Modeled Versus
Measured Variables
Not all model output variables could be directly
compared to the field measurements. In order to
compare these model variables to the field data, the
following calculations were performed:
APCP = phosphorus to carbon ratio (a constant
value of 0.01)
(A2.4.1.1)
Total Phosphorus
TP = LOP + POP + OOP + SRP
+ APCP (DIA + GFtE + ZOO)
where
Field measurements (expressed as mass/volume)
TP = total phosphorus
SRP = soluble reactive phosphorus (also a
model output variable)
Model output variables (expressed as mass/volume)
LOP = labile particulate organic phosphorus
ROP = refractory particulate organic phosphorus
OOP = dissolved organic phosphorus
DIA = diatoms (expressed as carbon)
GRE = non-diatoms (expressed as carbon)
ZOO = zooplankton (expressed as carbon)
Dissolved Phosphorus
DP = OOP + SRP
where
(A2.4.1.2)
DP = dissolved phosphorus (expressed as
mass/volume)(field measurement)
Total Kjeldahl Nitrogen
TKN = LON + RON + DON + NH3
+ ANCP(DIA + GRE + ZOO) (A2.4.1.3)
where
Field measurements (expressed as mass/volume)
TKN = total Kjeldahl nitrogen
NH3 = ammonia (also a model output variable)
Model output variables (expressed as mass/volume)
LON = labile particulate organic nitrogen
RON = refractory particulate organic nitrogen
DON = dissolved organic nitrogen
ANCP = nitrogen-to-carbon ratio (a constant value
of 0.25)
157
-------
Particulate Organic Carbon Total Algal Carbon
POC = LOG + ROC + DIA + GRE + ZOO Total a/0a/ carbon = DIA + GRE (A2.4.1.5)
(A2.4.1.4)
Total algal carbon was neither a field nor a model
where variable. It was a summed total of the diatoms and
greens (performed for both the field and model
Field measurements (expressed in mass/volume) output).
POC = particulate organic carbon
Model output variables (expressed as mass/volume)
LOC = labile particulate organic carbon
ROC = refractory particulate organic carbon
158
-------
PART 2
LM3-EUTRO
Chapters. Calibration
2.5.? Description of Process
After model equations were formulated and coded as
a computer program, model calibration was the next
step. The goal of calibrating water quality models
was to adjust the model coefficients in order to obtain
the best possible fit between the model output and
the field data. Challenges of calibrating
eutrophication models included the many degrees of
freedom (independent model coefficients) and the
uncertainty of many of these model coefficients. A
traditional model calibration approach was used for
LM3-Eutro. The model coefficients were initially
estimated using values and ranges reported in the
literature (see Table 2.4.7) and these parameters
were then adjusted to provide the best model fit to
the field data. In this study, values for many
coefficients were derived from available Lake
Michigan and Great Lakes historical data collected by
reputable agencies such as the United States
Environmental Protection Agency's (USEPA) Great
Lakes National Program Office (GLNPO), National
Oceanic and Atmospheric Administration's (NOAA)
Great Lakes Environmental Research Laboratory
(GLERL) and the University of Michigan. Very few
field and laboratory experiments were performed to
estimate kinetic coefficients for LM3-Eutro. Limited
14C primary production experiments were performed
and used in determining phytoplankton growth
coefficients. Phytoplankton (diatoms and non-
diatoms), particulate organic carbon (POC), total
phosphorus, and dissolved silica (DSi) were the
most important state variables in model
calibration. However, all variables were evaluated
during the calibration process.
The model was calibrated on the Level 3
segmentation framework (Figure 2.5.1). The main
calibration emphasis was placed on the main lake
due to inadequate Green Bay data. The high-
resolution (Level 3) segments were also collapsed to
the Level 2 segmentation scheme to provide a visual
representation of how well the model reflected the
field data in different areas of the lake and captured
expected trends, such as spring epilimnetic diatom
peaks and nutrient depletion. It was not
LEVEL2-LM-2
10 Surface Segments
41 Water Segments
Figure 2.5.1.
segmentation.
LEVEL 3 - LM-3
(High Resolution 5km X 5km Grid)
2318 Surface Segments
44,042 Water Segments
19 "Sigma" Levels
Level 2 and Level 3 model
159
-------
feasible to visually compare model output versus field
data on the Level 3 segmentation framework due to
the large number (44,042) of 5 km2 cells. Instead,
we regressed model output versus field data for each
of the 5 km2 cells where a field data point was
available. This enabled calculation of simple
statistical parameters such as square of the
correlation coefficient (r2) and slope and direct
comparison of the different model calibration runs.
During the calibration process, small changes to the
model initial conditions were made to observe the
effect on the model fit. This was done because of the
uncertainty of the initial condition estimates and the
fact that the initial conditions have a significant
influence on the model output.
Data from laboratory primary productivity
experiments were used to constrain and confirm
values for the growth coefficients that were used in
the model. Productivity experiment results; light and
temperature parameters; and the Lake Michigan
Mass Balance Project (LMMBP) field data were
applied to LM3-Eutro productivity equations to
generate model production estimates comparable to
those generated in the laboratory experiments.
Model constants were then adjusted in order to best
reflect the primary production trends observed in the
laboratory experiments (Figure 2.5.2).
2.5.2 Selection of Best Calibration
After performing several hundred model simulations,
we selected our best run based on statistical
parameters including square of correlation coefficient
(r2) and slope. The coefficients of the final run were
constrained to ensure that all model coefficients fell
within reasonable and reported ranges. The best
model fit was also evaluated visually on the Level 2
segmentation scheme. Important criteria included
the model fit with the overall field data, ability to
capture observed and expected phytoplankton
peaks, and how well hypolimnetic and epilimnetic
nutrient, carbon, and plankton trends and
concentrations were predicted. The calibrated final
model coefficients are listed in Table 2.5.1. Figure
2.5.3 shows the overall Level 3 fit for phytoplankton,
total phosphorus, POC, and DSi. Statistical results
are summarized in Table 2.5.2. Figure 2.5.4 shows
model versus field data plots for selected Level 2
segments. Selected Level 3 5 km2 cells representing
nearshore regions and offshore regions are
presented in Figure 2.5.5. Model output for the 5 km2
cells was much more dynamic than for the larger
Level 2 segments and there were far fewer data
points for model fit determination. The model
appeared to fit the available data very well in some of
the cells and not as well in others. Level 3 model
versus data comparisons in individual 5 km2 cells
were not used in our calibration exercise.
Although we attempted to calibrate all of the state
variables, less emphasis was placed on the nitrogen
states because Lake Michigan is phosphorus and
silica-limited. We also did not perform any
comparison of model output with the field data for
particulate silica (SU) or soluble reactive phosphorus
(SRP), since there was no SU field data and more
than 80% of the SRP field data fell below the
detection limit. As stated, we did not spend much
time calibrating Green Bay because of the limited
sampling done in the bay. As a consequence, the
final Green Bay calibration was not as good as the
rest of the lake. This was especially true for the
portion of the bay closest to the Fox River.
With the exception of zooplankton, the final model
calibration was reasonably good, and the model was
able to fit the field data well and capture important
spatial and seasonal trends. A brief discussion of
individual calibration results for phytoplankton, POC,
total phosphorus, and DSi follows.
2.5.2.1 Phytoplankton
The model somewhat underestimated the field data
for phytoplankton. Seabird chlorophyll a data were
used for all phytoplankton field values. Part of the
explanation for the underestimation was poor
chlorophyll a field data. The Seabird fluorescence
instrument, like many in vivo fluorescence methods,
is notorious for its inaccuracy in measuring
chlorophyll a (U.S. Environmental Protection Agency,
1997; Clesceri et a/., 1998). The square of the
correlation coefficient of 0.37 was acceptable,
especially given the inherent variation in
phytoplankton communities over space and time.
Our fit was in-range of other published eutrophication
models (Thomann, 1982; Cerco and Cole, 1994).
The model was able to capture spatial and temporal
trends such as the spring diatom blooms (Figure
2.5.4) and earlier phytoplankton blooms in the
160
-------
0.016'
0.014
0.012
O 0.01
D)
£ 0.008
t>
£ 0.006
Q
Q.
0.004
Phytoplankton carbon versus productivity (discrete samples)
D
D ^- ^
» ^ ^
* * D ^_ -•- "
^ — n
* _^^ LJ
* * -- * ""
V\i-«^^oD*
— ^F . — n ^ . _
^
a Chlorophyll-a based model
productivity results (mgC/L/hr)
4 Lab results (mgC/L/hr)
*
0.002
0.05
0.1 0.15 0.2
phytoplankton carbon (mg/L)
0.25
0.3
0.025
0.02
Seasonal productivity in Lake Michigan
n Model results
4 Lab results
1 2 3
5 6
LMMB cruises
Figure 2.5.2. LM3-Eutro model versus laboratory primary production.
161
-------
Table 2.5.1. Coefficients Used in the LM3 Model (Units Correspond to Required LM3 Model Output)
Coefficient
ANCP
APCP
ASCD
BMRD
BMRG
CCHLD
CCHLG
CGZ
FCDD
FCDG
FCLD
FCLG
FCRD
FCRG
FCDP
FCDZ
FCLP
FCLZ
FCRP
FCRZ
FNDD
FNDG
FNDP
FNDZ
FNID
FNIG
FNIP
FNIZ
FNLD
FNLG
FNLP
FNLZ
FNRD
FNRG
RNRP
RNRZ
FPDD
FPDG
FPDP
FPDZ
FPID
FPIG
FPIP
FPLD
FPLG
FPLP
FPLZ
FPRD
FPRG
FPRP
FPRZ
Value
0.25
0.01
2.3
8.6E-07
8.6E-07
40
40.
3.1E-06
0.05
0.05
0.3
0.3
0.3
0.3
0.35
0
0.5
0.4
0.15
0.1
0.5
0.5
0
0
0.5
0.5
0.5
0.5
0
0
0.4
0.4
0
0
0.1
0.1
0.1
0.1
0.2
0.2
0.3
0.3
0.5
0.5
0.3
0.15
0.15
0.3
0.3
0.15
0.15
Unit Description
Nitrogen:carbon ratio (mass basis)
Phosphorus:carbon ratio (mass basis)
Silica:carbon ratio (mass basis)
1/s Diatom mortality
1/s Greens mortality
Carboirchlorophyll ratio (diatoms)
Carboirchlorophyll ratio (greens)
m3/kg/s Zooplankton grazing rate coefficient
Dissolved organic carbon fraction from diatom mortality
Dissolved organic carbon fraction from greens mortality
Labile organic carbon fraction from diatom mortality
Labile organic carbon fraction from greens mortality
Refractory organic carbon fraction from diatom mortality
Refractory organic carbon fraction from greens mortality
Dissolved organic carbon fraction from algal predation
Dissolved organic carbon fraction from zooplankton mortality
Labile particulate dissolved carbon fraction from algal predation
Labile particulate dissolved carbon fraction from zooplankton mortality
Refractory particulate dissolved carbon fraction from algal predation
Refractory particulate dissolved carbon from zooplankton mortality
Dissolved organic nitrogen from diatom mortality
Dissolved organic nitrogen fraction from greens mortality
Dissolved organic nitrogen fraction from algal predation
Dissolved organic nitrogen fraction from zooplankton mortality
Dissolved inorganic nitrogen fraction from diatom mortality
Dissolved inorganic nitrogen fraction from greens mortality
Dissolved inorganic nitrogen fraction from algal predation
Dissolved inorganic nitrogen fraction from zooplankton mortality
Labile organic nitrogen fraction from diatom mortality
Labile organic nitrogen fraction from greens mortality
Labile organic nitrogen fraction from algal predation
Labile organic nitrogen fraction from zooplankton mortality
Refractory organic nitrogen fraction from diatom mortality
Refractory organic nitrogen fraction from greens mortality
Refractory organic nitrogen fraction from algal predation
Refractory organic nitrogen fraction from zooplankton mortality
Dissolved organic phosphorus fraction from diatom mortality
Dissolved organic phosphorus fraction from greens mortality
Dissolved organic phosphorus fraction from algal predation
Dissolved organic phosphorus fraction from zooplankton mortality
Dissolved inorganic phosphorus fraction from diatom mortality
Dissolved inorganic phosphorus fraction from greens mortality
Dissolved inorganic phosphorus fraction from algal predation
Dissolved inorganic phosphorus fraction from zooplankton mortality
Labile organic phosphorus fraction from greens mortality
Labile organic phosphorus fraction from algal predation
Labile organic phosphorus fraction from zooplankton mortality
Refractory organic phosphorus fraction from diatom mortality
Refractory organic phosphorus fraction from greens mortality
Refractory organic phosphorus fraction from algal predation
Refractory organic phosphorus fraction from zooplankton mortality
162
-------
Table 2.5.1. Coefficients Used in the LM3 Model (Continued)
Coefficient
FSAP
GREFF
ILUMO
ISMIN
KDC
KDCALG
KDN
KDNALG
KDP
KDPALG
KE
KECHL
KHND
KHNG
KHNNT
KHPD
KHPG
KHSD
KLC
KLCALG
KLN
KLNALG
KLP
KLPALG
KRC
KRCALG
KRN
KRNALG
KRP
KRPALG
KSUA
KSZ
KTBD
KTBG
KTGD1
KTGD2
KTGG1
KTGG2
KTHDR
KTMNL
KTNT1
NTNT2
KTSUA
NTM
NTM
PMD
PMG
TMD
TMG
TMNT
TRD
Value
0
0.6
25
400
1.16E-08
O.OOE+00
1 .74E-07
O.OOE+00
1.16E-09
6.0E-03
0.15
1.7E+04
2.50E-05
2.50E-05
0.0000
5.0E-07
5.0E-07
6.0E-05
1.0E-07
O.OOE+00
3.47E-07
O.OOE+00
1 .OOE-09
6.00E-03
1.00E--07
O.OOE+00
3.47E-08
OOE+00
1 .OE-09
6.0E+03
2.5E-07
1.0E-04
0.074
0.074
0.0025
0.006
0.0025
0.006
9.9E-02
7.4E-02
0.004
0.004
0.069
0.074
2.50E-11
2.90E-05
2.60E-05
18
18
30
20
Units
W/m2
W/m2
1/s
m3/kg/s
1/s
m3/kg/s
1/s
m3/kg/s
1/m
m2/kg
kg/m3
kg/m3
kg/m3
kg/m3
kg/m3
kg/m3
1/s
m3/kg/s
1/s
m3/kg/s
1/s
m3/kg/s
1/s
m3/kg/s
1/s
m3/kg/s
1/s
m3/kg/s
1/s
kg/m3
1/°C
1/°C
1/°C2
1/°C2
1/°C2
1/°C2
1/°C
1/°C
1/°C2
1/°C2
1/°C
1/°C
kg/m3/s
1/s
1/s
°C
°C
°C
°C
Description
Dissolved silica fraction from diatom predation
Zooplankton grazing coefficient
Constant illumination (first 90 days)
Optimum light illumination
Dissolved organic carbon mineralization coefficient
Dissolved organic carbon algal dependency coefficient
Dissolved organic nitrogen mineralization coefficient
Dissolved organic nitrogen algal dependency coefficient
Dissolved organic phosphorus mineralization coefficient
Dissolved organic phosphorus algal dependency coefficient
Background light attenuation
Light attenuation for chlorophyll a
Nitrogen half -saturation coefficients for diatoms
Nitrogen half-saturation coefficients for greens
Nitrate half-saturation coefficient for nitrification
Phosphorus half-saturation coefficients for diatoms
Phosphorus half-saturation coefficients for greens
Silica half-saturation coefficients for diatoms
Labile organic carbon hydrolysis coefficient
Labile organic carbon algal dependency coefficient
Labile organic carbon hydrolysis coefficient
Labile organic nitrogen algal dependency coefficient
Labile organic phosphorus hydrolysis coefficient
Labile organic phosphorus algal dependency coefficient
Refractory organic carbon hydrolysis coefficient
Refractory organic carbon algal dependency coefficient
Refractory organic nitrogen hydrolysis coefficient
Refractory organic nitrogen algal dependency coefficient
Refractory organic phosphorus hydrolysis coefficient
Refractory organic phosphorus algal dependency coefficient
Biogenic silica dissolution rate
Zooplankton half-saturation (for algae)
Diatom mortality temperature coefficient
Greens mortality temperature coefficient
Diatom growth temperature coefficient (< optimum)
Diatom growth temperature coefficient (> optimum)
Greens growth temperature coefficient (< optimum)
Greens growth temperature coefficient (> optimum)
Hydrolysis temperature dependency coefficient
Mineralization temperature dependency coefficient
Nitrification temperature coefficient (< optimum)
Nitrification temperature coefficient (> optimum)
Silica dissolution temperature coefficient
Diatom mortality temperature coefficient
Nitrification rate coefficient
Diatom growth coefficient
Greens growth coefficient
Optimum diatom growth temperature
Optimum greens growth temperature
Optimum nitrification temperature
Optimum diatom mortality temperature
163
-------
Table 2.5.1. Coefficients Used in the LM3 Model (Continued)
Coefficient
TRG
TRHDR
TRMNL
TRSUA
TZREF
ZDTH
ZTHET
VDIA
VGRE
VLOC
VROC
VLON
VRON
VLOP
VROP
VSU
Value
20
20
20
20
20
5.0E-07
1.0
1.15E-06
0.85E-06
2.0E-06
2.0E-06
2.0E-06
2.0E-06
2.0E-06
2.0E-06
2.0E-06
Units
°C
°C
°c
°c
°c
1/s
m/s
m/s
m/s
m/s
m/s
m/s
m/s
m/s
m/s
Description
Optimum greens mortality temperature
Optimum hydrolysis temperature
Optimum mineralization temperature
Optimum silica dissolution temperature
Optimum predation temperature
Zooplankton mortality rate coefficient
Arrhenius temperature coefficient for predation
Diatoms settling coefficient
Greens settling coefficient
Labile organic carbon settling coefficient
Refractory organic carbon settling coefficient
Labile organic nitrogen settling coefficient
Refractory organic nitrogen settling coefficient
Labile organic phosphorus settling coefficient
Refractory organic phosphorus settling coefficient
Biogenic silica settling coefficient
S
CO
P3 •«
0.25-
0.2-
0.15.
0.1.
0.05.
x
Phytoplankton (mg/L) x
X'
X
X
s'
X
X
X
^ x
' • • x
• • " "A
'?.:•;-',:? ^il
**• > -eT/" * ^* * *
•i&KpsV? " -''J •
'*»!?« ' " «
JJF?f-'*~
30.
25.
S20.
-D
T3
•5515-
i^=
10-
5-
0-
Total Phosphorus (M9/L) '
s'
S
/
/•
s
• , /
• /
; \; x
• .. X
x.
r: x
i • x
T " " •/
mZ' '
jgp
/
0.05 0.1 0.15 0.2 0.25 0.3
model results
0 5 10 15 20 25 30 35
model results
1.2
1.0
0.8-
0.6-
0.4-
0.2-
o.
Particulate Organic Carbon (mg/L) x '
x
x
x
x
. x
X
X
' x
X
• X
l X
'x
' /. X
•$$&?- •
**3lSi$ * ' "
/'aSS'**" * •* •
x "*"• '
2
oj1-5-
"ra
•a
m
•S 1-
0.5-
0-
Dissolved Silica (mg/L) x
x
x
x
x
x
; " x'
xx
.• x"
f:33*/
,
0.2 0.4 0.6 0.8 1.0 1.2
model results
0 0.5 1 1.5
model results
2.5
Figure 2.5.3. Level 3 LM3-Eutro model predictions versus field data, lake-wide.
164
-------
Table 2.5.2. Summary of Statistical Results of the Calibration
Variable Regression Coefficient (r2) Slope
Phytoplankton 0.37 0.67
Particulate Organic Carbon 0.39 0.95
Total Phosphorus 0.37 1 .4
Dissolved Silica 0.37 1 .2
Zooplankton 0.13 0.43
0.2
O.15
0.1
O.O5
O
J
1S
O.6
O.5
O.4
0.3
0.2
0.1
°j;
13
2O
15
10
5
°J:
19
O.2
O.15
O.1
O.O5
O
Phytoplankton (mg/L)
Segment 6
/v/v
an July Jan July Dec
94 1994 1995 19951995
Particulate
Organic Carbon (mg/L)
Segment 1
,/tv - yv
i
in July Jan July Dec
®4 1994 1995 19951995
Total Phosphorus (pg/L)
Segment 3
- • . >
} ' " i *
an July Jan July Dec
94 1994 1995 19951995
Dissolved Silica (mg/L)
Segment 2
^V! /'J~~V/" '
1^ ••*
O.2
O.15
O.1
O.O5
O
J<
19
O.6
O.5
O.4
0.3
O.2
0.1
°l:
19
2O
15
1O
5
°J<
19
O.2
0.15
O.1
O.O5
O
Phytoplankton (mg/L)
Segment 13
.
J\7V
an July Jan July Dec
94 1994 1995 19951995
Particulate
• Organic Carbon (mg/L) -
Segment 1 1
• :/sv /V n
1
in July Jan July Dec
94 1994 1995 19951995
Total Phosphorus (pg/L)
Segment 11
: . '
•TT - -T«-
sn July Jan July Dec
94 1994 1995 19951995
Dissolved Silica (mg/L)
Segment 11
^V -X^VJ^
-;
O.2
0.15
O.1
O.O5
O
j£
19
0.6
O.5
O.4
0.3
O.2
O.1
19
20
15
1O
5
9,
19
O.2
O.15
O.1
O.O5
Q
Phytoplankton (mg/L)
Segment 24
A A
AjV
in July Jan July Dec
94 1994 1995 19951995
Particulate
• Organic Carbon (mg/L)
Segment 21
/~^~\
,/rvy-.
n July Jan July Dec
94 1994 1995 19951995
Total Phosphorus (pg/L)
Segment 31
- 1 . i ' --.•
in July Jan July Dec
94 1994 1995 19951995
Dissolved Silica (mg/L)
Segment 29
'
~^^A~^-^—T^
•
O.2
O.15
O.1
O.O5
0
J;
19
O.6
O.5
O.4
O.3
0.2
O.1
9,
19
2O
15
1O
5
9=
19
O.2
O.15
O.1
O.O5
0
Phytoplankton (mg/L)
Segment 38
. -
t .
an July Jan July Dec
Q4 1994 1995 19951995
Particulate
- Organic Carbon (mg/L)
Segment 3O
1
an July Jan July Dec
94 1994 1995 19951995
Total Phosphorus (pg/L)
Segment 41
^— ~^~
an July Jan July Dec
94 1994 1995 19951995
Dissolved Silica (mg/L)
Segment 36
; , * •
^---•— j~^~
July Dec
1994 1994 1995 19951995
July
1994 1994 1995 19951995
water column
segment
numbers
Lake Michigan Level II
Water Column Seqmentation
10 surface water segments
41 total segments
1
11
20
29
36
1 0 average depth
1O of segment
10
20
103
Figure 2.5.4. Level 2 LM3-Eutro model output versus field data for selected segments.
165
-------
0.0928
non-diatoms
near-shore segment 10738
segment
10738
5 meter
depth
0.0025
i i i i i i i i i i i i i i i i i i i i i i i i
Jan.94 Jun.94 Dec.94 Jun.95 Dec.95
0.3408
0.2957
0.2506
j
n
0.2054
0.1603-
0.1152
particulate organic carbon
near-shore segment 10738
i i i i i—i i i i—i—i—i—r~i—i—n—i i i i '—TT
Jan.94 Jun.94 Dec.94 Jun.95 Dec 95
5.515-
5.120-
4.726-
4.332-
3.937-
3.543
total phosphorus
near-shore segment 10738
I I I I I I I I I I I I I
Jan.94 Jun.94 Dec.94
0.0799
0.0643-
0.0486-
0.0330-
|> 0.0173-1
0.0017
diatoms
off-shore segment 4096
\
I ii } i i i i r r i i r i i i i i i i ii i
Jan.94 Jun.94 Dec.94 Jun.95 Dec.95
0.331
0.282-
0.234-
0.185-
0.136-
0.087
* \
particulate organic carbon
off-shore segment 4096
i i i i i i i i i i i i i i i i i i i i i p t
Jan.94 Jun.94 Dec.94 Jun.95 Dec.95
5.298
4.887
4.475-
4.064-
3.652
3.241
total phosphorus
off-shore segment 4096
Jun.95 Dec.95
i i i i i i i i i i i i i r i i i i i i i i i i
Jan.94 Jun.94 Dec.94 Jun.95 Dec.95
Figure 2.5.5. Level 3 LM3-Eutro model output versus field data for selected nearshore and offshore
cells.
166
-------
nearshore cells than in the offshore cells (Figure
2.5.5).
2.5.2.2 Particulate Organic Carbon
The model fits the field data well (Figure 2.5.3), with
a slope of almost one (0.95) and a square of the
correlation coefficient of almost 0.4 (r2 = 0.39). The
POC data exhibited less scatter than the other
variables and were reflective of the phytoplankton in
the lake. Additionally, we had a great deal of
confidence in the POC data measurement technique.
The ability of the model to capture the POC data
trend increased our confidence in the model's overall
eutrophication predictions. Examination of Level 2
segments showed that the model captured important
trends, such as higher POC concentrations during
the spring diatom bloom (Figure 2.5.4).
2.5.2.3 Total Phosphorus
The model fits the total phosphorus data reasonably
well with a slope of 1.4 and a square of the
correlation coefficient of 0.37. In general, the total
phosphorus concentrations in the lake were fairly
constant, with most measurements falling between 4
and 5 ug/L and little seasonal variation observed.
The overall model fit was acceptable (Figure 2.5.4).
However, there were several higher total phosphorus
values measured in Green Bay close to the Fox River
and at other nearshore stations close to rivers and/or
areas where there were significant sediment
resuspension. In these cases, the model was not
able to mimic the data, probably due to initial
conditions that were too low and the lack of a
sediment resuspension term.
2.5.2.4 Dissolved Silica
The dissolved silica model fits the field data well, with
a slope of 1.2 (influenced by several very high silica
field data points) and a regression coefficient of 0.37.
The model predicted the expected trends, with
highest silica in the winter, a steep decline in the
epilimnion in the spring coinciding with the diatom
bloom, and a recovery toward the end of the year
(Figure 2.5.4).
In general, the model fit the field data reasonably
well. Seasonal and spatial trends for important
variables, such as phytoplankton and silica, were
captured. There were difficulties in accurately
predicting Green Bay phytoplankton and nutrient
concentrations, but this was not unexpected.
References
Cerco, C. and T. Cole. 1994. Three-Dimensional
Eutrophication Model of Chesapeake Bay. U.S.
Army Corps of Engineers, U.S. Army Engineer
Waterways Experiment Station, Vicksburg,
Mississippi. Technical Report Number EL-94-4,
658 pp.
Clesceri, L.S., A.E. Greenberg, and A.D. Eaton
(Eds.). 1998. Standard Methods for the
Examination of Water and Wastewater, 20th
Edition. American Public Health Association,
American Water Works Association, and Water
Environment Federation, Hanover, Maryland.
1,205 pp.
Thomann, R.V. 1982. Verification of Water Quality
Models. J. Environ. Engin., 108(EE5):933-940.
U.S. Environmental Protection Agency. 1997. Lake
Michigan Mass Balance Study (LMMB) Methods
Compendium, Volume 1: Sample Collection
Techniques. U.S. Environmental Protection
Agency, Great Lakes National Program Office,
Chicago, Illinois. EPA/905/R-97/012a, 1,440pp.
167
-------
PART 2
LM3-EUTRO
Chapter 6. Model Confirmation
The most common approach to model confirmation
is the comparison of the model output to an
independent field data set. Ideally, this independent
field data set describes a different year or years than
the calibration field data. We did not have the luxury
of access to a second field data set as thorough as
the 1994-1995 Lake Michigan Mass Balance Project
(LMMBP) data used in model calibration. However,
limited total phosphorus field data were available. In
addition, we were able to compare LM3-Eutro to the
historical MICH1 model, which was calibrated and
applied in the 1970s and 1980s (Rodgers and
Salisbury, 1981 a,b) and recently extended to predict
more recent chlorophyll and phosphorus
concentrations in Lake Michigan (Pauer et al.,
Submitted).
2.6.1 Additional Field Data
Limited Great Lakes National Program Office
(GLNPO) monitoring data were available for
comparison to model predictions. GLNPO data were
collected on an annual basis for the purpose of
monitoring long-term trends in the Great Lakes
(Barbiero era/., 2002). Samples were collected from
a set of stations that formed a north-south transect
through Lake Michigan (Barbiero et al., 2002). Green
Bay was not sampled as part of this lake monitoring
effort. Samples were taken from discrete depths
throughout the water column (Barbiero et al., 2002).
Data from the entire water column were averaged to
produce 1998 spring total phosphorus and 1998
summer total phosphorus lake-wide values. The
1998 spring and summer chlorophyll a data were
averaged to provide seasonal epilimnion (0-20 m in
depth) chlorophyll a values.
2.6.2 MICH1 Model
The Lake Michigan eutrophication model (MICH1)
was developed as part of the International Joint
Commission's (IJC) Great Lakes International
Surveillance Plan. The framework was constructed
by Rodgers and Salisbury (1981 a, b) based on the
Great Lakes model LAKE1 which was originally
developed and tested for Lake Ontario (Thomann ef
al., 1975). It is a four-segment model, simulating two
zooplankton classes, a single phytoplankton class (as
chlorophyll), and several nutrient species. However,
it does not have a sediment component and the
segmentation excludes Green Bay. MICH1 was
calibrated using field data from the Lake Michigan
intensive survey of 1976-1977 (Rockwell et al.,
1980). This model was recently resurrected and
extended to run from 1976 through 1995 and
compared to the LMMBP field data. Changes were
also made to the MICH1 model by reducing the
detrital settling rate by 20%, which results in a better
model fit with the LMMBP field data (Pauer et al.,
Submitted).
2.6.3 Comparison of LM3-Eutro to the
MICH1 Model and Field Data
In order to compare LM3-Eutro to the historical
MICH1 model, some modifications and qualifications
were necessary. The 1994 and 1995 loads were
repeated for the period 1996-2000 in both models.
However, the total phosphorus loads were averaged
for MICH1, while the loads were alternated in LM3-
168
-------
Eutro. Although the two approaches did not result in
any significant long-term differences, we observed
short-term differences. Because the two models
used very different segmentation schemes, all
comparisons were made on a lake-wide basis,
excluding Green Bay. LM3-Eutro algal carbon was
converted to chlorophyll a using a 40:1 carbon-to-
chlorophyll a ratio. All MICH1 simulations started in
1976 and ran through 2000, while LM3-Eutro was
only simulated from 1994 to 2000.
The results are shown in Figure 2.6.1. In general,
the two models compared reasonably well, which
was remarkable because the models are very
different in structure. MICH1 total phosphorus output
was lower than that of LM3-Eutro and the 1994-1995,
1998, and 2000 field data. The revised MICH1 (20%
reduced settling rate) compared more favorably with
the field data and LM3-Eutro.
The epilimnetic chlorophyll a concentration also
compared reasonably well between the models,
although LM3-Eutro predictions were higher than
both MICH1 predictions. The lower MICH1 output
values (as compared to LM3-Eutro) were probably
due to the absence of a sediment phosphorus
recycle mechanism. It was difficult to compare the
model versus field data for the chlorophyll due to the
steep peaks and large seasonal variation in the
chlorophyll a data.
The overall strength of the comparison between the
models and the model fit with limited 1998 and 2000
field data built confidence in the LM3-Eutro
framework and confirmed that the model was able to
represent the eutrophication state variables in Lake
Michigan.
References
Barbiero, R.P., M.L. Tuchman, G.J. Warren, and
D.C. Rockwell. 2002. Evidence of Recovery
from Phosphorus Enrichment in Lake Michigan.
Canadian J. Fish. Aquat. Sci., 59(10):1639-1647.
Pauer, J.J., K.W. Taunt, W. Melendez, and R.G.
Kreis, Jr. Submitted. Resurrection of the Lake
Michigan Eutrophication Model, MICH1. J. Great
Lakes Res, Submitted for publication.
Rockwell, D.C., D.S. DeVault, III; M.F. Palmer, C.V.
Marion, and R.J. Bowden. 1980. Lake Michigan
Intensive Survey, 1976-1977. U.S.
Environmental Protection Agency, Great Lakes
National Program Office, Chicago, Illinois.
EPA/905/4-80/003A, 155 pp.
Rodgers, P.W. and D. Salisbury. 1981 a. Modeling
of Water Quality in Lake Michigan and the Effect
of the Anomalous Ice Cover of 1976-1977. Great
Lakes Environmental Planning Study, Great
Lakes Basin Commission, Ann Arbor, Michigan.
Contribution Number 44, 53 pp.
Rodgers, P.W. and D. Salisbury. 1981b. Water
Quality Modeling of Lake Michigan and
Consideration of the Anomalous Ice Cover of
1976-1977. J. Great Lakes Res., 7(4):467-480.
Thomann, R.V., D.M. Di Toro, R.P. Winfield, and D.J.
O'Connor. 1975. Mathematical Modeling of
Phytoplankton in Lake Ontario, Part 1: Model
Development and Verification. U.S.
Environmental Protection Agency, Office of
Research and Development, ERL-Corvallis,
Large Lakes Research Station, Grosse lie,
Michigan. EPA/660/3-75/005, 177 pp.
169
-------
6.00
5.00-
4.00-
0.00
MICH1 and LM3-Eutro epilimnion chlorophyll-a
— MICH1
— MICH1 settling reduced 20%
LM3-Eutro
• 1994-1995 LMMB
D 1998
1976
1979
1982
1985
1988
1991
1994
1997
2000
12.0
10.0-
4.0-
2.0-
0
MICH1 and LM3-Eutro lakewide total phosphorus
— MICH1
— MICH1 settling reduced 20%
LM3-Eutro
• 1994-1995 LMMB
n 1998
o 2000
P
1976
—I 1 1 1 T
1979 1982
1 1 1 \ 1 1 1 T
1985 1988 1991
—I 1 1 1 \ T
1994 1997 2000
Figure 2.6.1. MICH1 versus LM3-Eutro model predictions and available field data.
170
-------
PART 2
LM3-EUTRO
Chapter 7. Results - Application of Model 2.7.1.2 Results and Discussion
A total of seven model scenarios were run to
evaluate future lake conditions under different total
phosphorus loads. Because total phosphorus is not
a model state variable, the individual phosphorus
state variables (soluble reactive phosphorus [SRP],
dissolved organic phosphorus [OOP], labile organic
phosphorus [LOP], and refractory organic
phosphorus [ROP]) were scaled accordingly to
accomplish the total phosphorus load increases or
reductions. The assumptions and conditions for each
model scenario are briefly described followed by a
Results and Discussion section. We evaluated future
lake-wide total phosphorus and particulate organic
carbon (POC) concentrations as well as epilimnetic
and hypolimnetic chlorophyll concentrations
(assuming a 20 m thermocline). Model simulation
time for the scenarios ranged between 20 and 30
years, with scenarios starting in 1994 and load
increases and decreases beginning in 2005.
2.7.7 Scenario 1 - Constant Conditions
2.7.1.1 Description of Assumptions
The scenario was started on January 1, 1994. A
constant user-specified net sediment total
phosphorus flux was applied in both space and time.
Total phosphorus tributary and atmospheric loads for
1994 and 1995 were repeated in a two-year cycle.
Hydrodynamics for 1994 and 1995 were similarly
repeated. The model was run until steady-state was
achieved.
The model reached steady-state within 28 years
(2021). The steady-state lake-wide concentrations for
total phosphorus and maximum POC were 4.3 ug/L
and 0.2 mg/L, respectively (Figure 2.7.1), while the
values for the spring maximum chlorophyll a
concentrations were 2.36 ug/L for the epilimnion and
1.07 ug/L for the hypolimnion (Figure 2.7.1).
2.7.2 Scenario 2 -
(Lower Bound)
Virtual Elimination
2.7.2.1 Description of Assumptions
The scenario was started on January 1, 1994. A
constant user-specified net sediment total
phosphorus flux was applied in both space and time.
The Constant Conditions scenario (Scenario 1) was
run from January 1, 1994 to December 31, 2004.
Beginning on January 1, 2005, tributary and
atmospheric total phosphorus loads were reduced by
100% and the sediment total phosphorus fluxes were
assumed to be zero. The model was run for a total
of 30 years (through December 2023).
2.7.2.2 Results and Discussion
As expected, the total phosphorus concentration
significantly declined once the atmospheric and
tributary total phosphorus loads were turned off. The
model predicted that after 30 years (December
2023), the lake-wide total phosphorus concentration
would be 0.54 ug/L (Figure 2.7.2). The epilimnetic
maximum chlorophyll a concentration after 30 years
would be less than 0.1 ug/L, while the hypolimnetic
171
-------
epilimnion chlorophyll-a
hypolimnion chlorophyll-a
1994 1998 2002 2006 2010 2014 2018 2021
1994 1998 2002 2006 2010 2014 2018 2021
total phosphorus
_J
ci
tn
0
a.
w
o
j:
n
JS
•*-'
'SO
4 n
^ n
20
1 0
0-
1
"^~— — ~— — __
•
994 1998 2002 2006 2010 2014 2018 2021
0.25
particulate organic carbon
1994 1998 2002 2006 2010 2014 2018 2021
Figure 2.7.1. Scenario 1: Constant Conditions.
epilimnion chlorophyll-a
1994 1998 2002 2006 2010 2014 2018 2022
1.2
1.0
i 0.6
Q. 0.4
0.2
hypolimnion chlorophyll-a
1994 1998 2002 2006 2010 2014 2018 2022
total phosphorus
particulate organic carbon
"S>
=>- 5.0-
™ 40-
o
J= 30-
CL ^'U
0 20
°- 1 o-
ro
o 0-
"" — ~_ ^
\
\
\
~^~
P 015-
" 0.10
O
°~ 005-
0-
M\M\*fi« n,n
AAI in! iAAAAAAi\
|\|BV\JvijVuv\i l/\
) \
\
\
^-—
1994 1998 2002 2006 2010 2014 2018 2022
1994 1998 2002 2006 2010 2014 2018 2022
Figure 2.7.2. Scenario 2: Virtual elimination.
172
-------
concentration would be approaching zero (Figure
2.7.2). In 2023, the lake-wide POC concentration
maximum was around 0.02 mg/L (Figure 2.7.2).
2.7.3 Scenario 3 - Best Estimate of
Current Trends Resulting From Previous
Actions
2.7.3.1 Description of Assumptions
The scenario was started on January 1, 1994. A
constant user-specified net sediment total
phosphorus flux was applied in both space and time.
Tributary and atmospheric total phosphorus loads
declined at rates observed over the last two decades
(1981-1995). The model was run for a total of 20
years (through December 2013).
The rate of total phosphorus decay was based on the
downward trend of total phosphorus in Lake Michigan
since the 1980s. In order to calculate the decay, it
was assumed that the total phosphorus equation was
an exponential. Figure 2.7.3 shows the Lake
Michigan historical total phosphorus loading.
Although historical total phosphorus loading data
c- 14000
CO
» 12000
•| 10000
a>
~ eooo
CD
O
O
.c
Q.
o
£ 2000
CO
6000
4000
—
-j
•-
- - target load
O 1974-1991
5600 mta (GLWQA 1978)
International Joint Commission
(1989), Dolan (pers.comm.)
D 1994-1995
USEPA, Lake Michigan
Mass Balance Study
-'.
-;'
_
"•"TTT
,
-
r.
.-,
n - n n
•3- in
Year
Figure 2.7.3. Historical total phosphorus loading
- Lake Michigan.
prior to 1981 were available, only the loading values
between 1981 and 1995 were used because it was
believed this provided the most realistic picture of the
present loading trend. No loading data were
available after 1995. Total phosphorus loads were
assumed to follow the equation:
= L(t0)exp[k(t-t0)]
(2-7.1)
where
t = time in units of years
t0 = initial time (1981)
L(t) - total phosphorus load at time t
L(t0) = total phosphorus load at time t0
k = total phosphorus decay rate in units of
1/year.
The decay rate k was calculated by applying the
Least Squares Fitting method and has a value of
-2.21 x 10'2/year.
2.7.3.2 Results and Discussion
The total phosphorus concentration steadily declined
over 20 years to approximately 3.5 ug/L (Figure
2.7.4). The epilimnetic chlorophyll a reached a value
of approximately 2.0 ug/L, while the hypolimnion fell
below 0.9 ug/L (Figure 2.7.4). In 2013, the lake-wide
POC maximum concentration was around 0.18 mg/L
(Figure 2.7.4).
2.7.4 Scenario 4 - Scenario 1 With
Instantaneous Reduction of Tributary
Loads to Zero
2.7.4.1 Description of Assumptions
The scenario was started on January 1, 1994. A
constant user-specified net sediment total
phosphorus flux was applied in both space and time.
The Constant Conditions scenario (Scenario 1) was
run from January 1, 1994 to December 31, 2004.
Beginning on January 1, 2005, tributary total
phosphorus loads were reduced by 100%. The 1994
and 1995 atmospheric load cycle was continued.
The model was run for a total of 30 years (through
December 2023).
173
-------
epilimnion chlorophyll-a
1994 1997 2000 2003 2006 2009 2012
hypolimnion chlorophyll-a
1994 1997 2000 2003 2006 2009 2012
total phosphorus
o.u
5 0
4.0
3.0
o n
1.0
n
~~~ — — — —
participate organic carbon
1994 1997 2000 2003 2006 2009 2012
1994 1997 2000 2003 2006 2009 2012
Figure 2.7.4. Scenario 3: Best estimate of current trends resulting from previous actions.
2.7.4.2 Results and Discussion
There was a significant decline in the total
phosphorus concentration when the tributary loads
were turned off in January 2005.(Figure 2.7.5).
However, this decline was not as steep as that
observed in Scenario 2 (Virtual Elimination). The
total phosphorus concentration at the end of 2023
was 0.91 ug/L, which was higher than the 0.54 ug/L
observed in the Virtual Elimination scenario (Scenario
2). Similarly, after 30 years, the maximum
chlorophyll a concentration fell to approximately 0.4
ug/L and less than 0.1 ug/L for the epilimnion and
hypolimnion, respectively (Figure 2.7.5), and the
lake-wide POC maximum was 0.03 mg/L (Figure
2.7.5). These values were all somewhat higher than
their equivalents in Scenario 2.
2.7.5 Scenario 5 - Scenario 1 With
Instantaneous Reduction of Atmospheric
Loads to Zero
2.7.5.1 Description of Assumptions
The scenario was started in January 1, 1994. A
constant user-specified net sediment total
phosphorus flux was applied in both space and time.
The Constant Conditions scenario (Scenario 1) was
run from January 1, 1994 to December 31, 2004.
Beginning on January 1, 2005, atmospheric total
phosphorus loads were reduced by 100%. The 1994
and 1995 tributary load cycle was continued. The
model was run for a total of 20 years (through
December 2013).
2.7.5.2 Results and Discussion
Turning off the atmospheric total phosphorus loads
had little effect on the total phosphorus, chlorophyll a,
and POC concentration (Figure 2.7.6) as compared
to the Constant Conditions scenario (Figure 2.7.1).
2.7.6 Scenario
Tributary and
increased 20%
6 - Scenario 1 With
Atmospheric Loads
2.7.6.1 Description of Assumptions
The scenario was started on January 1, 1994. A
constant user-specified net sediment total
phosphorus flux was applied in both space and time.
The Constant Conditions scenario (Scenario 1) was
run from January 1, 1994 to December 31, 2004.
174
-------
~ 3.0
epilimnion chlorophyll-a;
1994 1998 2002 2006 2010 2014 2018 2022
hypolimnion chlorophyll-a
"§> 1.0
tn|
1994 1998 2002 2006 2010 2014 2018 2022
total phosphorus
particulate organic carbon
O)
3 5.0
" 40
£ 30
Q. ->-u
S on
0. , „
i.u ^
fl n
— ^
\
\
\
"^--______
^ n TTI
f 0 15
O n 10
o
°" 0 05
0
,
AI\l\!\AAAF\i\AA'A
ilVvvVvUwUuui
Vv
\.
— -v—^^, ,
1994 1998 2002 2006 2010 2014 2018 2022
1994 1998 2002 2006 2010 2014 2018 2022
Figure 2.7.5. Scenario 4: Scenario 1 with tributary load elimination.
epilimnion chlorophyll-s
1994 1997 2000 2003 2006 2009 2012
hypolimnion chlorophyll-a
1994 1997 2000 2003 2006 2009 2012
total phosphorus
particulate organic carbon
-^ o.u ,
D)
3 5.0
1 4.0
o
•S 30
o 20
.c *"u
o- H n
z 1'°
3 0
"~~~~ — — ~— — —_______„__
s~. n 9n
ra rt . ,_
£ 0.15
O n in
O
°- 0.05 .
0
f\r\!\f\i\r\
A/V VV V V Wv vW\AAAAAA/^
i/
1994 1997 2000 2003 2006 2009 2012
1994 1997 2000 2003 2006 2009 2012
Figure 2.7.6. Scenario 5: Scenario 1 with atmospheric load elimination.
175
-------
Beginning on January 1, 2005, tributary and
atmospheric total phosphorus loads were increased
by 20%. The model was run for a total of 20 years
(through December 2013).
2.7.6.2 Results and Discussion
The 20% increase in total phosphorus loads had a
relatively small influence on the total phosphorus
concentration, the epilimnetic chlorophyll a
concentration, and the POC concentration (compare
Figure 2.7.7 [20% load increase] with Figure 2.7.1
[Constant Conditions]). The lake-wide total
phosphorus concentration after 20 years was 4.6
ug/L (Figure 2.7.7). Maximum chlorophyll a
concentrations after 20 years were 2.2-2.5 ug/L
(epilimnion) and 1.1 ug/L (hypolimnion) (Figure
2.7.7). The lake-wide POC maximum concentration
was 0.22 mg/L (Figure 2.7.7).
2.7.7 Scenario 7 - Application of Great
Lakes Water Quality Agreement Loads to
Model
2.7.7A Description of Assumptions
The scenario was started on January 1, 1994. A
constant user-specified net sediment total
phosphorus flux was applied in both space and time.
The Constant Conditions scenario (Scenario 1) was
run from January 1, 1994 to December 31, 2004.
Beginning on January 1,2005, the 1978 Great Lakes
Water Quality Agreement (GLWQA) specified total
phosphorus target loading of 5,600 MT/year was
applied (International Joint Commission, 1978). The
1994 and 1995 atmospheric load cycle was
continued. A new user-defined net sediment total
phosphorus flux was estimated, assuming that
approximately 95% of the phosphorus load was
retained in the sediment and 5% was recycled back
to the water column. The model was run to steady-
state.
epilimnion chlorophyll-a
1994 1997 2000 2003 2006 2009 2012
hypolimnion chlorophyll-s
1994 1997 2000 2003 2006 2009 2012
=d 6.0
I 5.0
c/l
S 4.0
o
-& 3.0
8 2.0
5 1.0
total phosphorus
1994 1997 2000 2003 2006 2009 2012
0.25
partculate organic carbon
1994 1997 2000 2003 2006 2009 2012
Figure 2.7.7. Scenario 6: Scenario 1 with tributary and atmospheric loads increased 20%.
176
-------
2.7.7.2 Results and Discussion
Applying the GLWQA total phosphorus load of 5,600
MT/year resulted in a lake-wide steady-state total
phosphorus concentration of 7.5 ug/L and an
epilimnetic chlorophyll a maximum of 4.0 ug/L (Figure
2.7.8). The hypolimnetic chlorophyll a maximum at
steady-state was 1.6 ug/L (Figure 2.7.8). Steady-
state lake-wide maximum POC was approximately
0.28 mg/L (Figure 2.7.8). Steady-state was reached
within 30 years.
2.7.5 Scenario 8 - Estimate of Total
Maximum Daily Loads to Reach
International Joint Commission's Target
Total Phosphorus Concentration
2.7.8.1 Description of Assumptions
The scenario was started on January 1, 1994. A
constant user-specified net sediment total
phosphorus flux was applied in both space and time.
The Constant Conditions scenario (Scenario 1) was
run from January 1, 1994 to December 31, 2004.
Through trial-and-error, a total phosphorus load
(tributary and atmospheric) that resulted in steady-
state total phosphorus concentration of 7 ug/L (the
lake-wide International Joint Commission's [IJC]
target) was determined (Great Lakes Research
Advisory Board, 1978). A new user-defined net
sediment total phosphorus flux was estimated,
assuming that approximately 95% of the phosphorus
load was retained in the sediment and 5% was
recycled back to the water column. The IJC total
phosphorus concentration target was chosen with the
goal of returning Lake Michigan to its "natural
oligotrophic state" under the GLWQA (International
Joint Commission, 1978). The model was run to
steady-state.
2.7.8.2 Results and Discussion
An average annual total phosphorus load of 5,020
MT resulted in a steady-state lake-wide total
phosphorus concentration of 7 ug/L (Figure 2.7.9).
This equated to a total phosphorus total maximum
daily load (TMDL) of 14 MT/d. This also resulted in
a spring epilimnetic maximum chlorophyll a
concentration of 3.7 ug/L and a spring hypolimnetic
chlorophyll a concentration of 1.6 ug/L (Figure 2.7.9).
Steady-state lake-wide maximum POC was around
0.33 mg/L (Figure 2.7.9). Steady-state was reached
within approximately 30 years.
2.7.9 Scenario Comparison and
Discussion
A summary of the final total phosphorus, chlorophyll
a, and POC concentrations is shown in Table 2.7.1.
Examining the scenarios revealed a number of
interesting conclusions regarding Lake Michigan. It
was apparent from comparing Scenarios 1, 4, and 5
that tributary loading was considerably more
important than atmospheric loading in driving Lake
Michigan total phosphorus and chlorophyll a (Figures
2.7.1, 2.7.5, and 2.7.6). However, as Scenario 6
revealed, a small increase in loading (tributary and
atmospheric) does not have a large impact on the
lake (Figure 2.7.7). Scenario 2 confirmed that Lake
Michigan's reaction to significant loading changes is
immediate but buffered by the large water volume
and slow retention time of the lake (Figure 2.7.2).
Scenario 3 suggested that if current loading trends
continue, lake-wide total phosphorus and chlorophyll
a will continue to slowly decline (Figure 2.7.4).
Scenarios 7 and 8 demonstrated the drastic
increases in chlorophyll a, total phosphorus, and
POC that would occur if loading to Lake Michigan
were allowed to increase to GLWQA/IJC limits
(Figures 2.7.8 and 2.7.9).
2.7.70 Mass Budget
The sources, sinks, and lake inventory of total
phosphorus was estimated. Figure 2.7.10 is a
graphical representation of the average annual loads,
sinks, and total phosphorus inventory for Lake
Michigan based on 1994 and 1995 modeled and
measured data. It was clear that the internal recycle
(settling and sediment feedback) accounted for the
majority of total phosphorus dynamics occurring in
the lake. A significant mass of phosphorus settled to
the lake bed, but a large percentage (~ 60%) is
recycled back to the water column. The monitored
tributaries made up the largest external total
phosphorus source to the lake, while the
unmonitored tributaries and atmospheric loads were
relatively small components. The total phosphorus
export at the Straits of Mackinac and the Chicago
diversion were estimated to be a small fraction of the
177
-------
epilimnion chlorophyll-a
hypolimnion chlorophyll-a
1994 1998 2002 2006 2010 2014 2018 2022
7.0
6.0
5.0
total phosphorus
1994 1998 2002 2006 2010 2014 2018 2022
particulate organic carbon
* 4.0
o
a.
at
o
a
3.0
2.0
o 1.0
1994 1998 2002 2006 2010 2014 2018 2022
1994 1998 2002 2006 2010 2014 2018 2022
Figure 2.7.8. Scenario 7: Application of the GLWQA loads to the model.
epilimnion chlorophyll-a
hypolimnion chlorophyll-a
.C
a
o
o
ra
u>
3
O
JC
a
in
o
3
o
8.0
7.0
6.0
5.0
4.0
3.0
2.0
1.0
0
1994 2004 2014
total phosphorus
2024
2034
2044
111111
2004 2014
1994
particulate organic carbon
2024
2034 2044
1994
2004
2014
2024
2034
2044
1994
2004
2014
2024 2034
Figure 2.7.9. Scenario 8: Estimate of the TMDL to reach the IJC's target total phosphorus
concentration.
178
-------
Table 2.7.1. Final Eutrophication Scenario Results
Scenario
and
Length
1 (28
years)
2(30
years)
3(20
years)
4(30
years)
5(30
years)
6(20
years)
7(30
years)
8(30
years)
Description
Constant Conditions Remain
From 1994-1 995
Virtual Elimination (Lower
Bound)
Best Estimate of Current
Trends Resulting From
Previous Actions
Scenario 1 With
Instantaneous Reductions of
Tributary Loads to Zero
Scenario 1 With
Instantaneous Reductions of
Atmospheric Loads to Zero
Scenario 1 With Tributary
and Atmospheric Loads
Increased 20%
Application of GLWQA
Loads to Model
Estimate of TMDL to Reach
IJC Target Total
Spring
Epilimnion
Chlorophyll
a(ug/L)
2.4
0.1
2.0
0.4
2.3
2.5
4.0
3.7
Maximum
Hypolimnion
Chlorophyll
afog/L)
1.1
~0
0.9
0.1
1.1
1.1
1.6
1.6
Maximum
Total Particulate
Phosphorus Organic
(ug/L) Carbon
(mg/L)
4.3 0.20
0.54 0.02
3.5 0.18
0.91 0.03
4.1 0.20
4.6 0.22
7.5 0.34
7.0 0.33
Phosphorus Concentration
179
-------
atmospheric
deposition
290,000
sediment recycle
4,000,000
export to
Lake Huron
140,000
i export via
Chicago
Diversion
13,000
Total Phosphorus Inventory
water column = 24,000,000 kg
sediment1
burial
3,000,000
tributary loading
monitored - 1,830,000
unmonitored - 670,000
(Lake Michigan watershed)
Figure 2.7.10. Annual average (1994-1995) Lake Michigan total phosphorus loading (kg/year).
total export. Overall, the best estimate using the
average of the 1994-1995 loads was that there was
a 5% annual loss of total phosphorus in the lake,
which suggested that there would be a small, but
steady, decrease in the total phosphorus lake
concentration given constant total phosphorus loads.
Figure 2.7.11 is a breakdown of the total phosphorus
loads and inventory of the main lake and Green Bay.
It depicts the loads entering the bay and lake
separately and the phosphorus exchange between
these two system. Phosphorus from Green Bay
accounted for approximately 13% of the total
phosphorus input into Lake Michigan. The bulk of
the Green Bay total phosphorus load was from the
Fox River.
References
Great Lakes Advisory Board. 1978. Annual Report
to the International Joint Commission.
International Joint Commission, Windsor, Ontario,
Canada. 44 pp.
International Joint Commission. 1978. Great Lakes
Water Quality Agreement of 1978, with Annexes
and Terms of Reference, Between the United
States and Canada, Signed at Ottawa, November
22, 1978. International Joint Commission,
Windsor, Ontario, Canada. 60 pp.
180
-------
Green Bay
monitored and
un monitored
tributary loading
930,000
atmospheric
loads
18,000
Green Bay Mass Budget
atmospheric
deposition
270,000
Green Bay
export
230,000
sediment recycle /
3,700,000 A
export to
Lake Huron
140,000
settling
6,500,000
export \i\a
^ Chicago
Diversion
13,000
Total Phosphorus Inventory
Main Lake:
water column = 23,300,000 kg
Green Bay:
water column = 700,000 kg
sediment'
burial
2,800,000
main lake monitored and
unmonitored tributary loading
(Lake Michigan watershed
excluding Green Bay)
1,570,000
Figure 2.7.11. Annual average (1994-1995) Lake Michigan and Green Bay total phosphorus loading
(kg/year).
181
-------
PART 2
LM3-EUTRO
Chapter 8.
Toxic
Results Provided for LM2- 2.8.2 Manipulation of Results
2.8.1 Description
LM2-Toxic was developed to simulate congener-
specific polychlorinated biphenyl (PCB) state
variables. Due to the importance of organic carbon
in the fate and transport of PCBs, the model also
simulated three carbon states: biotic carbon (BIG),
particulate detrital carbon (PDC), and dissolved
organic carbon (DOC). LM2-Toxic relied on external
calculations (measured or modeled) to estimate the
autochthonous and allochthonous carbon loads. The
internally produced carbon made up the majority of
carbon entering the lake. Thus, a reliable estimate of
this internal load was of utmost importance in
accurately simulating the organic carbon in the
system. The main purpose of the eutrophication
model (LM3-Eutro) in the Lake Michigan Mass
Balance Project (LMMBP) was to provide
autochthonous (internally produced) phytoplankton
carbon to the PCB fate and transport model (LM2-
Toxic).
LM3-Eutro and LM2-Toxic utilize very different
modeling frameworks, with different segmentation
schemes, hydrodynamics, and transport
mechanisms. Several modifications were made to
LM3-Eutro to ensure data compatibility when
exporting the autochthonous carbon to LM2-Toxic.
Because all necessary changes were made within
LM3-Eutro code, no post-processing was necessary.
LM3-Eutro generated carbon from primary production
at each model time step. The model used a variable
time step of approximately three hours. The carbon
was totaled on a daily basis, and the high-resolution
LM3-Eutro 5 km2 segments were collapsed to the
Level 2 segmentation scheme to generate daily
allochthonous carbon loads for each of the 41 Level
2 segments. These loads were generated for the
1994-1995 calibration years, as well as a long-term
simulation where the 1994 and 1995 loading and
hydrodynamics data were repeated for approximately
28 years (see Part 2, Chapter 7, Section 2.7.1 -
Constant Conditions Remain From 1994-1995).
182
-------
PARTS
LEVEL 1 MODELS
Douglas D. Endicott
Great Lakes Environmental Center
Traverse City, Michigan
and
Timothy J. Feist
Gregory Gerstner
Welso Federal Services, LLC
Large Lakes Research Station
9311 Groh Road
Grosse lie, Michigan 48138
and
Kenneth R. Rygwelski
U.S. Environmental Protection Agency
Office of Research and Development
National Health and Environmental Effects Research Laboratory
Mid-Continent Ecology Division
Large Lakes and Rivers Forecasting Research Branch
Large Lakes Research Station
9311 Groh Road
Grosse lie, Michigan 48138
Chapter 1. MICHTOX PCB Model
Executive Summary
MICHTOX is a toxic chemical mass balance and food
chain bioaccumulation model. In this project, the
model was used to provide a screening-level analysis
of the potential future trends in total polychlorinated
biphenyl (PCB) concentrations in Lake Michigan
water, sediment, and fish under a variety of
contaminant load scenarios. The model also
provided a comparison of an older, "off-the-shelf"
model with the more complex models developed as
part of the Lake Michigan Mass Balance Project
(LMMBP). Results of the MICHTOX modeling
indicate that atmospheric exchange is a dominant
loss process of total PCBs in Lake Michigan, and that
the reservoir of total PCBs in the sediment has a
significant impact on the future trends in
concentrations of total PCBs in lake trout.
MICHTOX was developed in the early 1990s to
provide guidance for the Lake Michigan Lake-wide
Management Plan (LaMP) and to assist with the
planning of the LMMBP (Endicott et al., 2005).
During the early part of the LMMBP, MICHTOX was
updated and used as a preliminary assessment tool
of the LMMBP PCB data (Endicott, 2005). For the
present study, the updated fate and transport
submodel was used to provide exposure
concentrations to the food chain submodel under
183
-------
seven scenarios of future PCB loadings to Lake
Michigan.
Chapter 3 summarizes the work with the MICHTOX
fate and transport submodel. This includes the
evaluation of historical loading trends in the earlier
project (Endicott, 2005). The preliminary modeling
suggested that the scenario of historical PCB loads
to the lake that best fit the available data was one in
which loads increased from zero at a start date of
1940, peaked in 1961-1963, and then declined to
present levels. This hindcast was later updated
using the LM2-Toxic model, as described in Part 4.
A Bayesian Monte Carlo (BMC) uncertainty analysis
was also conducted in the earlier project that
demonstrated that MICHTOX predicted PCB
concentrations should be within a factor of two of the
measured data.
For the present study, atmospheric and tributary
loads, including unmonitored tributary inputs, were
calculated for the 1994-1995 LMMBP sampling
period. The model was run using these inputs and
the previously developed parameterization, and the
applicability of MICHTOX as a screening model for
predicting Lake Michigan total PCB concentrations in
water, sediment, and fish was reconfirmed.
Chapter 4 summarizes the work with the MICHTOX
food chain submodel and the application of the model
for predicting potential total PCB concentrations
under different loading scenarios. Data to
parameterize the food chain model was obtained
from the LMMBP sampling effort. The applicability of
the food chain model was confirmed by applying it to
the previous hindcast scenarios and to 1994-2000
Lake Michigan lake trout data.
MICHTOX was run for seven scenarios to help
evaluate future loading trends and the impacts on
PCB concentrations of various loading sources.
These scenarios included:
> Continued loading at 1994-1995 levels
* Continued recovery - fast rate
*• Continued recovery - slow rate
»• Zero atmospheric deposition
>• Zero tributary loads
* Zero atmospheric deposition and zero tributary
loads
* Lake-wide sediment cleanup
The scenario model runs indicated that if declining
trends in loading sources occurred at the faster of
rates found in the scientific literature, the total PCB
concentrations in an average 5-6 year-old lake trout
in southern Lake Michigan would be reduced below
the fish consumption advisory target level by
approximately the year 2025. If loading sources
declined at the slower rates found in the literature,
total PCB concentrations in an average 5-6 year-old
lake trout would be reduced below the target level by
approximately 2053. The sensitivity scenarios
indicated that the system was more affected by
atmospheric vapor concentration and deposition than
tributary loadings, and that the sediment reservoir of
total PCBs played a large role in the concentrations
observed in lake trout.
References
Endicott, D.D. 2005. 2002 Lake Michigan Mass
Balance Project: Modeling Total PCBs Using the
MICHTOX Model. In: R. Rossmann (Ed.),
MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan, Part 2. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
EPA/600/R-05/158, 140 pp.
Endicott, D.D., W.L. Richardson, and D.J. Kandt.
2005. 1992 MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan. In: R. Rossmann (Ed.),
MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan, Part 1. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse Me, Michigan.
EPA/600/R-05/158, 140 pp.
184
-------
PARTS
LEVEL 1 MODELS
Chapter 2. MICHTOX Recommendations
MICHTOX was adapted from the general model,
WASP4, and has served well as a screening-level
model for Lake Michigan over the past several
decades. Much of the model development took place
prior to the availability of an extensive data set
collected for the Lake Michigan Mass Balance
Project (LMMBP) during 1994-1995 and, therefore,
depended heavily on existing historical data. In
contrast, the LM2-Toxic model and LM Food Chain
model were constructed using the most recent data
from the LMMBP. Some of the advantages of using
LM2-Toxic instead of MICHTOX as a screening-level
model for future contaminants of interest include the
following:
• LM2-Toxic has a significant amount of
documentation.
• LM2-Toxic algorithms are all contained within the
WASP code, whereas MICHTOX utilizes Excel
spreadsheets for some of the calculations. This
makes code modifying in LM2-Toxic easier.
• LM2-Toxic automatically corrects the Henry's Law
Constant for temperature.
• LM2-Toxic has a better treatment of carbon
(including having biotic and abiotic carbon and
carbon decay).
• LM2-Toxic is as easy and fast as MICHTOX in
preparing model runs for similar numbers of state
variables.
• LM2-Toxic utilizes output from the hydrodynamic
model to compute advective flows and vertical
exchanges.
• LM2-Toxic carbon state variables are from the
LM3-Eutro model for defining autochthonous
carbon generation.
• LM2-Toxic handles sediment as a limited source
for resuspension; whereas, MICHTOX does not.
• LM Food Chain has more organisms in its food
web.
• LM2-Toxic has a higher spatial resolution in both
the water and sediment. This higher resolution
allows one to utilize this resolution if data sets
related to a new contaminant of interest are well-
populated.
Therefore, future enhancements of MICHTOX are not
warranted.
185
-------
PARTS
LEVEL 1 MODELS
Chapter 3. MICHTOX PCB Fate and
Transport Modeling
MICHTOX is a toxic chemical mass balance and
bioaccumulation model for Lake Michigan. The
model was developed to simulate the transport and
fate of polychlorinated biphenyls (PCBs) and other
toxic chemicals in Lake Michigan, and it has served
as the screening-level model in the United States
Environmental Protection Agency (USEPA), Large
Lakes Research Station's (LLRS) suite of PCB
models.
MICHTOX contains both a fate and transport
submodel and a food chain bioaccumulation
submodel. The fate and transport submodel predicts
water and sediment concentrations that are used as
exposure concentrations by the food chain submodel.
MICHTOX was developed in the early 1990s to
provide guidance for the Lake Michigan Lake-wide
Management Plan (LaMP) and to assist with the
planning of the Lake Michigan Mass Balance Project
(LMMBP) (Endicott et al., 2005). The model was
later updated and used as a tool for a rapid,
preliminary assessment of LMMBP PCB data
(Endicott, 2005).
The focus of the MICHTOX modeling in the present
study was to provide a screening-level evaluation of
the effects of different loading scenarios on lake trout
PCB concentrations, using the MICHTOX food chain
submodel and the LMMBP data for organisms in the
lake trout food chain. The organism data and
descriptions of the scenarios are included in Chapter
4. For the present study, the same MICHTOX fate
and transport submodel parameterization was used
as that used for the early LMMBP preliminary mass
balance modeling assessment conducted by Endicott
(2005). Loads were adjusted to reflect later findings
of the LMMBP. This chapter summarizes this work
to provide background for the food chain modeling
assessments presented in Chapter 4. The
MICHTOX modeling also provided an opportunity to
compare an established, "off-the-shelf model to the
more complex Level 2 models developed during the
LMMBP.
The following chapter contains a brief description of
MICHTOX, a description of the data used in the
modeling, a discussion of the hindcast model
confirmation, a comparison of model results to the
1994-1995 data, and a discussion of the evaluation
of uncertainty in the model results using Bayesian
Monte Carlo (BMC) methods. A more complete
description of the model, the hindcast confirmation,
and the BMC evaluation can be found in Endicott et
al. (2005) and Endicott (2005).
3.3.1 Description
MICHTOX was implemented using the USEPA
WASP4 modeling framework (Ambrose et al., 1988).
The model was modified by the USEPA to run on
Hewett-Packard (HP) Alphas running Tru64 Unix. As
part of the early LMMBP data assessment (Endicott,
2005), MICHTOX was revised to reflect updated
information on the chemical properties of PCBs and
fate processes. In addition to using the new PCB
data, the following modifications or additions were
made at that time: the water balance was corrected
to maintain continuity for long model runs, the
186
-------
boundary condition at the Straits of Mackinac was
enhanced, the segment-specific atmospheric vapor
concentrations were specified, and the chemical
volatilization rate formulations and Henry's Law
Constant parameterizations were updated. This is
the version of the model used for the hindcast study
described in Sections 3.3.3.1 and 3.3.3.2. For the
most recent study discussed in Section 3.3.3.3 and
Chapter 4, MICHTOX predictions were improved by
updating atmospheric deposition PCB loadings and
adding unmonitored tributary PCB loadings.
MICHTOX used differential equations and the
concept of mass balance to solve for concentrations
of contaminants, in this case PCBs, both temporally
and spatially in the water column and sediments.
Water movement, contaminant transport, and
contaminant fate processes were used in the model
to track the mass of PCBs from sources (tributary
and atmospheric) to sinks. The mass balance
equation for PCBs included terms for loads to the
system, advective transport, vertical and horizontal
dispersive exchange, settling, resuspension,
atmospheric deposition, vapor exchange with the
atmosphere (volatilization and absorption), sediment
water diffusion, and burial from the surficial sediment
segments to deep sediment (Figure 3.3.1).
Contaminant concentrations were accounted for both
in their dissolved and particulate states. The
complete coupled mass balance equations for the
contaminant concentrations can be found in "1992
MICHTOX: A Mass Balance and Bioaccumulation
Model for Toxic Chemicals in Lake Michigan"
(Endicott et al., 2005).
Watershed
Atmosphere
Tributary
Loading
Epilimnion
Hypolimnion
Surficial
Sediment
Transport and
Exchange
Absorption
Volatilization
sorbed chemical
POC
-- Partitioning
Resuspension Settling
output to food
chain model
sorbed chemical
POC
Resuspension
output to food
chain model
Exchange
Figure 3.3.1. MICHTOX PCB mass balance schematic.
Deep
sediment
187
-------
MICHTOX simulates total PCBs as the sum of two
homologs (tetrachlorobiphenyl [PCB4] and
pentachlorobiphenyl [PCB5]). Although this is
technically incorrect (there are ten homologs), this
representation of total PCBs as two PCB homologs
was considered a reasonable compromise between
pre-LMMBP loading and concentration data, mostly
quantified as total PCBs or Aroclors, and congener-
specific estimates available for physicochemical
model parameters. Loads of total PCBs to the model
were evenly divided between the two homologs.
MICHTOX model segmentation included both water
column and surficial sediment segments. The water
column segmentation included ten segments of
varying geometry (Table 3.3.1), consisting of three
epilimnetic segments in the main lake, three
corresponding hypolimnetic segments in the main
lake, three water column segments in Green Bay,
and one segment for the lower Fox River (segment 4)
that was not used for the LMMBP modeling (Figure
3.3.2). There were seven sediment segments; each
corresponded to an epilimnetic/hypolimnetic segment
in the water column. The Fox River sediment
segment (segment 14), underlying the unused water
column segment, was not used for the LMMBP
modeling.
Table 3.3.1. MICHTOX Segment Geometry
Segment Volume (m3)
Depth (m)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
2.61 E+11
2.11E+11
7.49E+10
1.46E+07
2.23E+09
1.89E+10
4.61 E+10
1.70E+12
2.27E+12
1.91 E+11
5.81 E+08
4.09E+08
4.95E+07
1.97E+05
7.60E+06
4.20E+07
3.51 E+07
10
10
10
2.22
5.87
13.3
16.3
65.3
108
25.5
0.033
0.033
0.033
0.10
0.04
0.04
0.04
n
m
seasonally-stratified
water column
m completely-mixed
water column
frrf
4 °
surficial
sediment
sediment sampling
location
Figure 3.3.2. MICHTOX model segmentation.
The parameters used for all MICHTOX model runs
are shown in Table 3.3.2. Volatilization rates were
input as a monthly time series and were based on the
Wanninkhoff (1992) formulation for water mass
transfer resistance and the Schwarzenbach et al.
(1993) formulation for gas mass transfer resistance.
Henry's Law Constants were also input as a monthly
time series as a function of average monthly surface
water temperatures (Bamford et al., 2000). Endicott
(2005) contains a complete description of the revised
gas exchange equations in MICHTOX.
188
-------
Table 3.3.2. Model Parameters and Coefficients
Parameter
Value
Unit
Particle settling velocity
Pore water diffusion coefficient
Monthly fraction organic carbon:
for surface segments 1 -7
for deep segments 8-10
for sediment segments 11-17
Log(organic carbon partition
coefficientXKJ:
-PCB4
-PCB5
Monthly water temperatures
Monthly temperature-dependent
Henry's Law Constants
-PCB4
-PCB5
Monthly volatilization rates:
-PCB4
-PCB5
1.5
1.8e-5
0.127-0.290
0.039-0.090
0.023-0.052
6.18
6.46
1.7-19.2
1.2e-4-
2.6e-4
1.5e-4-
3.2e-4
0.55-1.72
0.56-2.02
m/day
m2/day
atm-
m3/mol
m/day
3.3.2 Description of Data Used in
MICHTOX
The following is a description of the data used in the
MICHTOX PCB modeling for the LMMBP.
3.3.2.1 Water Column PCB Concentrations
Water-column total PCB concentrations (McCarty et
al., 2004) were averaged for each water column
segment and cruise of the LMMBP using a volume-
weighted averaging (VWA) procedure (Appendix
4.4.1). Dissolved (filtered) and paniculate total PCB
concentrations were averaged separately. Average
dissolved total PCB concentrations are presented in
Table 3.3.3, and particulate total PCB concentrations
are presented in Table 3.3.4.
3.3.2.2 Surficial Sediment PCB Concentrations
Sediment total PCB concentrations were measured
in 133 surficial sediment samples; 50 were collected
from the top 1 cm increment of box cores, and 65
were Ponar samples (McCarty et al., 2004). All total
PCB data were interpolated onto a uniform grid using
a natural-neighbor algorithm and averaged for each
MICHTOX surficial sediment segment (Appendix
4.4.1). Average total PCB concentrations were also
calculated for the box core samples in each main
lake sediment segment (Segments 11 -13). Relatively
few sediment samples were collected in Green Bay;
therefore, surficial sediment total PCB concentrations
from cores collected during the 1989-1990 Green
Bay Mass Balance Project (GBMBP) (Manchester-
Neesvig era/., 1996) were used to calculate average
concentrations in Green Bay sediment (Segments
15-17). Segment-specific average total PCB
sediment concentrations are presented in Table
3.3.5. Model runs that started in 1994 used the
segment-average box core data for initial conditions.
3.3.2.3 Atmospheric and Tributary Loads
Forcing functions used in MICHTOX consisted of
atmospheric vapor concentrations, atmospheric (wet
and dry) deposition, and tributary loads. The forcing
functions developed for PCBs by the LMMBP were
believed to be accurate estimates for the 1994-1995
period, based upon the data quality objectives and
well-developed estimation procedures. Previous
studies (Endicott et al., 2005; Endicott, 2005),
including the hindcast scenarios described below,
used tributary loads from the major river systems.
The present study, including the 1994-1995 model
runs and the forecast scenarios, also included load
estimates from unmonitored tributaries.
A number of sources were available to characterize
trends of the usage and release of PCBs in the Great
Lakes during the 20th Century, and these were used
to extrapolate the LMMBP forcing functions both
backward in time for estimates of historical loading
trends (Endicott, 2005) and forward in time for model
scenarios (the present study, Chapter 4). The total
PCB forcing functions were based upon the LMMBP
estimated long-term trends in PCB usage and
loadings to and concentrations in Lake Michigan and
the Great Lakes (McCarty et al., 2004). Estimation
of hindcast forcing functions is described in the next
section, Section 3.3.3. Estimation of the forecast
forcing functions is discussed in Chapter 4.
189
-------
Table 3.3.3. Cruise and Segment-Specific Average Dissolved Total PCB Concentrations (ng/L)
Segment Segment Segment Segment Segment Segment Segment Segment Segment
Date 12356 7 8 9 10
May 94
Jun 94
Aug94
Oct94
Jan 95
Apr 95
Aug95
Sep95
0.173
0.124
0.167
0.204
0.162
0.196
0.216
0.244
0.145
0.114
0.172
0.184
0.160
0.169
0.183
0.172
0.072
0.110
0.129
0.146
0.160
0.141
0.189
0.127
0.912
0.567
0.404
0.478
0.234
0.253
0.916
0.565
0.400
0.473
0.234
0.253
0.152
0.348
0.190
0.261
0.234
0.142
0.133
0.131
0.181
0.214
0.173
0.270
0.254
0.121
0.106
0.167
0.194
0.150
0.221
0.176
0.064
0.100
0.138
0.154
0.137
0.210
0.138
Table 3.3.4. Cruise and Segment-Specific Average Particulate Total PCB Concentrations (ng/L)
Date
May 94
Jun 94
Aug94
Oct94
Jan 95
Apr 95
Aug95
Sep95
Segment
1
0.147
0.088
0.031
0.102
0.138
0.099
0.046
0.036
Segment
2
0.137
0.065
0.030
0.042
0.136
0.084
0.026
0.023
Segment
3
0.132
0.071
0.024
0.030
0.138
0.075
0.025
0.020
Segment
5
1.653
0.574
0.608
0.805
0.048
0.572
Segment
6
1.659
0.577
0.598
0.789
0.050
0.571
Segment
7
0.272
0.103
0.121
0.245
0.064
0.126
Segment
8
0.114
0.090
0.080
0.099
0.089
0.111
0.052
Segment
9
0.120
0.070
0.055
0.045
0.073
0.058
0.026
Segment
10
0.123
0.072
0.045
0.030
0.071
0.050
0.025
Table 3.3.5. Segment-Specific Average Surficial Sediment Total PCB Concentrations (ng/g)
All LMMBP Sediment Surficial Samples From Surficial Samples From
Segment Samples LMMBP Box Cores GBMBP Box Cores
11
12
13
15
16
17
56.2
35.2
4.99
17.1
127
52.9
103
63.4
27.9
695
643
07 3
190
-------
3.3.3 Model Confirmation
A thorough recalibration of MICHTOX was not
conducted after the enhancements were made to the
model as part of the preliminary assessment
modeling described in this chapter (Endicott, 2005).
However, the model predictions were confirmed
against data using two methods. Hindcast
simulations were produced to confirm the suitability
of and to establish confidence in MICHTOX model
parameters and model predictions over a long-term
model run. In addition, MICHTOX was run for the
1994-1995 LMMBP study period and model results
were compared to the LMMBP data. This section
contains discussion of the hindcast and comparison
to the LMMBP data for both of the MICHTOX
submodels.
3.3.3.1 Description of Hindcast Process
For the hindcast simulations developed by Endicott
(2005), estimated long-term trends were used to
develop the continuous total PCB forcing functions
from an uncontaminated initial condition in 1940.
Although somewhat speculative, a similar procedure
was demonstrated for PCBs in Lake Ontario
(Mackay, 1989; Gobas et al., 1995). The hindcast
forcing functions and model runs described in this
section were developed early in the LMMBP, before
the availability of the LMMBP sediment core data
used to develop the more rigorous hindcast forcing
functions (Part 1, Chapter 7) used for the LM2-Toxic
hindcast (Part 4, Chapter 5).
The development of hindcast forcing functions
required estimates of the date when contamination
began, the rate of increase in the magnitude of the
forcing function, the date and duration of the
loading/forcing function peak, and the rate of decline
in the magnitude of the forcing function. The
assumptions were that PCB contamination of Lake
Michigan commenced in 1940, the rate of increase in
vapor concentrations and tributary loadings was the
same as the rate of decline, atmospheric deposition
loadings followed the same long-term trends as
vapor concentrations, and monthly variability in the
magnitude of forcing functions followed the 24-month
pattern established by the LMMBP estimates.
Rates of change in vapor phase PCB concentrations
for Lake Michigan and the Great Lakes region have
been published by a number of researchers (Hillery
et al., 1997, 1998; Baker and Eisenreich, 1990;
Green et al., 2000; Schneider et al., 2001). Although
there is some disagreement as to whether
atmospheric measurements support the belief that
vapor phase PCB concentrations are declining over
Lake Michigan, Schneider et al. (2001) indicate that
PCB concentration profiles in highly-resolved
sediment cores from Grand Traverse Bay located in
northeastern Lake Michigan support the view that
vapor phase PCB concentrations have been
declining at a rate of about 0.115/year, which
corresponds to a six-year half-life, over the past 25
years.
Similarly, rates of change in PCB tributary loadings
were determined from loading estimates based upon
measurements for the Fox River in 1989-1990
(Velleux and Endicott, 1994) and from similar
estimates for other Lake Michigan tributaries for 1982
(Marti and Armstrong, 1990) and 1994-1995
(McCarty et al., 2004). This information yielded
estimates for the rate of decline in tributary loadings
of 0.053 to 0.054/year, corresponding to a 12- to 13-
year half-life.
The date and duration of the peak in the PCB forcing
function was not easily defined. Schneider et al.
(2001) suggested that forcing functions peaked in
1972 and declined with the decline in chemical
production after 1972 and a halt of production in
1977. On the other hand, Gobas et al. (1995)
estimated that PCB loadings to Lake Ontario peaked
in 1962. The reason for such a difference between
the lakes is unclear and perhaps reflects the
subjectivity of these estimates. Ultimately, three
different estimates for long-term hindcast total PCB
forcing functions were developed:
• Scenario A Total PCB forcing functions peaked
in 1970 and declined after 1972.
• Scenario B - Total PCB forcing functions peaked
in 1961 and declined after 1963.
• Scenario C - Total PCB forcing functions peaked
in 1961 and declined after 1972.
Plots of the three forcing function scenarios were
provided on a whole-lake basis for vapor
concentrations (Figure 3.3.3), atmospheric
191
-------
CO
E
a
0>
o
c
o
CJ
co
o
CL
CO
16
14 •
12 -
10 -
8 •
6 •
4
2
total PCB
vapor-phase
concentration estimates
- - - scenario A
scenario B
scenario C
0
1940 1950 1960
2000
1970 1980
year
1990 2000
Figure 3.3.3. Long-term estimates of
Michigan total PCB vapor concentrations.
Lake
^
"en
en
"
m
1800-
1600-
1400-
1200-
1000-
800-
600-
400-
200-
total PCB
tributary loading
estimates
— scenario A
scenario B
scenario C
0
1940 1950 1960
1970 1980 1990 2000
year
Figure 3.3.5. Long-term estimates of Lake
Michigan total PCB tributary loadings.
deposition (Figure 3.3.4), and tributary loadings
(Figure 3.3.5).
3.3.3.2 Hindcast Results
Hindcast simulations, commencing in 1940 from zero
concentration (i.e., clean) initial conditions for total
PCBs, were conducted for each of the three long-
term forcing function scenarios (Endicott, 2005). The
results of these simulations were compared to the
LMMBP data (1994-1995) and long-term PCB
concentration data with the goals of confirming
parameters and determining which long-term loading
scenario simulated the data most accurately.
noooo
total PCB
wet and dry
atmospheric estimates
- • scenario A
scenario B
- scenario C
1990 2000
Figure 3.3.4. Long-term estimates of Lake
Michigan total PCB atmospheric deposition
loadings.
Predictions for Scenario A were plotted and
compared to available data in Figures 3.3.6-3.3.9.
Predictions in the main Lake Michigan segments
were compared to whole lake volume-weighted
average water column concentrations in Figure 3.3.6.
Clearly, the model predictions of total PCB
concentrations in the water column were low for this
scenario. Predicted surficial sediment total PCB
concentrations were also low in comparison to the
data (Figure 3.3.7) (see Figure 3.3.2 for sediment
core locations). The same tendency of Scenario A to
underpredict total PCBs concentrations occurred with
en
c
O
c
03
O
C
o
o
CD
O
Q.
scenario A
predictions
segment 1
— segment 2
segment 3
segment 8
segment 9
segment 10
a water data
1940
2000
Figure 3.3.6. Long-term Scenario A predictions
of main lake total PCB concentrations.
192
-------
-S*
"5>
c^
c.
o
c
8
c
o
o
03
o
Q.
onn -,
GUU
250-
200-
150"
100-
50-
A -
0
19
3
OC
scenario A predictions
r- A si rv, n _, * H -4
- segment 12
— segment 13
n site 18 field data
A site 47s field data
o site 68k field data
C
0 D
o o n
n D
<-, O A
1920 1940
O
0°AAA
O
o .0
A
A
' D Aa D DL^
A
/ ^
/ *v
t/'""^Vv
A A y/ * '<^>-
-^^~*2*^' "-'<
1960 1980 2000
year
Figure 3.3.7. Comparison of long-term Scenario
A predictions to main lake sediment total PCB
concentrations (sediment cores collected in
1991-1992).
-3 6000-
— 5000-
c:
O
"•5 4000-
§ 3000-
c
8 2000-
CD
CL 1000-
0-
Sheboygan Reef
scenario A PCB
n average
HI model
fti fte 1 fl
1
I
I
CO
(D
Figure 3.3.9. Comparison of MICHTOX Scenario
A total PCB concentrations to Sheboygan Reef
data.
)
o
c
03
0
C
o
o
CD
O
Q_
25
20-
15-
10-
5-
n
scenario A
lake trout predictions
- ----- 3np 1 0
aye i u
2qQ Q
age 8
---••-; age 7
0 GLNPO data
\D
0 0
o
o
0
-------
0.3
=J 0.25 -
en
c:
c 0.2
g
1o
~ 0.15
"Bi
c
0}
o
c
o
o
CO
o
CL
20-
10-
5-
\o o
scenario B
lake trout predictions
age 10
age 9
age 8
•-.-•r. ..- age 7
O GLNPO data
0
1940 1950 1960
1970
year
1980 1990 2000
Figure 3.3.14. Comparison of long-term Scenario
B predictions to GLNPO lake trout data.
194
-------
7000
S6000-
"cn
— 5000
g
'"Si 4000 H
-»->
g 3000
c
°2000
CO
o
a. 1000
Sheboygan Reef
scenario B PCB
[] average
HI model
o
CD
^
CO
cn
co
o
0>
CO
o
-------
300-r
250-
200-
4= 150-
c
(D
O
C
o
o
CD
o
Q_
100-
50
scenario C predictions
segment 11
- - - segment 13
D site 18 field data
A site 47s field data
o site 68k field data
on
0 0
0-
1900
1920 1940 1960 1980 2000
year
Figure 3.3.17. Comparison of Scenario C
predictions to main lake sediment total PCB
concentrations (sediment cores collected in
1991-1992).
3
2
CD
JO
CO
cn
co
CD
CO
cn
co
CD
JO
in
0)
cn
co
jo
co
(D
cn
co
CD
JB
N-
CD
cn
co
CD
JO
co
CD
cn
co
CD
JO
O)
CD
cn
CO
I UUU -
'36000-
^>
£-5000-
c
0
To 4000-
[U ' '-' ** "
"c
aa 3000-
o
c
8 2000-
03
CL 1000-
n.
Sheboygan Reef
scenario C PCB
D average
10 model
rh I^W
rk II
\
i
I
j.
rf,
T
^
1
I
1
J
Fl
b
1
I
CD
CO
CD
cn
CO
3
o
01
jg
CM
CD
cn
(0
Figure 3.3.19. Comparison of MICHTOX Scenario
C total PCB concentrations to Sheboygan Reef
fish data.
25-t
f 20-
1 15-
PCB concentre
D cn o
19
0 0
O
0
„ _ — ^^-^^
1 1 1
40 1950 1960 1970 1
scenario C
lake trout prediction
age 10
- - - age 8
.: -.- age 7
0 GLNPO data
O
i 1 r— '
980 1990 200C
s
year
Figure 3.3.18. Comparison of long-term Scenario
C predictions to GLNPO lake trout data.
In general, the Scenario B long-term simulations tend
to agree most favorably with the available PCB data.
The model predictions for this scenario were
probably at least as accurate as the forcing functions
themselves; this was judged to be an adequate level
of model confirmation for this assessment. However,
further refinement of forcing functions and model
parameters could improve the agreement between
data and predictions.
3.3.3.3 Comparison to the LMMBP Data
For the most recent study, MICHTOX simulations
were conducted for the 1994-1995 LMMBP period as
a check on model performance and in preparation for
the forecast scenarios discussed in Part 3, Chapter
4.
The total PCB forcing functions used for this model
run included both monitored and unmonitored
tributary loads and the revised atmospheric
deposition estimates. They were calculated following
the same procedure used for the LM2-Toxic forcing
functions described in Section 4.4.3.1 and Section
4.6.3. For comparison to model-predicted water
column PCB concentrations, the LMMBP cruise
sample data were extrapolated using the natural
196
-------
neighbor method and aggregated to the MICHTOX
water column segments (Appendix 4.4.1).
Differences between model predictions and sampling
data concentrations were evident, although the
differences were probably reasonable for a
screening-level model application. Modeled total
PCB concentrations for the main lake epilimnetic
segments were consistently lower than observed
data (Figure 3.3.20). Model-predicted main lake
epilimnetic total PCB concentrations averaged over
the 1994-1995 period were 30-40% lower than
sample data average concentrations.
Predicted hypolimnetic PCB concentrations were
important as these were passed to the MICHTOX
food chain submodel as exposure concentrations.
Modeled main lake hypolimnetic total PCB
concentrations exhibited substantial seasonal
variation, but the data showed little seasonal
variation (Figure 3.3.21). While the baseline
concentrations predicted by the model were in
general agreement with the observed data, the
seasonal peaks resulted in study-period averaged
total PCB concentrations from the model being 50%
higher in southern Lake Michigan, 7% higher in
central Lake Michigan, and 25% higher in northern
Lake Michigan.
The results of the model simulations for the 1994-
1995 period were also interrogated in terms of total
PCB mass transport fluxes and inventories (Figures
3.3.22-3.3.23). Mass balance diagnostics
demonstrated that air-water fluxes clearly dominated
the transport pathways for PCBs in Lake Michigan.
The net volatilization (gross volatilization minus gas
absorption) of 3,078 kg/year was the most significant
net gain or net loss to the system. Particulate settling
and resuspension resulted in large amounts of PCBs
cycling between the water column and sediments.
The general agreement between model and data,
coupled with the hindcast simulation conducted in the
Endicott (2005) study, confirmed the applicability of
MICHTOX for screening-level modeling assessments
of PCBs in Lake Michigan. A robust calibration of the
model should be conducted if it is desired to use
MICHTOX in a future management context. The
results from this study suggested that important
areas for future review included MICHTOX
parameterization related to partitioning,
settling/resuspension, and volatilization/absorption.
Further confirmation of the model is presented in the
following discussion of the MICHTOX food chain
submodel, which uses the fate and transport
submodel water-column and sediment PCB output as
exposure concentrations.
0.4
£>0.3H
c
o
're
I °-2
o
o
o
gO.1-
CL
0'
0.4
Northern Lake Michigan (segment 3)
en
c
g
CD
0.3-
-------
1.0
Northern Lake Michigan (segment 10)
0.8-
c
o
0.6-
0
1.0
Central Lake Michigan
(segment 9)
hypolimnetic total PCB
model
o cruise data
0.8-
c
o
0.6-
c
03
y
o
o
CO
O
Q.
0.4-
0.2-
Southern Lake Michigan (segment 8)
Jan
1994
Jan
1995
Jan
1996
during model development reduced the errors
associated with model conceptualization and coding.
Errors associated with model parameterization and
forcing functions can be examined through an
uncertainty analysis.
Uncertainty in MICHTOX model predictions had been
examined through both conventional Monte Carlo
analyses (Endicott et ai, 2005) and Bayesian Monte
Carlo (BMC) analyses (Endicott, 2005). The
uncertainty analyses were conducted on a steady-
state version of the model because of run time
limitations. Both analyses were conducted before the
availability of the LMMBP data; however, the results
should be representative. The BMC analyses
indicated that total PCB concentrations should be
well within a factor of two of model predictions
(Endicott, 2005). It is possible that repeating the
BMC analysis with the LMMBP data and forcing
functions would result in smaller confidence intervals
for prediction, and less uncertainty.
References
Ambrose, R.B., T.A. Wool, J.P. Connolly, and R.W.
Schanz. 1988. WASP4, a Hydrodynamic and
Water Quality Model - Model Theory, User's
Manual and Programmer's Guide. U.S.
Environmental Protection Agency, Office of
Research and Development, Environmental
Research Laboratory, Athens, Georgia.
EPA/600/3-87/039, 297 pp.
Baker, J.E. and S.J. Eisenreich. 1990.
Concentrations and Fluxes of Polycyclic Aromatic
Hydrocarbons and Polychlorinated Biphenyls
Across the Air-Water Interface of Lake Superior.
Environ. Sci. Technol., 24(3):342-352.
Figure 3.3.21. Comparison of MICHTOX
hypolimnetic total PCB concentrations to the
LMMBP cruise data.
3.3.4 Model Uncertainty
Model predictions contain uncertainty for numerous
reasons: conceptual errors and/or simplifications,
errors in parameterization, uncharacterized system
variability, and systematic errors in forcing functions
and calibration data. Peer review and quality control
Bamford, H.A., D.L. Poster, and J.E. Baker. 2000.
Henry's Law Constants of Polychlorinated
Biphenyl Congeners and Their Variation with
Temperature. J. Chem. Engin., 45:1069-1074.
DeVault, D.S., R. Hesselberg, P.W. Rodgers, and
T.J. Feist. 1996. Contaminant Trends in Lake
Trout and Walleye From the Laurentian Great
Lakes. J. Great Lakes Res., 22(4):884-895.
198
-------
gross volatilization
5387
gas absorption
2309
traits of Mackinac
export
-2
atmospheric
deposition
wet deposition 226
dry deposition 781
water column = 1511 kg
sediment = 27,045 kg
sediment
burial
984
379
monitored and unmonitored
tributary loading
(Lake Michigan watershed)
Figure 3.3.22. MICHTOX predicted mass balance fluxes and inventories (kg/year) for 1994-1995, whole
lake results.
Green Bay
monitored and
unmonitored
tributary loading
atmospheric 226-
deposition"
wet deposition
dry deposition
Green Bay
volatilization
804
Green Bay
gas absorption
70
Green Bay
sediment burial
53
Green Bay Mass Budget
atmospheric
deposition
wet deposition 212
dry deposition 531
main
gas absorption
2240
gross
volatilization
4583
Green Bay
export
resuspension
2966
Straits of Mackinac
export
-2
PCB Inventory
Main Lake:
water column = 1411 kg
sediment = 18,749 kg
main lake monitored and
unmonitored tributary loading
(Lake Michigan watershed
excluding Green Bay)
153
Green Bay:
water column = 100 kg
sediment = 8,296 kg
931
Figure 3.3.23. MICHTOX predicted mass balance fluxes and inventories (kg/year) for 1994-1995, Green
Bay and main lake results.
199
-------
Endicott, D.D. 2005. 2002 Lake Michigan Mass
Balance Project: Modeling Total Polychlorinated
Biphenyls Using the MICHTOX Model. In: R.
Rossmann (Ed.), MICHTOX: A Mass Balance
and Bioaccumulation Model for Toxic Chemicals
in Lake Michigan, Part 2. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
EPA/600-R-05/158, 140 pp.
Endicott, D.D., W.L. Richardson, and DJ. Kandt.
2005. 1992 MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan. In: R. Rossmann (Ed.),
MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan, Part 1. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
EPA/600-R-05/158, 140pp.
Gobas, F.A.P.C., M.N.Z. Graggen, and X. Zhang.
1995. Time Response of the Lake Ontario
Ecosystem to Virtual Elimination of PCBs.
Environ. Sci. Techn., 29(8):2038-2046.
Green, M.L., J.V. DePinto, C.W. Sweet, and K.C.
Hornbuckle. 2000. Regional Spatial and
Temporal Interpolation of Atmospheric PCBs:
Interpretation of Lake Michigan Mass Balance
Data. Environ. Sci. Technol., 34(9): 1833-1 841.
Hillery, B.L., I. Basu, C.W. Sweet, and R.A. Hites.
1997. Temporal and Spatial Trends in a Long-
Term Study of Gas-phase PCB Concentrations
Near the Great Lakes. Environ. Sci. Technol.
Hillery, B.L, M.F. Simcik, I. Basu, R.M. Hoff, W.M.J.
Strachan, D. Burniston, C.H. Chan, K.A. Brice,
C.W. Sweet, and R.A. Hites. 1998. Atmospheric
Deposition of Toxic Pollutants to the Great Lakes
as Measured by the Integrated Atmospheric
Deposition Network. Environ. Sci. Technol
32( 1 5) :22 16-2221.
Mackay, D. 1989. Modeling the Long-Term Behavior
of an Organic Contaminant in a Large Lake:
Application to PCBs in Lake Ontario. J. Great
Lakes Res., 15(2):283-297.
Manchester-Neesvig, J.B, A.W. Andren, and D.N.
Edgington. 1996. Patterns of Mass
Sedimentation and of Deposition of Sediment
Contaminated by PCBs in Green Bay. J. Great
Lakes Res., 22(2):444-462.
Marti, E.A. and D.E. Armstrong. 1990.
Polychlorinated Biphenyls in Lake Michigan
Tributaries. J. Great Lakes Res., 16(3):396-405.
McCarty, H.B., J. Schofield, K. Miller, R.N. Brent, P.
Van Hoff, and B. Eadie. 2004. Results of the
Lake Michigan Mass Balance Study:
Polychlorinated Biphenyls and frans-Nonachlor
Data Report. U.S. Environmental Protection
Agency, Great Lakes National Program Office,
Chicago, Illinois. EPA/905/R-01/011, 289 pp.
Schneider, A.R., H.M. Stapleton, J. Cornwell, and
J.E. Baker. 2001. Recent Declines in PAH, PCB,
and Toxaphene Levels in the Northern Great
Lakes as Determined from High Resolution
Sediment Cores. Environ. Sci. Technol.,
35(19):3809-3815.
Schwarzenbach, R.P., P.M. Gschwend, and D.M.
Imboden. 1993. Environmental Organic
Chemistry. John Wiley and Sons, Incorporated,
New York, New York. 681 pp.
Velleux, M.L. and D. Endicott. 1994. Development of
a Mass Balance Model for Estimating PCB Export
from the Lower Fox River to Green Bay. J. Great
Lakes Res., 20(2):416-434.
Wanninkhoff, RJ. 1992. Relationship Between Gas
Exchange and Wind Speed Over the Ocean. J
Geophys. Res., 97:7373-7381.
200
-------
PARTS
LEVEL 1 MODELS
Chapter 4.
Modeling
MICHTOX Food Chain
MICHTOX is a toxic chemical mass balance and
bioaccumulation model. The model was used at the
beginning of the Lake Michigan Mass Balance
Project (LMMBP) as a planning tool (Endicott et al.,
2005). After the LMMBP data were collected, the
model provided preliminary mass balance modeling
assessments for polychlorinated biphenyls (PCBs) in
Lake Michigan (Endicott, 2005). For the present
phase of the LMMBP, MICHTOX was used to provide
a screening-level analysis of the effects of various
pollutant loading sources on bioaccumulation in Lake
Michigan lake trout, and to predict the length of time
until PCB concentrations in lake trout declined below
health advisory target levels under different forecast
scenarios. Chapter 4 discusses this aspect of the
MICHTOX application. MICHTOX also provided a
comparison of an established "off-the-shelf" model to
the more complex Level 2 models developed during
the LMMBP.
MICHTOX contains both a fate and transport
submodel and a food chain bioaccumulation
submodel. This chapter provides information on the
application of the MICHTOX food chain
bioaccumulation submodel to Lake Michigan. The
food chain submodel is briefly described, with a
complete description included in Endicott et al.
(2005). The model coefficients and data are also
briefly described. The remainder of the chapter
includes a discussion of seven model scenarios that
were conducted as a screening-level assessment of
the fate and sources of PCBs in the system, as a
preliminary evaluation of the potential range of future
PCB concentrations in Lake Michigan under different
possible loading scenarios, and for a comparison to
predictions of the Level 2 models.
3.4.1 Model Development
MICHTOX was originally developed in the early
1990s as a screening model for the Lake Michigan
Lake-wide Management Plan (LaMP) and for
development of the LMMBP (Endicott et al., 2005).
The model was later updated with newer process
formulations, parameters, and the LMMBP data
(Endicott, 2005), as described in Part 3, Chapter 3.
The screening-level food chain bioaccumulation
modeling for Part 3 was completed using the food
chain submodel of MICHTOX. This submodel was
essentially unchanged from the original version
(Endicott et al., 2005), except for adaptations to the
program code to accommodate 62-year model runs
for the forecast modeling.
The MICHTOX food chain bioaccumulation submodel
was adapted from version 3.20 of the Manhattan
College Food Chain Model, which was based upon
the WASTOXv4 food chain model (Connolly and
Thomann, 1985; Connolly, 1991). It used the time-
variable water column dissolved and particulate PCB
concentrations output from the MICHTOX fate and
transport submodel as the PCB exposure
concentrations for the trophic levels of the food chain.
The MICHTOX food chain submodel was applied
separately to paired water column and sediment
segment output from the fate and transport submodel
for each area of interest. As with the fate and
transport submodel, the food chain submodel
201
-------
simulated total PCBs as the sum of two homologs:
tetrachlorobiphenyl (PCB4) and pentachlorobiphenyl
(PCBS).
MICHTOX treats bioaccumulation as a chemical
mass balance within individual organisms, and a
bioaccumulation differential equation was solved for
each individual age class of organism (Equation
3.4.1 ) (Endicott ef a/., 2005):
= kuicfd
E
- K',v,
where
/ = the organism of interest
/ = the prey organism
v, = chemical concentration in organism i
(Mcherr/Mwet)
kui = uptake rate (L3/T/Mwet)
c = chemical concentration in water (M/L3)
fd = dissolved chemical fraction in the water
column
n
Pij - feeding preference factor £ (p^ = 1 ) of
organism i for organism j y=i
a/y = chemical assimilation efficiency across gut
Cv = food consumption rate (Mprey,W9t/MprediW9t)
K'; = chemical elimination rate (1/T)
In general, PCB concentrations in an organism was
equal to the sum of the PCB uptake from water
(across the gill) and from consumption minus the
PCB concentrations lost through elimination
(excretion and dilution through growth). Equations
for the consumption rate, PCB uptake rate, and PCB
excretion rate are fully described in "1 992 MICHTOX:
A Mass Balance and Bioaccumulation
Model for Toxic Chemicals in Lake Michigan"
(Endicott, 2005).
For phytoplankton, PCB accumulation was assumed
to be a partitioning process, assuming 2% organic
carbon composition on a wet weight basis. The PCB
concentrations in detritus, the food source for benthic
organisms, was assumed to be equal to that of the
surficial sediment.
3.4.2 Description of the Data Used in
MICHTOX Food Chain
3.4.2.1 Description of Data
Fish and lower food chain organism data used for
MICHTOX modeling were collected in three biota
zones (Figure 3.4.1) (McCarty et al., 2004). The
biota zones were geographical areas on the lake
chosen to compare and contrast the fish population
characteristics in different regions of Lake Michigan.
0,-
Figure 3.4.1. The LMMBP biota sampling zones.
202
-------
The Saugatuck biota zone is located on the eastern
side of Lake Michigan (MICHTOX water column
segments 1 and 8). The Sturgeon Bay biota zone is
located east of the Door Peninsula (MICHTOX water
column segments 2 and 9). The Sheboygan Reef
biota zone in southern Lake Michigan is located just
south of the segment partition between MICHTOX
water column segments 1 and 2. The Sheboygan
Reef data are only shown for information purposes.
Because forage fish were collected in a substantially
different location than the lake trout at this site and
were possibly not representative of prey items on the
reef, the Sheboygan Reef data were not used for
model confirmation. Tables 3.4.1, 3.4.2, and 3.4.3
show the biota data for their respective zones. In the
higher trophic levels, it can be seen that Saugatuck
organisms had a consistently higher amount of total
PCB concentrations than organisms at other
locations (Figure 3.4.2).
The representation of the Lake Michigan food chain
in the MICHTOX food chain modeling included five
organisms: phytoplankton, Mysis, Diporeia, alewife,
and lake trout.
3.4.2.2 Sources and Choice of Constants
For the most recent study, LMMBP data were used
to update alewife growth rates and biota zone-
specific lipid concentrations. In addition, biota zone-
specific growth rates and biota zone-specific lipid
concentrations were used for lake trout. The age-
and species-specific weight, growth rate, and lipid
concentrations for all organisms are shown in Table
3.4.4. Table 3.4.5 includes the food assimilation
efficiencies and the chemical assimilation coefficients
for the organisms and PCB homologs used in the
MICHTOX food chain model.
3.4.3 Model Confirmation
In previous work with the MICHTOX model and the
LMMBP data (Endicott, 2005), a hindcast
confirmation of the MICHTOX fate and transport
submodel and the food chain submodel was
conducted to establish confidence in the model and
model parameters. The hindcast simulations of both
submodels were discussed in Part 3, Chapter 3.3.
The food chain organism weights, specific growth
rates, and lipid concentrations were subsequently
Table 3.4.1. Average Total PCB Concentrations in Fish in the Saugatuck Biota Zone
Species
Age (Years)
Average PCB
Concentrations (ng/g)
PCB Standard Deviation
(ng/g)
Alewife < 120 mm
Alewife > 1 20 mm
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
1-2
3-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
304
592
175
904
883
1287
2068
3185
3609
4511
5728
8209
7477
8116
6666
6799
4014
167
140
171
288
241
532
1126
809
921
1645
4101
2515
2997
872
794
3268
203
-------
Table 3.4.2. Average Total PCB Concentrations in Fish in the Sheboygan Reef Biota Zone
Species
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Age (Years)
3
4
5
6
7
8
9
11
12
13
14
Average PCB
Concentrations (ng/g)
547
706
1202
1395
1974
2668
3102
5322
4692
4466
3483
PCB Standard Deviation
(ng/g)
184
217
204
192
320
1001
1022
1215
1234
217
Table 3.4.3. Average Total PCB Concentrations in Fish in the Sturgeon Bay Biota Zone
Species
Alewif e < 1 20 mm
Alewife > 120 mm
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Lake Trout
Age (Years)
1-2
3-7
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Average PCB
Concentrations (ng/g)
170
589
350
395
889
1268
1707
2487
2656
3360
4211
5283
5939
4420
4324
5254
7192
PCB Standard Deviation
(ng/g)
71
171
163
107
159
270
309
577
509
559
757
1168
1543
1185
438
1345
683
204
-------
Table 3.4.4. MICHTOX Food Chain Age- and Species-Specific Weight, Growth Rate, and Lipid
Concentrations
MYSIS
Age
1
2
3
4
Weight (g)
0.00021
0.0022
0.0081 1
0.01977
Growth
Rate (1/day)
0.0193
0.0107
0.0073
0.0056
Lipid %
4
4
4
4
DIPOREIA
Age
1
2
3
Weight (g)
0.00007
0.00300
0.00630
Growth
Rate (1/day)
0.00398
0.00203
0.00313
Lipid %
3
3
3
ALEWIFE
Age
1
2
3
4
5
6
7
Weight (g)
3
15
27
37
45
50
53
Growth
Rate (1/day)
0.00441
0.00161
0.00086
0.00054
0.00029
0.00016
0.00010
Average
Lipid %
Sheboygan
Reef
7.2
8.5
9.0
10.5
11.5
12.0
12.2
Average
Lipid %
Saugatuck
5.5
5.5
6.0
7.5
9.0
10.0
11.0
Average
Lipid %
Sturgeon
Bay
4
6
6
6
6
6
6
STOCKED TROUT
Age
1
2
3
4
5
6
7
8
9
10
11
12
Weight (g)
Sheboygan
Reef
20
128
244
490
900
1378
1900
2600
3400
4000
4400
4700
Growth
Rate (1/day)
Sheboygan
Reef
0.005082
0.001766
0.001909
0.001665
0.001166
0.000879
0.000859
0.000734
0.000445
0.000261
0.000181
0.000114
Lipid %
Sheboygan
Reef
2.30
3.66
7.90
9.36
12.48
15.56
18.60
19.36
19.34
19.10
20.73
22.40
Weight (g)
Saugatuck
90
180
550
1100
2050
2850
3400
4000
4500
5400
6500
6900
Growth
Rate
(1/day)
Saugatuck
0.001898
0.003058
0.001898
0.001704
0.000902
0.000483
0.000445
0.000322
0.000499
0.000508
0.000164
0.000078
Lipid %
Saugatuck
2.30
3.66
7.13
9.52
14.77
18.96
21.05
18.56
19.12
20.68
22.00
23.00
Weight
(9)
Sturgeon
Bay
98
120
350
800
1500
2700
3200
3700
4400
5000
5500
5600
Growth
Rate
(1/day)
Sturgeon
Bay
0.000554
0.002931
0.002263
0.001721
0.001609
0.000465
0.000397
0.000474
0.000350
0.000261
0.000049
0.000096
Lipid %
Sturgeon
Bay
4.80
4.68
9.21
11.81
17.04
18.30
19.13
20.52
20.15
22.63
22.50
20.53
205
-------
9000-
8000-
7000-
6000-
5000-
4000-
3000-
2000-
1000-
0-
J3
El Saugatuck
• Sheboygan Reef
D Sturgeon Bay
R n
] n
u
1
1
..•-L-P
H
i;| f~
u
*•
ll
^^
I
in
ll
^1
il
LMJ
I
1
=
'.
I
I
^
-
T
I
|
1
?
f
\
8
1
f
1
~]
,£ E
'5 E s E
_0)0 .0)0
03 (N TOtN
lake trout age class
Figure 3.4.2. Total PCB concentrations of organisms in Lake Michigan biota zones.
Table 3.4.5. MICHTOX Food Chain Model Parameters and Coefficients
Parameter
Value (Unitless)
Mysis Assimilation of Ingested Food
Diporeia Assimilation of Ingested Food
Alewife Assimilation of Ingested Food
Stocked Lake Trout Assimilation of Ingested Food
Mysis Chemical Assimilation (PCB4)
Diporeia Chemical Assimilation (PCB4)
Alewife Chemical Assimilation (PCB4)
Stocked Lake Trout Chemical Assimilation (PCB4)
Mysis Chemical Assimilation (PCB5)
Diporeia Chemical Assimilation (PCB5)
Alewife Chemical Assimilation (PCB5)
Stocked Lake Trout Chemical Assimilation (PCB5)
0.400
0.0288
0.800
0.800
0.800
0.405
0.800
0.600
0.575
0.165
0.575
0.600
updated as described in Section 3.4.2. Estimates of
unmonitored tributary PCB loads were also added.
The hindcast model runs were repeated with the new
food chain parameterization, and the results were
again in general agreement with the available data.
While previous applications of MICHTOX used
hindcast model runs to confirm model performance
and evaluate past loading (Endicott et al., 2005;
Endicott, 2005), the most recent application of the
model focused on forecast model runs. For the latest
study, the model performance representing Lake
Michigan food chain PCB dynamics was confirmed
by comparing forecast model predictions to lake trout
data collected from the Saugatuck biota zone. The
forecast confirmation model run was compared
against the same Saugatuck biota zone historical
lake trout data (1970 to 2002) as the hindcast runs
previously described (DeVault et al., 1996;
Swackhamer, 2003). For these studies, lake trout
were collected in a specific size range with an
average weight of 2,600 grams, which is
approximately the average weight of the five year-old
and six year-old lake trout at Saugatuck (Table 3.4.4)
collected during the LMMBP. The average of the
five year-old and six year-old lake trout will be
referred to as "5.5 year-old" lake trout. The
MICHTOX model was set up with Constant
Conditions for the 1994-2050 time frame: vapor
concentrations, tributary loadings, and atmospheric
206
-------
loads were repeated from those measured during
1994-1995. The original food chain model
coefficients were found to give an acceptable fit
between model and the 5.5 year-old lake trout total
PCB data (Figure 3.4.3). Chemical assimilation
efficiencies were adjusted to see if a better fit could
be obtained, but no substantial improvement was
obtained.
25
20
= 15
10
age 5.5 Saugatuck lake trout
• total PCB data
— MICHTOX
1970 1980 1990 2000 2010 2020 2030 2040 2050 2060
year
Figure 3.4.3. Total PCB concentrations in 5.5
year-old lake trout at Saugatuck biota zone.
3.4.4 Results - Forecast Scenarios
The ability to forecast future pollutant concentrations
based upon changes in pollutant loadings is one of
the most useful capabilities of models. MICHTOX
was used to forecast the reduction in total PCB
concentrations in the Lake Michigan food chain,
especially those trophic levels that would impact
human health risk by consumption such as lake trout.
The forecast simulations in Sections 3.4.4 and 3.4.5
were run for 62 years, from January 1, 1994 through
December 31, 2055. Measured LMMBP PCB
concentrations were used to define initial conditions
in water, sediment, and fish (McCarty et al., 2004).
All simulations used the 1994-1995 forcing functions
for the first two years of the model run. The forcing
functions were determined from measured values of
the LMMBP and included atmospheric vapor
concentration, wet and dry atmospheric deposition
loads, and monitored and unmonitored tributary
loads. These forcing functions were the same as
those used in the MICHTOX model run for Section
3.3.3.3 and were calculated in the same manner as
the congener-specific functions of the LM2-Toxic
model. The calculation procedures are described in
Sections 4.4.3.1 and 4.6.3. After January 1, 1996,
forcing functions were varied according to the
specified conditions of the forecast scenario.
The scenario results were evaluated against a fish
advisory consumption guideline target level for the
total PCB body burden in lake trout. The Protocol for
a Uniform Great Lakes Sport Fish Consumption
Advisory (Great Lakes Sport Fish Advisory Task
Force, 1993) derived a target concentration in the
edible portion of lake trout of 0.05 ppm. For
comparison to the LMMBP model output, this value
was converted to a whole fish concentration of 0.075
ug total PCBs/g fish (Appendix 3.4.1). Model output
for the average of five and six year-old lake trout
("5.5 year-old fish") was selected for evaluation
against the target level to be consistent with the size
range of fish in the long-term data set (DeVault et al.,
1996).
In this section, MICHTOX was applied to three
loading scenarios. The first scenario assumed
constant loads at the 1994-1995 level. The second
and third scenarios had decreasing loads based upon
decline rates observed in the literature. The results
of the scenario simulations for Saugatuck are
displayed in Figures 3.4.4a and 3.4.4b, and the
Sturgeon Bay results are displayed in Figures 3.4.5a
and 3.4.5b. The figures include an expanded
concentration scale to allow a comparison of
predicted total PCB concentrations to the 0.075 ug
total PCBs/g fish consumption advisory target
concentration. While the model results are
referenced to data from the Saugatuck and Sturgeon
Bay biota zones, MICHTOX has relatively coarse
segmentation and the model results only represent
the southern and middle sections of Lake Michigan.
3.4.4.1 Conditions Remain the Same as 1994-
1995 (Constant Conditions)
For this scenario, 1994-1995 forcing functions
(tributary loads, atmospheric deposition loads, and
atmospheric vapor concentrations) were assumed to
remain constant from January 1, 1994 through
December 31, 2055. The two-year cycle of forcing
functions was repeated for the entire 62-year period
of the scenario. This scenario provides insight into
the equilibrium status of the system to the 1994-1995
conditions, but it likely overestimates future
207
-------
— constant conditions
slow recovery
— fast recovery
-i 1 r
1990 2000 2010 2020 2030 2040 2050 2060
Figure 3.4.4a. Sensitivity scenario predicted total
PCB concentrations in 5.5 year-old lake trout
from Saugatuck biota zone.
1.0-
•° 06-
a
c:
-------
Total PCB concentrations in lake trout were predicted
to decrease at an exponential rate (Figures 3.4.4a
and 3.4.5a). Total PCB concentrations in 5.5 year-
old lake trout were predicted to be less than the
0.075 ug total PCBs/g fish consumption advisory
target concentration in the year 2025 for Saugatuck
and 2018 for Sturgeon Bay (Figures 3.4.4b and
3.4.5b).
Recent studies (Buehler et at., 2002) have suggested
that historical rates of decline have recently slowed,
and thus this scenario, while realistic, may
overestimate the future rate of decline.
3.4.4.3 Continued Recovery - Slow
This scenario also simulated the system response to
declines in PCB loads and atmospheric
concentrations, but assumed that a slower observed
decline rate of PCB loadings (Section 1.7.2) will
continue for the 62-year simulation period. Forcing
functions were assumed to decrease from 1994-1995
levels at a 20-year half-life for atmospheric
components (vapor phase PCB concentrations and
wet and dry atmospheric deposition loadings) and a
13-year half-life for PCB tributary loadings).
Total PCB concentrations in lake trout again declined
exponentially but at a slower rate than in the previous
scenario (Figures 3.4.4a and 3.4.5a). Total PCB
concentrations in 5.5 year-old lake trout were
predicted to take 28 years longer to reach the
consumption advisory target concentration at
Saugatuck. The 0.075 ug total PCBs/g fish target
concentration was achieved in the year 2053 for
Saugatuck and 2037 for Sturgeon Bay (Figures
3.4.4b and 3.4.5b).
The atmospheric vapor and deposition decline rate
selected for this scenario was on the conservative
side of possible rates, and the predicted dates of
compliance with the consumption advisory target are
likely the upper bound of possible dates.
3.4.5 Model Sensitivity
Four model sensitivity runs were conducted to
analyze the importance of different total PCB loading
sources to the Lake Michigan system. Each
sensitivity run eliminated one or more loading
sources to determine the impact of the sources on
PCB concentrations in the water and lake trout. The
remaining loading sources repeated the 1994-1995
values, similar to the constant conditions scenario.
The loading sources removed included all tributary
loads, atmospheric deposition loads, the combination
of tributary and atmospheric deposition loads, and
internal loads from the sediment. Results were
displayed for the Saugatuck area, but the trends
were similar for all areas of the main lake.
3.4.5.1 No Atmospheric Wet and Dry Deposition
Loadings
This simulation analyzed the sensitivity of MICHTOX
to the elimination of atmospheric wet and dry
deposition loadings. The first two years of the
simulation used the 1994-1995 forcing functions.
After January 1, 2005, the atmospheric deposition
loadings were set to zero for the remaining 60 years
of the model run. The remainder of the forcing
functions repeated the two-year 1994-1995 values.
The 5.5 year-old lake trout total PCB concentrations
decreased at a rate considerably faster than the
Constant Conditions Scenario and achieved a
significantly lower steady-state concentration (Figure
3.4.6). The scenario with no atmospheric deposition
was predicted to reach an approximate steady-state
total PCB concentrations in lake trout of 0.30 ug/g,
compared to the constant conditions steady-state
concentration of 0.47 ug/g.
3.0
^2.0
o
-£ 1.5-
<§ 1.0
CO
o
2=0.5-
constant conditions
no atmospheric deposition
no tributary loading
— .no atmospheric deposition
or tributary loads
—— clean sediment
0
1990 2000 2010 2020 2030 2040 2050 2060
Figure 3.4.6. Sensitivity scenario total PCB
concentration predictions for 5.5 year-old lake
trout at Saugatuck.
209
-------
3.4.5.2 No Tributary Loadings
This simulation analyzed the sensitivity of MICHTOX
to the elimination of total PCB tributary loads to the
Lake Michigan system. The first two years of the
simulation used the 1994-1995 forcing functions, but
after January 1,2005, the tributary loadings were set
to zero for the remainder of the model run. All other
forcing functions repeated the two-year 1994-1995
values.
Compared to the Constant Conditions Scenario, the
5.5 year-old lake trout total PCB concentrations
declined faster and reached a lower steady-state
concentration, but the reduction was small (Figure
3.4.6). The scenario with no tributary loadings
reached an approximate steady-state total PCB
concentrations of 0.42 ug/g, compared to the
constant conditions steady-state concentration of
0.47 ug/g.
Based upon these results, the system is predicted to
have a greater sensitivity to wet and dry atmospheric
deposition loadings than to tributary loadings. This
result was not surprising, because MICHTOX treats
tributary loadings and atmospheric deposition
loadings in the same manner and the atmospheric
deposition loadings were more than 2.6 times greater
than tributary loadings in 1994-1995.
However, the relative magnitudes of tributary and
atmospheric deposition loads were confounded by
the available load estimate methodologies. The
atmospheric deposition loads included estimates of
the coarse particle loads, which were not directly
measured. While comprising a large portion of the
total load, the estimate of the coarse particle load is
only approximate and may be subject to significant
error.
The results of this scenario do not suggest that
tributary loadings are not important. While they have
a relatively small impact when looking at Lake
Michigan on a large scale, such as MICHTOX does,
tributary loadings have a large impact on the local
receiving waters which they enter. Tributary loadings
and their watershed sources may also have a
significant effect on atmospheric vapor
concentrations and deposition to the lake. Thus,
clean-up of watershed PCB sources may have an
effect on loadings which are not directly quantified by
the water quality model.
3.4.5.3 No Atmospheric Deposition and No
Tributary Loadings
This simulation combined the removal of tributary
loadings and atmospheric deposition loadings. The
first two years of the simulation used the 1994-1995
forcing functions. After January 1, 2005, the
atmospheric deposition loadings and the tributary
loadings were set to zero. Other forcing functions
repeated the 1994-1995 values throughout the
simulation period.
The 5.5 year-old lake trout total PCB concentrations
decreased at a rate and achieved a steady-state
concentration only slightly below those of the
scenario with only atmospheric deposition loads
removed (0.25 ug/g versus 0.30 ug/g) (Figure 3.4.6).
As with the previous two scenarios, this suggested
that atmospheric components had a greater effect on
MICHTOX-predicted total PCB concentrations in
Lake Michigan lake trout than tributary loadings.
3.4.5.4 Sediment Total PCB Concentration Initial
Conditions Set to Zero
This scenario was conducted to evaluate the
sensitivity of the total PCB concentrations in lake
trout to the reservoir of total PCBs in the sediment of
Lake Michigan. Measured LMMBP PCB
concentrations were used to define initial conditions
in water, sediment, and fish (McCarty et al., 2004).
The 1994-1995 forcing functions were repeated for
the entire period of the model run. On January 1,
1996, the sediment total PCB concentrations were
re-set to zero, after which the model simulation was
allowed to run normally for a 60-year period.
The shape of the lake trout PCB concentration
curves overtime was influenced by PCB dynamics in
the water column and sediments and a time lag in the
MICHTOX food chain. Water column concentrations
dropped for two years as PCBs in the water column
settled out and volatilized faster than loadings
entering the system. After two years, however, water
column concentrations began to recover due to
tributary and atmospheric deposition loadings,
resuspension of newly contaminated sediments, and
absorption of PCBs from the atmosphere.
210
-------
While the lower levels of the food chain immediately
responded to the reduced PCBs in water and
sediment, the higher food chain organisms had a
residual body burden, and the response was slower.
MICHTOX predicted a drop in the 5.5 year-old lake
trout total PCB concentrations for a period of six
years, then the concentrations steadily increased
until reaching the same steady-state concentration as
the Constant Conditions Scenario (Figure 3.4.6).
The lake trout total PCB concentrations were within
5% of the concentrations of the Constant Conditions
Scenario within a period of 30 years, and reached
steady-state concentration about 45 years after the
sediment clean-up. This was about the same time
period required for the Constant Conditions Scenario
to reach steady-state concentrations.
This scenario demonstrated the importance of the
reservoir of total PCBs in the sediments on the total
PCB concentrations in the higher levels of the food
chain. Response time would be greatly influenced by
the sediment settling and resuspension dynamics in
the model.
References
Buehler, S.S., I. Basu, and R.A. Hites. 2002. Gas-
Phase Polychlorinated Biphenyl and
Hexachlorocyclohexane Concentrations Near the
Great Lakes: A Historical Perspective. Environ.
Sci. Technol., 36(23):5051-5056.
Connolly, J.P. and R.V. Thomann. 1985. WASTOX,
A Framework for Modeling the Fate of Toxic
Chemicals in Aquatic Environments. Project
Report. U.S. Environmental Protection Agency,
Office of Research and Development, ERL
Duluth, Large Lakes Research Station, Grosse
lie, Michigan. 52 pp.
Connolly, J.P. 1991. Documentation for Food Chain
Model, Version 4.0. Manhattan College,
Riverdale, New York.
DeVault, D.S., R. Hesselberg, P.W. Rodgers, and
T.J. Feist. 1996. Contaminant Trends in Lake
Trout and Walleye From the Laurentian Great
Lakes. J. Great Lakes Res., 22(4):884-895.
Endicott, D.D. 2005. 2002 Lake Michigan Mass
Balance Project: Modeling Total Polychlorinated
Biphenyls Using the MICHTOX Model. In: R.
Rossmann (Ed.), MICHTOX: A Mass Balance
and Bioaccumulation Model for Toxic Chemicals
in Lake Michigan, Part 2. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
EPA/600/R-05/158, 140 pp.
Endicott, D.D., W.L. Richardson, and D.J. Kandt.
2005. 1992 MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan. In: R. Rossmann (Ed.),
MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan, Part 1. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
EPA/600/R-05/158, 140 pp.
Great Lakes Sport Fish Advisory Task Force. 1993.
Protocol for a Uniform Great Lakes Sport Fish
Consumption Advisory. 86 pp.
McCarty, H.B., J. Schofield, K. Miller, R.N. Brent, P
Van Hoff, and B. Eadie. 2004. Results of the
Lake Michigan Mass Balance Study:
Polychlorinated Biphenyls and frans-Nonachlor
Data Report. U.S. Environmental Protection
Agency, Great Lakes National Program Office,
Chicago, Illinois. EPA/905/R-01/011, 289 pp.
Swackhamer, D. 2003. Personal communication.
University of Minnesota, Madison, Wisconsin.
211
-------
PART 3
LEVEL 1 MODELS
Appendix 3.4.1 Derivation of a
Hypothetical Lake Michigan Lake Trout
Fish Consumption Criteria for PCBs
Brent Burman
Welso Federal Services, LLC
Large Lakes Research Station
9311 Groh Road
Grosse lie, Michigan 48138
and
Kenneth R. Rygwelski
U.S. Environmental Protection Agency
Office of Research and Development
National Health and Environmental Effects
Research Laboratory
Mid-Continent Ecology Division
Large Lakes and Rivers Forecasting Research
Branch
Large Lakes Research Station
9311 Groh Road
Grosse lie, Michigan 48138
The hypothetical fish consumption criteria that we
derived for Lake Michigan lake trout has not been
officially adopted by any Federal, State, or local
authorities. We proceeded to calculate this target
concentration because we could not find any other
polychlorinated biphenyl (PCB) consumption
concentrations for Lake Michigan lake trout that we
could readily relate to our model-predicted
concentrations in whole fish. This fish consumption
criteria was used in our hypothetical PCB forecast
scenarios for Lake Michigan found elsewhere in this
report.
A reduction factor to convert whole fish PCB
concentrations to fillet PCB concentrations was
needed for the comparison of model output data to
fish consumption advisories. The data gathered for
the Lake Michigan Mass Balance Project (LMMBP)
for lake trout PCB concentrations was based on the
whole body of the fish, less the stomachs. Available
consumption advisories are based on the edible
portion of the fish (Great Lakes Sport Fish Advisory
Task Force, 1993). The edible portion refers to a
fillet which includes all flesh from the back of the
head to the tail, including the skin and fatty belly flap.
The Health Protection Value for this fillet portion has
been established at 0.05 ug PCBs/(kg/day) and is
sufficient to keep cancer incidence at less than 1 per
10,000. Using a standard of 225 meals per year, the
amount of PCBs allowable in the edible portion for a
70 kg individual is 0.05 ppm. The derivation of this
figure is described in the Protocol for a Uniform Great
Lakes Sport Fish Consumption Advisory (Great
Lakes Sport Fish Advisory Task Force, 1993).
Research was necessary to determine if PCBs are
concentrated equally in the fillet and whole body of
lake trout. If it was determined that PCBs are
concentrated either higher or lower in the fillet than in
the whole fish, then the target level of 0.05 ppm
would have to be adjusted accordingly.
Limited research was performed on Lake Michigan
lake trout in this regard. Relevant research was done
on Lake Superior lake trout (Miller and Schram,
2000) and Lake Michigan rainbow trout and coho
salmon (Amrhein etai, 1999), but nothing was found
that specifically looked at Lake Michigan lake trout.
Amrhein era/. (1999) found whole fish:fillet derived
212
-------
ratios of 2.47 for rainbow trout, and 2.7 for coho
salmon, but it was uncertain as to whether the ratio
would be similar for lake trout. Miller and Schram
(2000) found a whole fish:fillet ratio for siscowet lake
trout of 2.5, but the lipid content of siscowet and Lake
Michigan lake trout are vastly different. A data set of
lean lake trout from Lake Superior was provided by
the Michigan Department of Environmental Quality
(MDEQ) which had PCB concentrations for both
whole fish and fillets (Day, 1997). The fish in the
data set provided were nearly identical in length,
weight, and lipid content, so we believed that they
would be sufficient to use in our comparison to the
Lake Michigan data set (Table A3.4.1).
Statistical analysis of the MDEQ data set resulted in
a whole fishrfillet ratio of 1.525; that is, PCB levels
were found to be 1.525 times higher in the whole fish
than in the fillets. Based upon this calculation, we
concluded that a factor of 1.5 would be justified in
converting the target level of 0.05 ppm PCBs for
fillets to 0.075 ppm PCBs as the new target level for
comparison of our model output for whole fish.
Further examination of the MDEQ data set revealed
that the whole fish:fillet PCB ratio was closely related
to the whole fish:fillet lipid ratio. A ratio of 1.50 was
found for the lipid concentrations and is shown in
Figure A.3.4.1. The result was not surprising as it is
known that PCBs are concentrated in the lipids.
Further research was initiated to validate the factor of
1.5 for lake trout in light of the work of Amrhein et al.
(1999). Coho salmon and rainbow trout were both
found to have much higher ratios, which if used for
lake trout, would raise the target level for a fish
consumption advisory to 1.25 ppm of PCBs, or even
higher. To add validity to the factor we had
calculated, a review of several common Great Lakes
sport fish was conducted. Because the PCB ratio in
question was shown in Figure 3.4.1 to be closely
related to lipid concentrations, a comparison was
made between eight Great Lakes sport fish using
additional data. The lake trout from the MDEQ data
set and the rainbow trout and coho salmon from
Amrhein et al. (1999) were compared with five
additional species of fish from the Fox River and
Green Bay. The additional species examined were
carp, walleye, northern pike, smallmouth bass, and
yellow perch (Fox River Model Evaluation
Workgroup, 1999). It was found that as whole fish
lipid concentrations decreased, that the relative ratio
of whole fish:fillet PCBs increased (Figure 3.4.2). It
can be reasoned that fish with higher lipid contents
store more lipids in the fillet portion than
comparatively less fatty fish, which store most of their
lipids in the viscera and head which are not included
in the edible portion. Because Lake Michigan lake
trout have higher lipid contents than the rainbow trout
and coho salmon studied by Amrhein et al. (1999),
they will also have a relatively lower whole fish:fillet
PCBs ratio.
References
Amrhein, J.F., C.A. Stow, and C. Wible. 1999.
Whole-Fish Versus Fillet Polychlorinated Biphenyl,
Concentrations: An Analysis Using Classification
and Regression Tree Models. Environ. Toxicol.
Chem., 18(8): 1817-1823.
Day, R. 1997. Michigan Fish Contaminant
Monitoring Program Annual Report. Surveillance
Water Quality Division, Michigan Department of
Environmental Quality, Surface Water Quality
Division, Lansing, Michigan. Report Number
MI/DEQ/SWQ-97-125.
Fox River Model Evaluation Workgroup. 1999.
Analysis of Bioaccumulation in the Fox River.
Technical Memorandum Document, Number
8600B6A.001 1001 0299 DN08, 24 pp.
Table A3.4.1. Comparison of the LMMBP Lake Trout to MDEQ Lake Superior Lake Trout
Length (cm) Weight (g)
Whole Fish PCBs
Lipid
Lake Superior (MDEQ)
Lake Michigan (LMMBP)
58.1
57.83
1519
1943.34
0.24075
2.03646
15.63
16.07
213
-------
2.5
2.0-
co
o
1.5
Lake Trout
PCB
3 lipid
1.52 - average PCB ratio
1.50 - average lipid ratio
o
.c
5
1.0-
0.5-
50.8 51.1 51.1 51.8 52.1 52.3 55.9 56.9 57.4 57.9 58.7 59.9 61.0 61.2 62.5 62.7 63.2 63.5 66.0 66.0
individual fish (by length - cm)
Figure A3.4.1. Whole fish to edible portion of fish PCBs and lipid ratios for lake trout.
18.00%
_g
"ro
CD
O
CL
-C
CO
JD
O
PCB ratio
—A— lipid content
lake
trout
carp
rainbow
trout
coho
salmon
walleye
northern
pike
T r
smallmouth
bass
-0.00%
yellow
perch
fish species
Figure A3.4.2. Comparison of whole fish to fillet PCB ratios and lipid content for various fish species.
214
-------
Great Lakes Sport Fish Advisory Task Force. 1993. Miller, M.A. and S.T. Schram. 2000. Growth and
Protocol for a Uniform Great Lakes Sport Fish Contaminant Dynamics of Lake Superior Lake
Consumption Advisory. 86 pp. Trout. J. Great Lakes Res., 26(1 ):102-111.
215
-------
PART 4
LM2-TOXIC
Xiaomi Zhang
Welso Federal Services, LLC
Large Lakes Research Station
9311 Groh Road
Grosse He, Michigan 48138
Chapter 1. Executive Summary
As one of the components in the overall Lake
Michigan Mass Balance Project (LMMBP) modeling
framework, a comprehensive polychlorinated
biphenyl (PCB) congener-based water quality model,
LM2-Toxic, was developed to simulate fate and
transport of PCBs in both water and sediment of
Lake Michigan. The main focus of this model was to
address the relationship between sources of toxic
chemicals and their concentrations in water and
sediments of Lake Michigan, and provide the PCB
exposure concentrations to the bioaccumulation
model (LM2 Food Chain) to predict PCB
concentrations in lake trout tissue. This report
provides detailed model description and
development, model input and field data, model
calibration procedures and confirmation, PCB mass
budget analysis, the results of model predictions, and
sensitivity analyses.
LM2-Toxic is a revision of the United States
Environmental Protection Agency (USEPA)-
supported WASP4 water quality modeling framework.
It incorporates the organic carbon dynamics featured
in GBTOX and the sediment transport scheme, a
quasi-Lagrangian framework, used in the IPX. Both
GBTOX and IPX were WASP4-type models and
major components in the Green Bay Mass Balance
Project (GBMBP) modeling framework. Another
important modification was the addition of updated
air-water exchange formulations to the model.
There were 94 segments in the spatial segmentation
for the LM2-Toxic. Forty-one of them were water
column segments, and 53 of them were surficial
sediment segments. Temporal resolution for the
model input was on a daily time scale. Most of the
kinetic functions were segment-specific time
functions. Good representation of water circulation
was essential for the accuracy of outputs from the
water quality model. The results at 5 x 5 km2 grid
generated by Princeton Ocean Model (POM) for the
Great Lakes were linked to the transport fields for
LM2-Toxic. Due to an affinity of PCBs for organic
carbon, three organic carbon sorbents were
simulated as state variables in LM2-Toxic. They
were biotic carbon (BIC), particulate detrital carbon
(PDC), and dissolved organic carbon (DOC). The
model simulated 54 PCB congeners which accounted
for roughly 70% of the total PCB mass in Lake
Michigan. Four phases were simulated in LM2-Toxic
for the congeners. The four phases were dissolved,
sorbed to PDC, sorbed to BIC, and bound to DOC.
LM2-Toxic is a coupled mass balance of organic
carbon solids and toxic chemical (PCBs) dynamics.
Prior to the organic carbon dynamics and PCB
216
-------
dynamic calibrations, vertical dispersion coefficients
were determined using a thermal balance model.
Using the LMMBP-generated field data, the organic
carbon solids dynamics were first calibrated. This
was followed by the independent calibration of PCB
dynamics. The temporal variations of both BIG and
PDC resulted from an algal bloom in late spring and
early summer. Primary production was the dominant
organic carbon load to Lake Michigan. The
eutrophication model (LMS-Eutro)-generated primary
production accounted for over 90 percent of the total
paniculate organic carbon (POC) load to the lake.
The PCB concentrations in the dissolved phase was
about double the concentration in the particulate
phase in the main lake. There was some degree of
temporal variation in the water column PCB
concentration controlled by a combination of
seasonal variation of external loads, atmospheric
concentrations, and sediment resuspension events.
There was also a slight longitudinal concentration
gradient throughout the main lake. The highest
concentrations were found in the southern segments
due to higher PCB atmospheric deposition and
concentrations observed in the area close to
Chicago. There was little vertical gradient of PCB
concentrations found based on main lake cruise
mean data. As an important part of the modeling
effort to reduce uncertainties associated with water
transport, settling, resuspension, and sedimentation,
a chloride model, a long-term simulation using a
137Cs and 239'240pu model, a long-term organic carbon
simulation, and a 47-year PCB hindcast simulation
using LM2-Toxic were developed and run for LM2-
Toxic confirmation. These confirmation steps were
crucial and laid down a credible foundation for long-
term projections using LM2-Toxic.
After calibration of organic carbon and PCB congener
dynamics and model confirmation, a mass budget
analysis was done for the LMMBP period (1994-
1995) to identify the critical contaminant sources,
sinks, and key environmental processes in Lake
Michigan. Figure 4.1.1 provides a summary of the
results of the total PCB mass budget diagnosis in
Lake Michigan. The average masses of total PCBs
presented in the water column and the surficial
sediments (0-1 cm) of the lake during 1994-1995
were 1,216 kg and 13,085 kg, respectively. The
inventories divided by the volumes of the water and
the surficial sediment layer of the lake lead to an
average concentration of total PCBs equal to 0.259
ng/L in the water column and 12,037 ng/L in the
surficial sediment layer. The information on the
fluxes of total PCBs in Figure 4.1.1 shows the single
largest flux leaving the Lake Michigan system was
gross volatilization. This flux was countered by the
flux from gas absorption as the largest source to the
lake. The air-water exchange was the most
important process for Lake Michigan. It produced the
largest PCB net loss out from the lake.
Resuspension was a major influx of PCBs to the
water column offset by the flux from settling.
Resuspension and settling were very important
processes in the lake system. The results of these
processes made the net flux between resuspension
and settling the second largest net source. The total
external load (tributary loads + atmospheric loads) to
the water column of the lake was the largest net PCB
source to the lake water column. The flux by burial
was the largest net loss from the surficial sediment
layer. There was a net loss of 1,863 kg/year of total
PCBs for the entire Lake Michigan system (the water
column + the surficial sediment layer of both Green
Bay and Lake Michigan). This indicated both the
water column and the surficial sediment layer of the
lake were not at steady-state during the LMMBP
period.
The model was also applied for forecasting the long-
term responses (62-year simulation, starting on
January 1, 1994) of the PCBs in Lake Michigan
under various forcing functions and load reduction
scenarios. Seven PCB forecast and sensitivity
scenarios were conducted. These seven long-term
forecast and sensitivity scenarios were:
A. Constant Conditions (Upper Bound) - Repeat
1994-1995 conditions.
B. Continued Recovery (Fast) - Atmospheric
components (vapor phase concentration, dry and
wet depositions) decline with a six-year half-life
(Hillery et al., 1997; Schneider et at., 2001).
Tributary loads decline with a 13-year half-life
(Endicott, 2005; Marti and Armstrong, 1990).
C. Continued Recovery (Slow) - Atmospheric
components (vapor phase concentration, dry and
wet depositions) decline with a 20-year half-life
(Buehler et al., 2002). Tributary loads decline
with a 13-year half-life.
217
-------
volatilization
3439
gas absorption
1507
atmospheric
deposition
980
.input from
Lake h(uron
V. 4
resuspension
export to
Lake Huron
12
PCB Inventory kg
water column = 1216
active sediment = 13,085
(0-4 cm interval)
monitored and unmonitored
tributary loading
(Lake Michigan watershed)
Chicago,
River
export)
8
sediment
burial
1284
Figure 4.1.1. Mass budget average for 1994-1995 total PCBs in the Lake Michigan system (including
Green Bay). Unit of the masses transported (arrows) is in kg/year.
D. No Atmospheric Deposition - Stop all PCB
atmospheric deposition (dry and wet).
E. No Tributary Loadings - Stop all tributary PCB
loads.
F. Lake Sediment Clean-up - Remove all PCB
mass from the surficial sediments.
G. No Atmospheric Deposition and No Tributary
Loadings - Stop all PCB atmospheric deposition
(dry and wet) and tributary loads.
All of the control actions started on January 1,1996.
Figure 4.1.2 presents the long-term responses of
total PCBs in the water column of Lake Michigan
under the seven forecast and sensitivity scenarios.
The results from the Constant Conditions Scenario
simulation clearly demonstrated that, during the
LMMBP period, the Lake Michigan system was not at
steady-state with respect to the 1994-1995 loads,
vapor phase concentrations, and the level of
sediment total PCB inventory. The results from this
scenario also indicate that, if there is no decline in
the current (1994-1995) forcing conditions, the water
column PCB concentration in the future will never
meet the USEPA water quality criteria for the
protection of wildlife (0.12 ng/L) (U.S. Environmental
Protection Agency, 2005) and human health (0.026
ng/L) (U.S. Environmental Protection Agency, 1997)
in the Lake Michigan system. The long-term
response from Scenario B - Continue Recovery
(Fast) shows that it takes about five years for the
water column concentration to reach the USEPA
water quality criterion for the protection of wildlife and
more than two decades to reach the USEPA water
quality criterion for the protection of human health.
The water column PCB concentrations predicted in
218
-------
0.30
? 0.25-
c
•B
cc
§ 0.15H
"
CD
o
0.10-
0.05-
Constant conditions - scenario a
Continue recovery (slow) - scenario c
«™= Continue recovery (fast) - scenario b
» LMMBP data -1994-95; EEGLE data 2000
— • EPA water quality criteria for protection of
wildlife (2005)
EPA water quality criteria for protection of
human health (1997)
0
1994 2004 2014 2024
year
2034 2044
2054
0.30
c
.g
~m
•B 0.20-
| 0.15H
o
0 0.10-
00
O
°- 0.05-
No atmospheric deposition - scenario d
No tributary loadings - scenario e
Lakewide sediment cleanup - scenario f
—— No atmospheric deposition & tributary
loadings - scenario g
* LMMBP data - 1994-95; EEGLE data 2000
1994
2004
2014
2024
year
2034
2044
2054
Figure 4.1.2. Annual long-term responses of total PCB concentrations in the water column of Lake
Michigan for the forecast and sensitivity scenarios.
Scenario C - Continue Recovery (Slow) declined at
a much slower pace. The model results indicated
that it takes about 12 years for the water column PCB
concentrations in the lake to reach the USEPA water
quality criterion for the protection of wildlife and that
the water column PCB concentration will reach the
USEPA water quality criterion for the protection of
human health around 2046 (five decades after 1996).
The rates used in Scenarios B and C may not be
realistic rates for the Great Lakes in the future. With
the addition of more recent data, it appears that the
rate of decline in the atmospheric components could
be slower than the decline rate used in Scenario C.
The results from the sensitivity scenarios (Scenarios
D, E, F, and G) suggested that the long-term PCB
concentrations in the water column are more
sensitive to atmospheric deposition (dry and wet)
than to tributary loads. By eliminating PCB total
inventory in the lake sediments on January 1,1996,
the water column concentration had a steep drop
initially, and then gradually increased and reached a
value close to the steady-state concentration
predicted by Scenario A - Constant Condition.
LM2-Toxic is a sophisticated and state-of-the-art
toxic chemical fate and transport model for Lake
Michigan. There are still many improvements that
can be made to the modeling framework and more
systematic tests could be conducted to address the
impacts of each process conceptualized in the LM2-
Toxic on the model outcomes. The results and
219
-------
predictions from the LM2-Toxic clearly demonstrate
the ability of the model to quantitatively analyze the
behavior of PCBs in the Lake Michigan system and
to forecast the long-term PCB dynamics under
various external forcing conditions.
References
Buehler, S.S., I. Basu, and R.A. Hites. 2002. Gas-
Phase Polychlorinated Biphenyl and
Hexachlorocyclohexane Concentrations Near the
Great Lakes: A Historical Perspective. Environ.
Sci. Techno!., 36(23):5051-5056.
Endicott, D.D. 2005. 2002 Lake Michigan Mass
Balance Project: Modeling Total PCBs Using the
MICHTOX Model. In: R. Rossmann (Ed.),
MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan, Part 2. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
EPA/600/R-05/158, 140 pp.
Hillery, B.L., I. Basu, C.W. Sweet, and R.A. Hites.
1997. Temporal and Spatial Trends in a Long-
Term Study of Gas-Phase PCB Concentrations
Near the Great Lakes. Environ. Sci. Technol.,
Marti, E.A. and D.E. Armstrong. 1990.
Polychlorinated Biphenyls in Lake Michigan
Tributaries. J. Great Lakes Res., 16(3):396-405.
Schneider, A.R., H.M. Stapleton, J. Cornwell, and
J.E. Baker. 2001. Recent Declines in PAH,
PCB, and Toxaphene Levels in the Northern
Great Lakes as Determined From High
Resolution Sediment Cores. Environ. Sci.
Technol., 35(19):3809-3815.
U.S. Environmental Protection Agency. 1997.
Revocation of the Polychlorinated Biphenyl
Human Health Criteria in the Water Quality
Guidance for the Great Lakes System. Federal
Register, October 9, 1997, Volume 62, Number
196. [DOCID:fr09oc97-9]. From the Federal
Register Online via GPO Access
[wais.access.gpo.gov].
U.S. Environmental Protection Agency. 2005. Water
Quality Guidance for the Great Lakes System.
Code of Federal Regulations, Title 40, Volume
21, Chapter 1, Part 132. http://www.access.
gpo.gov/nara/.
220
-------
PART 4
LM2-TOXIC
Chapter 2. Recommendations
LM2-Toxic was developed for its efficient conduct of
model calibrations and its capability to forecast long-
term impacts resulting from a variety of control
actions applied to an aquatic system. There are still
quite a few improvements that can be done to
enhance the credibility of the model predictions. The
recommendations for the further improvements and
tests are listed below.
1. The results from the long-term cesium hindcast
simulation should not be limited to comparison to
the water column concentration and total
sediment cesium inventory. They should include
comparisons with the available 137Cs sediment
core profiles. This will provide an additional
confirmation on the solid vertical transport
dynamics (settling, resuspension, and burial
rates) in Lake Michigan.
2. Instead of the empirical approach to estimate
sediment resuspension of each sediment
segment, a more sophisticated mechanistic
sediment transport model (e.g., SEDZL or
SEDZLJ), should be used to provide a more
accurate and realistic sediment resuspension.
3. Model verification (a post-audit) should continue
using the latest field data collected around Lake
Michigan to verify 1) parameters and rates used
in the model and 2) some of the conclusions
made from the long-term forecast scenarios.
4. As an extended confirmation process, 1)
comparison between net settling fluxes of organic
carbon generated from the model with available
sediment trap data, and 2) more research
conducted on carbonrchlorophyll a ratio including
its spatial variation in the lake are needed.
5. To investigate the potential impacts on the
outcomes of the model by critical environmental
processes and under different physical, chemical,
and meteorological conditions, more systematic
analyses are needed, including:
A. Sensitivity analysis of the fluxes across the
air-water interface by using different gas and
liquid transfer formulations.
B. Sensitivity analysis of model responses by
changing surficial sediment initial conditions.
C. Investigation of model responses in both
water column and sediment to various mass
fluxes across the sediment-water interface by
changing diffusion coefficient and/or mixing
length between the water column and surficial
sediment.
D. Investigation of potential impacts of ice-cover
and water surface elevation on the model
outcome.
E. Investigation of the impacts of carbon internal
loads on the results of LM2-Toxic PCB
hindcast.
F. Investigation of system responses to different
hydrodynamic transport fields.
221
-------
G. Sensitivity analysis of organic carbon decay
rates in the surficial sediment.
All of the above analyses should be conducted on
both short-term and long-term scales to see the
short-term and long-term effects on the model
simulation results.
6. Explore the effect of a huge resuspension
event on outcomes of the model hindcast and
forecast.
7. Addition of a benthic nepheloid layer (BNL) to
the model configuration to address the
importance, effects, and benefits of this
compartment on the overall organic carbon and
hydrophobic organic chemicals cycling in the
lake system.
8. Expand the model to more applications and
additional contaminants including a) mercury
and frans-nonachlor modeling for Lake
Michigan and b) finer model resolutions (e.g.,
Level 3) in both spatial and kinetic processes.
9. Couple an air model to the LM2-Toxic to
compute vapor phase concentration
dynamically for more accurate calculation of the
fluxes across the interface between air and
water.
10. Collect a higher density of sediment samples
from shallow high-energy areas of Lake
Michigan that will greatly enhance the
representativeness of carbon sorbent or toxic
chemical dynamics in these zones. These may
significantly influence sediment-water exchange
and toxic chemical dynamics in Lake Michigan.
11. Apply a different segmentation (still similar to
the spatial resolution as the current
segmentation used in LM2-Toxic) to the model
to 1) more efficiently utilize nearshore and
offshore data, 2) document nearshore and
offshore gradients observed in the data
collected for the Lake Michigan Mass Balance
Project (LMMBP) for most congeners, and 3)
investigate the impacts of using different
segmentations (even on the same spatial
resolution) on the model outcomes.
12. Conduct sensitivity analysis on potential
polychlorinated bipheny! (PCB) decay rates in
both water column and sediments of Lake
Michigan. The model currently assumes no
PCB decay in both compartments.
13. I nvestigate the uncertainty associated with the
selected parameters for specific processes
conceptualized in the LM2-Toxic using Monte
Carlo or other uncertainty analyses.
Considering the feasibility of these
recommendations, they can be categorized as
follows:
A. The recommendations that can be done in a
relatively short time period with limited effort
include Numbers 5A, B, C, G; and 12.
B. The recommendations that can be done within
a relatively moderate time frame and effort
once the necessary data are available include
Numbers 3; 4; 5D, E, F; and 6.
C. The recommendations that can be done in a
long time frame and significant effort include
Numbers 1, 2, 8, and 13.
D. The recommendations that can be done in a
very long time period and demand full effort
include Numbers 7, 9, 10, and 11.
222
-------
PART 4
LM2-TOXIC
Chapter 3. Model Description
As one of the submodels in the overall Lake Michigan
Mass Balance Project (LMMBP) modeling
framework, LM2-Toxic was specifically developed to
simulate transport and fate of toxic chemicals, such
as polychlorinated biphenyl (PCB) congeners, in both
the water and sediment of Lake Michigan. The
principal focus of this model was to quantitatively
define the relationship among external carbon and
toxic chemical loads, internal cycling of organic
carbon and toxic chemicals, and toxic chemical
concentrations in the water and sediments of the
lake. The PCB exposure concentrations predicted
from the LM2-Toxic model were then used as a
forcing function to compute PCB concentrations in
fish tissue.
LM2-Toxic evolved by combining the IPX and
GBTOX models in an attempt to better represent and
integrate processes considered to be important in
Lake Michigan (Velleux et at., 2000; Settles, 1997;
Bierman etal., 1992). Both the IPX and GBTOX are
descendants from the WASP4 water quality modeling
framework (Ambrose et a/., 1988). LM2-Toxic
incorporates the organic carbon dynamics highlighted
in GBTOX and the sediment transport scheme, a
quasi-Lagrangian framework, used in the IPX model.
It allows decay and transformation between organic
carbon states in both the water column and sediment
bed and variation of the surficial sediment layer
thickness in response to net settling or net
resuspension of sediments. In addition to the above
features, an updated air-water exchange formulation
(Bamford et ai, 1999; Wanninkhof et a!., 1991;
Schwarzenbach etal., 1993) was implemented in the
model.
4.3.1 Model Framework
LM2-Toxic was based on the principle of
conservation of mass. It used the same finite
segment modeling approach used in the United
States Environmental Protection Agency (USEPA)-
supported WASP4 modeling framework. The mass
of a chemical or solid in each segment of a water
system is controlled by water movement between
adjacent segments, solid dynamics and chemical
dynamics within the system, internal and external
loads, and boundary concentrations. A group of
mass balance equations representing the above
processes was used in the model to compute change
of mass for a state variable (constituent) in a
segment at a certain time. The model traced and
described where and how a mass of constituent was
transported and transformed. The general time-
dependent finite differential equation in a given
segment can be written to describe the change of
mass for a state variable at a certain time.
In a water column segment:
dC
(4.3.1.)
In a sediment segment (assume mixing only in
vertical direction):
223
-------
dt
where
Vj = volume of segment j (L3)
Cj = concentration of water quality constituent in
segment j (M/L3)
Cjj = concentration of water quality constituent at
the interface between segment i and j
(M/L3)
QJJ - net flow across the interface between
segment i and j (defined as positive when
entering segment j and negative when
leaving segment j) (L3/T)
n = number of adjacent segments
Ay =
AX,
W, =
'aw,/ -
, bulk dispersion/diffusion
coefficient (L3/T)
mixing (dispersion/diffusion) coefficient
(L2/T)
interfacial area between segment i and j
(L2)
characteristic mixing length between
segments i and j (L)
internal and external loading rate of
segment j (M/T)
mass change rate due to air-water
exchange process between segment j and
air directly above segment j (M/T)
mass change rate due to sediment-water
exchange processes between segment j
and adjacent sediment segments (M/T).
The processes include settling and
resuspension
mass change rate due to sum of kinetic
transformation processes within segment j
(M/T), positive is source, negative is sink
Sb = mass change rate due to burial process
from surficial mixing layer to deeper
sediment layer (M/T)
Note: L = length; M = mass; T = time.
The following sections of this chapter will provide a
detailed description of model segmentation, water
circulation, solid dynamics, and chemical dynamics.
The direct loads, including external and internal
loads, and boundary conditions are discussed in
detail in Chapter 4.
4.3.2 Model Configuration
4.3.2.1 Spatial Resolution - Segmentation
Compared to MICHTOX (Level 1 contaminant
transport and fate model developed for Lake
Michigan) segmentation (Figure 3.3.2), the LM2-
Toxic Level 2 model had finer resolution. Most water
column segments in the LM2-Toxic model
segmentation share the same or portion of the
segment boundaries used in the MICHTOX-Toxic
model. Green Bay segments had similar features as
those used in GBTOX (Bierman et a/., 1992) and
MICHTOX (Endicott et at, 2005). The significant
differences between Level 1 and Level 2 models
were with respect to model structure, state variables,
and related physical and chemical processes. They
are discussed in detail in the following sections.
The spatial segmentation for the LM2-Toxic model
was developed from digitized bathymetric (5x5 km2
grid) and shoreline data for Lake Michigan provided
by Dr. David Schwab, National Oceanic and
Atmospheric Administration (NOAA) (Schwab and
Beletsky, 1998). The lake, including Green Bay, was
divided into 10 horizontal columns, five water column
layers, and one surficial sediment layer. A detailed
spatial and cross sectional segmentation of the LM2-
Toxic model is illustrated in Figures 4.3.1 to 4.3.3.
There are 94 segments in total. Segments 1 -41 were
water column segments. Most of them had an
interface with surficial sediments. Segments 1-10
were surface water segments with an interface with
the atmosphere. Segments 42-94 were the surficial
sediment segments with sizes not identical to the
upper water column segments. The sediment
segmentation of the LM2-Toxic model was very
224
-------
5
15
24
33
40
10
8
9
15
19
3
13
22
31
38
10
10
10
19
80
1
11
20
29
36
10
10
10
18
42
4
14
23
32
39
10
10
10
20
84
2
12
21
30
37
10
10
10
18
48
6
16
25
34
41
10
10
9
16
21
water
column
segment
numbers
1
11
20
29
36
10
10
10
20
100
average
segment
thickness
in meters
Figure 4.3.1. Water column segmentation for the LM2-Toxic model.
225
-------
92
92
i
1
i
!
»
45
•ib
Jb
4b
4b
47
4/
J6
^
is
.*
i
V
i
•IS
•li
•se
at,
4b
47
4/
47
4E
48
47
Jb
Jt
«
15
>
47
4;
44
4B
4/
44
47
k/
4S
47
«/
47
47
-47
47
4/
f*
4£
8
B
H
i
^8
^
55
59
59
59
59
69
,B
56
56
54J
58
5U
46
47
4/
47
47
47
47
4J
41
1!
47
47
47
47
47
4/
4«
56
jS
58
59
59
59
5C
y
6C
6C
59
59
59
$e
58
58
5B
58
16
4/
47
«S
46
46
47
—
4!
4j
47
47
4/
47
41
41
8
59
BC
iC
6C
it
it
if
6_t
it
it
6t
be
58
59
58
58
58
M
*&
^6
46
atj
•Ib
•*E
fiO
47
47
4/
H16
»6
47
47
41
iO
iC
SO
it
Et
iC
iC
LC
S
it
>t
ea
50
58
58
58
5«
5U
Sfc
5E
16
46
Jb
Jb
4e
4b
4b
46
40
db
4b
16
4f
4/
^f
4!
>t
SI
6(
i(
i!
Bt
B(
>_(
6f
6C
Bl
59
50
if
58
8!
62
61
i-
!2
>1
61
5'
51
5.1
&1
61
'•'
,~,'
>1
bl
51
bu
4S
it
if
it
if
»f
Bt
6t
if
60
bt
>t
59
^B
58
59
i!
61
b1
J-
92
51
51
bl
51
51
bl
£
11
SI
5,
SI
49
bt
if
if
if
if
bt
5(
5(
31
61
6Q61
6Q6(
if
60
BU
61
59
!3
62
«2
S1
J-
B1
J
!2
SO
bU
50
Ml
iO
K
bu
5C
50
fjl
51
6C
50
bO
50
4S
i(
61
b'l
il
59
«2
62
01
i'
82
32
82
bO
bU
4!
4|
it
if
i1
i-
B
bllbl
idet
>1|61
>1B1
i1
i1
J1
3ilei 31
6lST61|61
6l!61'6161
51 >1 51
61
61
s
J2
g
61
H
>1
61
6'
61
61
i1
61
61
6^6*6'
6161
61
«2
i
02
82
5()
49
K
4<
4! 49
4945
48
49
50
bU
bt
bU
50
5C
bO
50
btl
50
)2
«2
82
82
61
51
— '
4!
4=
4<
4<
82
S2
62
62
02
61
6f
it
i1
6'
J1
W
6J
f-..
82
62
12
6161
sp
4;
49
4!
4!
4£
I49494_g
1434^49
<9434!
13494!
49494<
»P
134^
50
bO
50
bU
10
4i
50
50
X
50
4S
49
4S
50
50
5C
bt
4!
49
4S
4j
49
49
49
4!
4!
4!
4a
49
50
50
50
bl)
50
51
>1
61
i1
61
61
31|81
51
61
61
61
61.61
61J61
61
6161
61
61
61
b1
61
61
63
63
63
63
5
B
62
50
50
bt)
50
50
bu
50
5C
bO
bt
5U
50
5t
50
50
5U
4<
61
61
61
b1
8'
S3
D3
.f
&3
i3
bJ
al
13
W
D3
53
b
50
50
5C
bU
SO
61
60
50
61
61
Bl
61
H2
•=,i
63
M
64
M
»1
w
l.J
5
M
bl
51
M
W
SO
50
50
49
41
61
6'
>1
B1
61
63
63
bl
53
03
',1
50
50
50
4S
4P
4S
61
82
yi
02
bi
i3
tJ
03
SI
53
33
S3
52
SO
52
52
63
C3
13
6i
63
63
63
S3
a
H
JT
H53
H54
'5
:4 15
J4 14
; j 44
*
13
15
5
4
4
J
2
47
45
45
45
14
43
4)
43
43
47
47
I/
45
18
lj
14
4~4
4J
43
*3
13
4J
47
47
4/
4/
IS
Wj
IS
H.
«
43
43
43
43
4J
47
147
4/
47
47
ij
i?
4
(j
]
43
12
.'
41
47
47
47
47
47
47
45
45
44
4-1
4i
47
47
47
47
41
it
47
45
45
4
43
3
3
|
4!
«
47
4/
47
41
47
4/
47
45
45
45
47
43
43
43
41
41
41
\l
47
47
4>
47
47
17
5
5
44
3
3
3
41
4(
47
41
4V
4/
47
47
37
0
Ji
50
bO
if
JO
ic
SO
49
49
4!
4>
62
i2
gr
^5
-J
4<
4<
4!
4J
bO
50
)C
43
49
45
4!
52
52
JT
55
55
4<
4!
4!
4!
50
50
4!
4<
4<
4S
4<
52
^7
^-i
55
4<
4!
4!
4i
4!
50
4J
4E
4£
4S
b2
52
•i?
57
4«
4£
4J
4!-
4?
4(
4£
41
4!
52
5^
b2
»
55
4<
4<
4f
46
4<
49
4!
4J
45
4?
4<4<
494t
4<
55
52
S2
s
52
55
52
^
M
','.,
4349
4^49
4i
41
4i
4<
52
52
52
57
04
S4
SS
4J
W
s?
52
52
52
b2
M
SS
Sc.
4«
52
b2
S?
57
5^
V
'•4
•i?
52
b2
S?
Si
bJ
U
5262
•»?S4
52c-i
M M
*^ 55
3
Lake Michigan 5 Kilometer Grid
Level 2 Model
[#] depositional zones
(¥] transitional zones
0 non-depositional zones
# segment numbers
Figure 4.3.2. Surface sediment segmentation for the LM2-Toxic model.
226
-------
| | water column
[g] non-depositional area
|H transitional area
•I depositional area
Figure 4.3.3. Cross-sectional sediment segmentation and overlying water column segments for 10
Lake Michigan and four Green Bay water columns.
227
-------
different from the one used in MICHTOX. The
surficial sediment was differentiated into non-
depositional areas, transitional areas, and
depositional areas (Figure 4.3.2). The areas in a
surficial sediment segment were not necessarily
adjacent to each other. The surficial mixing sediment
layer was not uniformly distributed and its thickness
varied from 1 to 4 cm in transitional and depositional
areas.
The principal criteria considered for the LM2-Toxic
model water column segmentation were the
following:
1. Circulation patterns (Schwab and Beletsky,
1998).
2. Bathymetry (Schwab and Beletsky, 1998).
3. Horizontal and vertical gradients of temperature
and concerned constituent concentrations.
4. Comparability to MICHTOX segmentation and its
results.
The principal criteria considered for the LM2-Toxic
model surficial sediment segmentation were the
following:
1. Bathymetry data - 5 x 5 km grid (Schwab and
Beletsky, 1998).
2. Distribution of sediment characteristics (Cahill,
1981; Bobbins etal., 1999).
3. Results generated from the LMMBP sediment
measurements - box cores, ponar, and gravity
samples (Robbins et a/., 1999; Eadie and
Lozano, 1999).
Geometry-related data used in the LM2-Toxic model
such as volumes, surface areas, and average
thickness of all segments are listed in Tables 4.3.1
and 4.3.2.
4.3.2.2 Temporal Resolution
Two levels of temporal resolution for model inputs
and outputs were applied to LM2-Toxic. One was a
daily time scale that was used for the LM2-Toxic
calibrations. Another was a monthly time scale that
was used for the purpose of long-term forecasts to
be used as a lake management tool. Both temporal
resolutions were detailed enough for physical and
chemical processes in the lake that occur at monthly
or seasonal time scales, such as thermal
stratification, general water circulation patterns,
carbon internal loads (primary production), PCB
atmospheric concentrations, and resuspension. Final
interpretation of carbon and PCB mass budgets and
the results from long-term load reduction scenarios
on a lake-wide basis were at an annual time scale
using annually averaged results.
4.3.3 Water Balance
Water balance is one of the major components in a
traditional water quality modeling framework. Water
movement directly controls the transport of solids and
chemicals in dissolved and particulate phases in a
water system. In terms of LM2-Toxic model inputs,
the data in transport fields such as advective flows
and dispersive exchanges or mixing were used to
describe the water balance in the model. The
components and their sources used in LM2-Toxic
model transport fields are listed below:
1. Bi-direction horizontal advective flows (provided
by David Schwab, NOAA; originally based on
Schwab and Beletsky, 1998).
2. Net vertical advective flows (provided by David
Schwab, NOAA; originally based on Schwab and
Beletsky, 1998).
3. Tributary flows and flows across the Straits of
Mackinaw (Endicott etal., 2005; Quinn, 1977).
4. Water balancing flows.
5. Vertical dispersion coefficients.
Components such as precipitation, evaporation, and
groundwater infiltration were not considered in the
water transport fields used in the LM2-Toxic model.
Correct water circulation is essential for the accuracy
of outputs from the LM2-Toxic model. The Princeton
Ocean Model (POM) has been demonstrated for its
ability to accurately simulate water movement for a
given large water body (Schwab and Beletsky, 1997;
Blumberg and Mellor, 1987). Using an extensively
228
-------
Table 4.3.1. Geometry Data for Water Column Segments and Lake Michigan (Total)
Segment
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
Volume (m3)
1 .2729E+1 1
1 .3000E+1 1
1.1475E+11
8.6000E+10
4.0850E+10
3.8500E+10
1.1925E+10
1.1475E+10
1.3213E+10
1 .7250E+09
1.2123E+11
1 .2924E+1 1
1.1475E+11
8.6000E+10
3.1775E+10
3.7263E+10
8.4125E+09
7.7625E+09
6.4375E+09
1.1214E+11
1 .2303E+1 1
1.1444E+11
8.6000E+10
1.9625E+10
3.1988E+10
3.7625E+09
2.1500E+09
5.3750E+08
1.9198E+11
2.1649E+11
2.1339E+11
1.6908E+11
2.481 3E+10
4.2725E+10
2.8750E+08
3.6038E+1 1
4.6783E+1 1
7.8266E+1 1
6.8133E+11
1.6913E+10
3.4625E+10
Average
Thickness (m)
9.97
10
10
10
9.95
10
9.25
9.7
9.49
4.36
9.66
9.93
10
10
8.05
9.63
8.44
7.48
6.34
9.6
9.69
9.97
10
8.99
9.1
6.36
4.53
2.13
18.01
18.22
18.85
19.63
14.55
15.63
4.5
42.01
48.1
79.58
84.05
19.38
21.11
Surface Area (m2)
1.2750E+10
1.3000E+10
1.1475E+10
8.6000E+09
4.1000E+09
3.8500E+09
1 .2750E+09
1.1750E+09
1 .3750E+09
3.5000E+08
1.2500E+10
1.3000E+10
1.1475E+10
8.6000E-1-09
3.8500E+09
3.8500E+09
9.7500E+08
1.0000E+09
9.5000E+08
1.1625E+10
1.2650E+10
1.1475E+10
8.6000E+09
2.1500E+09
3.4750E+09
5.5000E+08
4.2500E+08
2.0000E+08
1.0600E+10
1.1825E+10
1.1275E+10
8.6000E+09
1 .6750E+09
2.7000E+09
5.0000E+07
8.4750E+09
9.6250E+09
9.8000E+09
8.0750E+09
8.5000E+08
1 .6000E+09
Water Segment
Above
1
2
3
4
5
6
7
8
9
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
29
30
31
32
33
34
Total
4.8148E+12
5.7950E+10
229
-------
Table 4.3.2. Initial Geometry Data for Surficial Sediment Segments and Surficial Sediment Layer (Total)
Segment
Number
42
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
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
Total
Volume (m3)
2.5000E+05
8.7500E+05
1.0250E+06
2.1250E+06
2.6975E+07
6.8850E+07
6.4375E+07
1 .8058E-I-08
7.6850E+07
2.1850E+07
1 .6500E+07
8.2500E+05
8.2500E+05
3.5000E+05
2.0000E+05
1 .4750E+06
4.9725E+07
2.1850E+07
1.3510E+08
1.0985E+08
5.0375E+07
3.5600E+07
5.2500E+06
2.5000E+05
1 .7000E+06
4.7500E+05
8.2500E+05
1.7500E+06
1 .2600E+07
4.5000E+06
2.0000E+06
6.8750E+06
7.7500E+06
1.1000E+06
7.7500E+05
3.7500E+05
7.0000E+06
2.3000E+07
1 .2000E+07
3.0000E+06
2.0000E+06
1.0000E+06
1.0000E+06
2.0000E+07
1.7000E+07
1.2000E+07
1 .7000E+07
2.0000E+07
5.0000E+06
1.3000E+07
1.2000E+07
1.0000E+06
1 .OOOOE+06
1.0871E+09
Average
Thickness (m)
0.001
0.001
0.001
0.001
0.013
0.018
0.025
0.031
0.029
0.019
0.012
0.001
0.001
0.001
0.001
0.001
0.013
0.019
0.028
0.026
0.031
0.016
0.010
0.001
0.001
0.001
0.001
0.010
0.024
0.030
0.010
0.011
0.010
0.001
0.001
0.001
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
Surface Area (m2)
2.5000E+08
8.7500E+08
1.0250E+09
2.1250E+09
2.0750E+09
3.8250E+09
2.5750E+09
5.8250E+09
2.6500E+09
1.1500E+09
1.3750E+09
8.2500E+08
8.2500E+08
3.5000E+08
2.0000E+08
1.4750E+09
3.8250E+09
1.1500E+09
4.8250E+09
4.2250E+09
1.6250E+09
2.2250E+09
5.2500E+08
2.5000E+08
1 .7000E+09
4.7500E+08
8.2500E+08
1.7500E+08
5.2500E+08
1.5000E+08
2.0000E+08
6.2500E+08
7.7500E+08
1.1000E+09
7.7500E+08
3.7500E+08
1 .7500E+08
5.7500E+08
3.0000E+08
7.5000E+07
5.0000E+07
2.5000E+07
2.5000E+07
5.0000E+08
4.2500E+08
3.0000E+08
4.2500E+08
5.0000E+08
1 .2500E+08
3.2500E+08
3.0000E+08
2.5000E+07
2.5000E+07
5.7950E+10
Water Segment
Above
1
11
20
29
36
36
36
37
37
37
30
30
21
12
22
31
38
38
38
39
39
39
32
5
15
24
33
40
40
40
41
41
41
34
25
16
8
18
27
27
27
35
35
26
17
7
9
19
19,28
19,28
10
10
10
230
-------
tested version of POM for the Great Lakes (POMGL),
transport fields were generated for Lake Michigan at
different spatial and temporal resolutions for use in a
series of mass balance models adapted for LMMBP
(Schwab and Beletsky, 1998). The hydrodynamic
model for Lake Michigan had 20 vertical levels and a
uniform horizontal grid size of 5 km (Schwab and
Beletsky, 1998). Because the LM2-Toxic model
segmentation was constructed based on the 5 x 5
km2 grid used in the POMGL for Lake Michigan, the
hydrodynamic model results were easily aggregated
to the resolution used in LM2-Toxic (Schwab and
Beletsky, 1998). The aggregated horizontal bi-
direction flows at each interface provided a good
approximation of horizontal advective and dispersive
transport components at the interface. The
advantage of using bi-direction flows at an interface
was that it bypassed the tedious and necessary
horizontal dispersion coefficient calibration procedure
required when only net flow is available at the
interface.
The vertical transport field was calculated in the form
of net vertical flow (provided by David Schwab,
NOAA; originally based on Schwab and Beletsky,
1998). Therefore, vertical exchange coefficients
were calculated and calibrated to define the vertical
mixing process between vertically adjacent
segments. A summer period of strong stratification
and a non-stratified period of intense vertical mixing
were the most important limnological features of the
Great Lakes (Chapra and Reckhow, 1983; Thomann
and Mueller, 1987). Therefore, determining the
dynamics of vertical mixing was considered as an
important task in modeling development for the
LMMBP. A thermal balance model was constructed
to calibrate the vertical exchange coefficients at the
interfaces (Zhang et al., 1998, 2000). The
coefficients were calibrated using 250 observed
vertical temperature profiles collected at 40 stations
in Lake Michigan during the 1994-1995 LMMBP
period. The model calibration results versus
temperature measurements in each water column
segment and temporal plots of calibrated exchange
coefficients are listed in Appendix 4.5.1. A detailed
discussion on how the thermal model was run can be
found in Part 4, Chapter 4.
Water balancing flow was another advective
component added into the water transport field for
LM2-Toxic. The aggregated advective flows
provided by NOAA were not balanced in individual
segments over the two-year LMMBP period.
However, the total water mass was perfectly
balanced on a whole lake basis. Over the two-year
LMMBP period, some segments lost or gained a
certain amount of water. This problem could be very
significant for long-term simulations for the LM2-
Toxic. It stops the model simulation once the volume
of a segment reached zero. To counter the amount
lost or gained in each segment, a balancing flow was
introduced to keep the volume of water unchanged in
each segment at any time during the simulation. The
balancing flows were generated based on the
aggregated advective flows (provided by David
Schwab, NOAA; originally based on Schwab and
Beletsky, 1998), original volume of each segment,
and the general water circulation patterns during the
LMMBP period.
Tributary flows and flows through the Straits of
Mackinac were based on MICHTOX model inputs
(Endicott et al., 2005), literature (Quinn, 1977), and
water circulation patterns during the LMMBP period
(provided by David Schwab, NOAA; originally based
on Schwab and Beletsky, 1998).
After vertical exchange coefficients were calibrated,
a conservative constituent, chloride, was simulated
using the LM2 model configuration to verify that the
water transport components described above were a
good representation of the overall water transport
field for the LM2-Toxic. The chloride model was run
just once without adjusting any parameter or
coefficient. The model results had very good
agreement with the observations during the LMMBP
period (Appendix 4.5.4). A detailed discussion of the
chloride results can be found in Part 4, Chapters 4
and 5.
4.3.4 Solid Balance
PCBs have an affinity for organic carbon. Each type
of organic carbon sorbs PCBs differently. Settling
rates and resuspension rates, decay rates, etc.,
impact the organic carbon fractions. Figure 4.3.4
shows the conceptual framework and the processes
related to organic carbon sorbent dynamics in Lake
Michigan. Three organic carbon sorbents were
simulated as state variables in LM2-Toxic. They
were biotic carbon (BIG), particulate detrital carbon
(PDC), and dissolved organic carbon (DOC). BIG
231
-------
Primary
WBIC Production
1 1
Advection In -
Water
Surface
Mixed
Sediment
Layer
~B
r-*' BIC
Net
Settling
HTC°2
BIC to ^
PDC
Wpoc
1
4-
PDC
Gross
Settling
>,
>
;
r
k
Primary
WDOC Production
1 1
PDC to ^
DOC
i-t
DOC
i
Resuspension
PDC
^TC°2
PDC to
DOC
\
k
^ DOC
i> Decay
Diffusion
r
DOC
DOC
> Decay
->• Advection Out
Burial >
BIG = Biotic Carbon; PDC = Particulate Detrital Carbon;
DOC = Dissolved Organic Carbon; W = Load
Figure 4.3.4. Conceptual framework of organic carbon sorbent dynamics used in the LM2-Toxic model.
represented participate organic carbon (POC) in live
phytoplankton biomass. PDC represented participate
detrital carbon derived from phytoplankton
decomposition, zooplankton excretion, and
allochthonous sources. DOC represented colloidal-
sized particles that pass through ashed 47 mm
diameter glass fibers (U.S. Environmental Protection
Agency, 1997).
Other than the components related to the water
transport field, numerous processes were considered
important in controlling the three organic carbon
solids concentrations in either the water column or
sediment or both (Table 4.3.3).
The segment-specific internal primary production
load generated from the eutrophication model (LM3-
Eutro) was a crucial input to the LM2-Toxic. Primary
production is the dominant organic carbon load to
Lake Michigan. LM3-Eutro generated primary
production accounted for over 90 percent of the total
POC load to the lake. Further discussions on
parameterization of processes such as organic
carbon sorbent decay and resuspension are
presented in the following sections. In Chapter 4,
detailed discussion of the input data used in the
processes related to organic carbon dynamics will be
provided.
4.3.4.1 Solid Kinetics
Decay of organic carbon sorbent was the only kinetic
process considered for solid dynamics in the LM2-
Toxic. This process transforms significant amounts
of carbon species in both the water column and
sediment segments. To capture general carbon
sorbent loss mechanisms in the water column without
introducing too much complication, pseudo-first-order
carbon decay rates for BIG, PDC, and DOC were
formulated as Michaelis-Menten functions of
respective carbon concentrations and then
temperature-corrected according to an Arrhenius
relationship with a temperature coefficient, 6 = 1.05
(Bierman et al., 1992). The decay equations for the
three organic carbon sorbents in water column are:
* 0(7- 20)
»0(7--20)
(4.3.3)
(4.3.4)
232
-------
Table 4.3.3. Processes Considered in Organic Carbon Sorbent Dynamics Constructed for the LM2-
Toxic
Biotic Carbon (BIC)
Particulate Detrital Carbon (PDC) Dissolved Organic Carbon (DOC)
External Tributary Loads
Internal Primary
Production Loads
Net Settling
Decay and Yield to PDC
External Tributary Loads
Yield From BIC Decay
Gross Settling
Resuspension
Decay and Yield to DOC in Both
Water Column and Sediments
Burial to Deeper Subsurface
Sediment Layer
External Tributary Loads
Internal Loads Derived From Primary
Production Loads
Yield From PDC Decay
Diffusion at Sediment-Water
Interface
Decay in Both Water Column and
Sediments
d(DOC)
^1/2(000)
'DOC
where
*0(T-20)
(4.3.5)
kd = decay rate of a carbon sorbent in water
column (d~1)
k = substrate saturated decay rate of a carbon
d sorbent in water column at 20°C (d~1)
k1/2 = Michaelis-Menten half-saturation constant for
a carbon sorbent in water column (mg C/L);
where C = carbon, L = liter
C = segment-specific concentration of a carbon
sorbent in water column (mg C/L); where C =
carbon, L = liter
Q = Arrhenius temperature coefficient
(dimensionless)
T = measured segment-specific temperature (°C)
Therefore, the final decay rates for the carbon
sorbents were temperature- and spatially-dependent.
The rates were calculated at each time step during
the model simulation period. Simpler equations were
used for decay of carbon species in the surficial
segments. BIC in sediments was assumed to be
zero. The equations used for PDC and DOC decay
in sediments were:
-
ds(DOC) ~ Kds(DOC)
Q(7"-20)
°
If -If *fl(r-2°)
Kds(PDC) ~ *ds(PDC) °
(4.3.6)
where
kds = segment-specific and temperature-dependent
decay rate of a carbon sorbent in sediments
~k = decay rate of a carbon sorbent in sediment at
20°C
T = measured segment-specific temperature (°C)
4.3.4.2 Sediment Transport
The transport and fate of PCBs in natural waters and
sediments are governed by the transport and fate of
POC. Based on the data collected during the
233
-------
LMMBP period, sediment PCB concentrations were
approximately four or five orders of magnitude higher
than water column PCB concentrations. The
interaction between the water column and surficial
sediments for POC was critical to PCB dynamics,
mass budget, and especially long-term concentration
in the lake. Therefore, it was necessary to describe
the important processes involved in vertical particle
transport, including settling velocity, resuspension
rate, and burial rate within the LM2-Toxic. There is
no unique set of carbon solids transport rates that
could be determined without the aid of a solids tracer
compound (Thomann and Di Toro, 1983; DePinto,
1994). Even though the burial rate could be fixed
based on the Pb-210 core dating technique, there still
was an infinite set of settling and resuspension rates
that could close the solids mass balance for vertical
particle transport.
4.3.4.2.1 Steady-State Resuspension Calibration
The overall modeling design for the LM2-Toxic was
intended to minimize the parameters needing to be
calibrated. It was originally planned not to calibrate
any particle transport parameters, including the
sediment resuspension rate. The original plan was
that LM3-Eutro and SEDZL would provide carbon
sorbent settling velocities and estimate POC
resuspension rates, respectively. However, it was
not clear in late 1999 that the particle transport rates
would be available in time for the LM2-Toxic
execution due to the departures of key project
personnel. After a series of discussion among staff,
it was decided that a steady-state PDC mass balance
approach would be used. A similar approach was
used for the Green Bay Mass Balance Project
(GBMBP) (Bierman et al., 1992). This approach was
used to estimate the segment-specific sediment POC
resuspension. In the water column, POC = BIG +
PDC, while in the sediment, POC = PDC.
There was only one sediment layer specified in LM2-
Toxic model segmentation (Figure 4.3.3). The
thickness of each surficial mixing sediment segment
was estimated based on the surficial sediment mixing
layer thickness derived from the LMMBP box cores
(Robbins et al., 1999). Figure 4.3.5 shows a
schematic of the concepts used for the steady-state
PDC mass balance analysis. Due to the simple
sediment segmentation within the LM2-Toxic, the
procedure for determining the segment-specific
water
V,
surficial sediment
Cs = f (O, FD, p)
Figure 4.3.5. Schematic of conceptualization for
the steady-state mass balance analysis for PDC
vertical transport where 4> = porosity
(dimensionless), FD = fraction of organic carbon,
9dv/9» P = bulk density of surficial sediments, wet
weight, g/cm3.
sediment PDC resuspension velocities was much
more simplified. The set of PDC mass balance
equations for each sediment segment for PDC
vertical transport was reduced to only one. The
simplified equation can be written as:
dt
-vb*A*Cs-kds*Vs*Cs
dC,
(4.3.7)
At steady-state, —p = 0, and Vs = A*z, the
resuspension velocity can be solved as:
(4.3.8)
where
Vs = volume of the surficial sediment segment (m3)
Cs = bulk concentration of PDC in the surficial
sediment segment (mg C/m3); where C =
carbon
vs = PDC gross settling velocity (m/d)
234
-------
A = PDC settling area for a surficial sediment
segment (m2)
Cw = effective PDC concentration in water column
segment right above the sediment segment
(mg C/m3); where C = carbon. Cw = CWPDC +
(VSBIC/VSPDC)*CWBIC, where vsBIC = BIC gross
settling velocity, m/d; VSPDC = vs = PDC gross
settling velocity, m/d; CWPDC = PDC
concentration in the water column segment,
mg C/m3; CwBIC = BIC concentration in the
water column segment, mg C/m3.
vr = PDC resuspension velocity (m/d)
vb = sediment (PDC) burial rate (m/d)
kds = PDC decay rate in sediments (d"1)
z = thickness of the surficial sediment mixing
layer (m)
Equation 4.3.8 was manipulated differently for
different sediment zones. In the areas where the
water depth was greater than 100 m, resuspension
was considered as non-wave-induced resuspension.
The resuspension rates in these areas were directly
calculated using Equation 4.3.8. Although it is
arguable that there is any resuspension in the area
above which the water is deeper than 100 m (Part 1,
Chapter 4), the resuspension rate used in LM2-Toxic
was estimated. This was because of a lack of
available sediment transport models that were well
tested for not only accurately computing sediment
resuspension rate, but also satisfying the particle
mass balance in both water column and sediments
under this kind of coarse spatial resolution used in
LM2-Toxic. After combining wave height information
for Lake Michigan estimated for the LMMBP period,
resuspension velocities in non-depositional areas
were computed using empirical wave-induced
resuspension derived from Equation 4.3.8. Further
detailed data reductions and discussion of the
parameters used in Equation 4.3.8 will be provided in
Chapters 4 and 5.
4.3.4.2.2 Empirical Wave-Induced Resuspension
Calculation
An equation similar to the one used in the GBMBP
(Bierman etal., 1992) was developed for LM2-Toxic
to estimate the resuspension rates in
non-depositional zones. The equation is:
vr= a(W-Wcr)
where
(4.3.9)
vr = estimated daily, segment-specific
resuspension velocity (m/d)
W = segment-specific surface daily average
wave height (m)
Wcr = segment-specific critical wave height for the
segment below which there is no wave-
induced resuspension (m)
cr = segment-specific empirical wave coefficient
(d-1)
Due to lack of accurate ice cover information and
winter concentrations of BIC and PDC during the
LMMBP period, it was not feasible to estimate the
segment-specific base (W < Wcr) resuspension
velocity (v^, i.e., non-wave-induced resuspension in
non-depositional area) formulated in the original
wind-induced resuspension calculation presented in
the GBMBP report (Bierman etal., 1992). Therefore,
vr0 was removed from the original formula used in the
GBMBP report (Bierman etal., 1992). This made the
resuspension in the non-depositional areas a function
of wave heights, as shown in Equation 4.3.9.
As part of the LMMBP, Schwab and Beletsky, using
the NOAA/Great Lakes Environmental Research
Laboratory (GLERL) Donelan wave model, generated
surface wave heights (Schwab and Beletsky, 1998).
These values were made available on a high-
resolution grid of 5 x 5 km2 and on time scales of
one-hour. These data were then averaged and
aggregated on a daily basis for the 53 sediment
segments used in LM2-Toxic. The segment-specific
empirical wave coefficient (a) was estimated using
the following equation:
v,=
a
(4.3.10)
235
-------
where
vr = segment-specific average resuspension
velocity calculated for Equation 4.3.8 (m/d)
n = number of days during LMMBP period
Wj = segment-specific daily wave height (m)
Note: The term (Wj - Wcr) in the Equation 4.3.10
becomes zero when Wj < Wcr.
The Equation 4.3.10 was derived under the
assumption that the cumulative resuspension flux
computed from Equation 4.3.9 on a daily basis for the
LMMBP period was equal to the total resuspension
flux computed for the same period from the steady-
state sediment carbon mass balance in Equation
4.3.8. The empirical wave coefficient (a) was then
formulated as:
a =
4.3.11
n
The segment-specific critical wave height (Wcr) was
a very crucial parameter for calculating wave-induced
resuspension flux. The availability of field data
relating local resuspension to surface critical wave
heights was extremely important for defining the
segment-specific or depth-specific critical wave
heights in this approach. Fortunately, a relatively
good set of the wave-induced resuspension field data
(Appendix 4.3.1) was available to use with a great
deal of help from Nathan Hawley at NOAA/GLERL
(Hawley, 1999). The data set included the critical
wave heights required for resuspension and the
water depths at which local sediment resuspension
was observed. These data were based on 30
deployments during 1994-2000 for which sediment
concentration near the bed were plotted against the
wave heights from the GLERL/Donelan wave model
(Schwab and Beletsky, 1998). Figure 4.3.6 shows
deployment locations in Lake Michigan during this
period (Lesht and Hawley, 1987; Hawley, 1999;
Hawley, 2001). The details on this method are
discussed in Lesht and Hawley (1987) and Hawley
(1999,2001). The deployment locations were limited
to the southwest region of the lake and the number of
the deployments were relatively small. There were
Figure 4.3.6. Locations of the 30 deployments
between 1994 and 2000.
indications that for similar water depths, different
surface critical wave heights were required to
resuspend sediment in different regions of the lake
(Hawley, 2001). However, the data set was the best
available data (Appendix 4.3.1) and was applied to
the sediment segments within the depositional area.
Based on this data set, a simple linear regression
was performed on the data (deployment depth,
critical wave height) for stations at which local
resuspension occurred (Figure 4.3.7). Visual
examination of Figure 4.3.7 revealed that the linear
regression line was a reasonable approximation of
236
-------
Relation b/w Critical Wave Height and Station Depth
(Data provided by Nathan Hawley, NOAA, GLERL)
Vllcr = 0.0417Depth + 0.5688
R2 = 0.6ls41
20
40
100
120
60 80
Station Depth (m)
resuspension observed e no resuspension Linear (resuspension observed)
140
Figure 4.3.7. Regression analysis on the data set (resuspension observed only). For the convenience
of the viewer, the data at the stations with no resuspension observed are also put on this plot.
the relationship between water depth and critical
wave heights.
Assuming this linear relationship between the critical
wave height and water depth, the parameterization of
the segment-specific critical wave height (Wcr) is:
Wcr = 0.0417 *Depth + 0.5688
where
(4.3.12)
Depth = deployment depth at the resuspension
station (m)
Therefore, the daily wave-induced resuspension in
the non-depositional area was calculated using
Equation 4.3.9 given the segment-specific
parameters a and Wcr.
For the 30 deployments used to calculate the
relationship, there were 11 cases with no
resuspension observed. These deployments were
either at very large depths or occurred during the
stratified period. Previous observations (Lesht and
Hawley, 1987; Hawley, 1999; Hawley and Lesht,
1995; Hawley and Murthy, 1995), combined with
those listed in Appendix 4.3.1, lead to the conclusion
that resuspension events during the stratified period
are confined to shallow water regions (< 13 m) where
the epilimnion was the entire water column. It is
arguable whether Equation 4.3.12 should be applied
to both unstratified and stratified periods. Based on
Nathan Hawley's suggestion (Hawley, 2001) and
data in Appendix 4.3.1, separate criteria for
unstratified and stratified periods were unnecessary.
Therefore, it was decided that the Equation 4.3.12
would be used in the wave-induced resuspension
calculation without considering seasonal variability.
4.3.4.2.3 The Sediment Bed - Semi-Lagrangian
Option
Another important aspect related to sediment
transport was constructing the sediment bed. The
Semi-Lagrangian sediment bed option in IPX Version
2.74 (Velleux et a\., 2000) was incorporated into the
237
-------
LM2-Toxic model by taking advantage of the
flexibility in IPX, which allows a moving sediment-
water interface. Unlike the Eulerian option, the Semi-
Lagrangian option does not allow material to move
into or out of the modeled system across the
sediment bottom boundary. Therefore, the system
being modeled would not be artificially introducing or
losing mass through the sediment bed, especially for
cases when the total inventory of sediments is not
well-defined. In this option, the solids concentration
in all sediment segments was held constant. The
volume (thickness) of all surficial sediment segments
was allowed to vary from their initial values in
response to net settling and net resuspension fluxes.
In response to net settling (deposition), the thickness
of the surficial sediment segment increased, and the
segment was split into two vertically adjacent
segments when its maximum thickness was reached.
In response to net resuspension (scouring), the
thickness of the surficial sediment segment
decreased, and the subsurface sediment segment
replaced the surficial sediment segment when its
minimum thickness was reached. When there was
not enough sediment to be resuspended, the
resuspension was stopped regardless of the
resuspension velocity specified. Once sediment was
deposited, the resuspension was resumed in the
surficial sediment segment.
4.3.5 Chemical Balance
The transport and fate of hydrophobic organic
chemicals such as RGBs are closely linked to the
movement of carbon sorbents. The approaches
used in the past to model PCBs for the Great Lakes
were based on either homolog or total PCBs. LM2-
Toxic is a PCB congener-based model. It simulates
54 PCB congeners that accounted for roughly 70% of
the total PCB mass in Lake Michigan. Figure 4.3.8
shows the conceptual framework and processes
related to PCB dynamics in Lake Michigan. Four
phases were simulated in the LM2-Toxic for the
congeners. The four phases were dissolved, sorbed
to PDC, sorbed to BIG, and bound to DOC. The
processes considered important for PCB dynamics
in Lake Michigan water column and sediments are:
Conceptual Framework of Toxic Chemical Dynamics
Atmosphere
c
Advection in
Surface
Water
Layer
Tributary +
ther inflow
loading
_>. *
voiaiinzation
Deposition + Absorption
X A
bound chemical
BIG | PDC
1
DOC
T
TCDOC
A A
i
unbound
chemical
(dissolved)
Advection out
>
Exchange
Settling
Net advection
Exchange
Advection in
Sub-Water
Layers
>,.
*
I
i i
bound chemical
BIG | PDC
tl
DOC
t
Resuspension 1 [settling
II >
TTpoc unbc
^ ^- chen
T^QOC (dissc
y
3iffusion
>
und
nical
>lved)
k
Diffusion
f
Advection out
>-
Exchange
Surface
Sediment
Layer
bound chemical
PDC DOC
f Sedimentation
Figure 4.3.8. Conceptual framework used by the LM2-Toxic model for PCB congeners in Lake
Michigan.
238
-------
' Equilibrium partitioning between dissolved phase
and sorbed phases of PCBs.
• Air-water exchange of dissolved PCBs including
both volatilization and absorption.
• External loadings of PCBs including tributary
loads and atmospheric wet and dry deposition.
• Gross settling of particulate PCBs.
• Resuspension of particulate PCBs.
• Pore-water diffusion of dissolved PCBs at the
sediment-water interface.
• Advection across interfaces between water
segments.
• Dispersion of PCBs across interfaces between
water segments.
• Sediment burial of PCBs.
The following subsections present detailed
descriptions of PCB equilibrium partitioning, PCB air-
water exchange, and parameters associated with
these processes.
4.3.5.1 PCB Partitioning
Due to PCBs' hydrophobic nature, the partitioning
process and the movement of organic carbon
particles were very important for describing the
transport and fate of PCBs in Lake Michigan.
Therefore, estimation of PCB partitioning coefficients
was one of the major steps to secure a successful
calibration of LM2-Toxic.
Partitioning of a hydrophobic organic contaminant
such as PCBs in a dilute water system is, in general,
governed by the following relationship (Eadie et al.,
1990,1992; Bierman et al., 1992):
V
where
Kp = partitioning coefficient (L/kg)
(4.3.13)
Cp = particle associated contaminant (PCBs)
concentration (kg/L)
m = sorbent (either total suspended solid, total
suspended matter (TSM), or POC in LM2-
Toxic concentration (kg/L)
Cd - dissolved PCBs concentration (kg/L)
Based on this fundamental equation, two of the many
partition theories have been widely used to describe
the distribution of PCBs in a diluted water system
such as a lake or a river. These were a two-phase
PCB partitioning model and a three-phase PCB
partitioning model (Swackhamer and Armstrong,
1987; Eadie etal., 1990,1992; Bierman etal., 1992).
In the two-phase PCB partitioning model, PCBs is
either in dissolved phases or in a particulate phase.
The dissolved phases include both dissolved PCBs
and PCBs sorbed to DOC. The particulate phase is
the PCBs sorbed to POC (POC = BIC + PDC). The
partition coefficient is described as:
1/-I
A POC,a
where
'POC
(4.3.14)
'd,a
K'
POC, or
'POC ~
in situ partition coefficient for PCBs on
POC (L/kg OC)
concentration of PCBs bound to POC
(kg/L)
[POC] = concentration of POC (kg OC/L)
'a,a
concentration of PCBs in an analytically
defined dissolved phase (i.e., dissolved
and sorbed to DOC) (kg/L)
The two-phase model is a very simple, straight-
forward method and most commonly used to quickly
compute the in situ POC partition coefficients (K'POc.a)
for PCB congeners without using sophisticated
statistical analysis. The computed K'POc,a can tnen
be conveniently compared to the measured octanol-
water partition coefficient Kow (Karickhoff, 1981;
Baker and Eisenreich, 1986; Swackhamer and
Armstrong, 1987; Bierman et al., 1992).
239
-------
In the three-phase PCB partitioning model, PCBs
was distributed between the dissolved, POC
participate, and DOC participate phases. The
partition coefficients are defined as follows:
•'DOC
we
and
[DOC]Cd
(4.3.15)
K"
•'POC
POC
[POC]Cd
(4.3.16)
where
- partition coefficient for PCBs on DOC
(L/kg OC)
K'poc = in situ partition coefficient for PCBs on
POC (L/kg OC)
[DOC] = DOC concentration (kg OC/L)
CDOC = concentration of PCBs bound to DOC
(kg/L)
Cd - dissolved PCB concentrations (kg/L)
The total concentration of PCBs (CT) equals the sum
of the three phases:
CT=Cd+ C
DOC + CPOC
= Cd(1 + KDOC[DOC} + K'POC[POC])
(4.3.17)
Because of the difficulty of directly measuring
dissolved PCB concentrations (Cd), the POC (K'POc)
partition coefficients had to be estimated using either
a simple linear regression technique (Brannon etal.,
1991; Bierman et al., 1992) or a statistical analysis
such as a combination of the Levenberg-Marquardt
nonlinear least squares routine and root mean
square error algorithm (Bierman et al.', 1992). The
initial estimation of the K'pnr and Knnr for each
XDOC
selected PCB congener (total 40 congeners) was
done by applying the simple linear regression
technique to the following equation for the LMMBP-
generated data (see Karickhoff et al., 1979 and
Bierman et al., 1992 for detailed derivation of this
equation).
•'POC
K'
POC
K'
POC
The terms on the left-hand side and [DOC] on the
right-hand side are the measured concentrations for
each water column sample. The simple linear
regression was applied to the above equation to yield
a slope of KDOC/K'poc and an intercept of 1/K'POC-
From this, both partition coefficients K'Poc anc' KDQC
were calculated. Pairs of partition coefficients were
estimated by repeating the regression analysis for
each PCB congener selected.
After a thorough analysis of the PCB congener
partitioning coefficient results estimated using both
the two- and three-phase partitioning models, the
two-phase partitioning model was selected as the
approach to initially compute the POC partition
coefficients (K'POc,a) that would be used in LM2-Toxic.
It was selected because 1) the two-phase partitioning
model was simple and very efficient in terms of data
analysis procedure; 2) though the regression
technique used in the three-phase partitioning model
worked well for a well-behaved system, it was not
applicable to a natural water system with
heterogeneous organic carbon concentrations, PCB
characteristics, and PCB concentrations; and 3) the
three-phase model-estimated PCB congener
partitioning coefficients on DOC (KDOC) did not have
similar trends as the ones on POC (K'POc)- Previous
publications suggested that the KDOC and K'POc °f
PCB congeners should have similar variation trends,
and that the value of KDOC should be within 1 to 2
orders of magnitude less than the value of K'POc
(Carter and Suffet, 1982; Landrum et al., 1984,1987;
Hassettand Milicic, 1985; Chiou et al., 1986,1987;
Eadie et al., 1990,1992; Bierman et al., 1992).
4.3.5.2 PCB Air-Water Exchange
Previous studies (Endicott et al., 2005; Endicott,
2005; DePinto etal., 2003) have suggested that net
volatilization to the atmosphere may be the
predominant loss mechanism for hydrophobic
organic contaminants such as PCBs in Great Lakes.
Therefore, it was very important to precisely compute
240
-------
the net PCB mass transfer across the water-air
interface in order to satisfy the overall PCBs
inventory and mass budget in the Lake Michigan
system for the LMMBP period, and forecast PCB
concentrations in both the water column and
sediments with a certain degree of confidence. The
mass change rate term (Sawj) for PCBs due to air-
water exchange processes in Equation 4.3.19 was
calculated as a product of the overall net mass
exchange flux and surface area of the water segment
'dwj
H
f)
(4.3.19)
where
kol = the overall mass exchange rate coefficient
(m/d)
Crfn, = dissolved PCB concentrations in water
'dm
(ng/m3)
Ca = atmospheric PCB concentrations (ng/m3)
H' = temperature dependent Henry's Law
Constant for a PCB congener
(dimensionless)
Aj = surface area of the water segment j (m2)
The overall mass exchange rate coefficient (k0[) was
calculated using the Whitman two-film theory
formulation (Whitman, 1923) given as:
1
(4.3.20)
k, K,*H'
where
k, = the liquid film mass transfer rate coefficient
(m/d)
kg = the gas film mass transfer rate coefficient
(m/d)
The LMMBP Atmospheric Workgroup recommended
that the Wanninkhoff (Wanninkhoff, 1992) formulation
for water mass transfer resistance and the
Schwarzenbach (Schwarzenbach et al., 1993)
formulation for gas mass transfer resistance were the
most appropriate for modeling the air-water
exchange of PCBs in Lake Michigan. The
Wanninkhoff equation for k,, with correction for PCBs
molecular diffusivity in reference to carbon dioxide
(CO2) molecular diffusivity across the air-water
interface, is given as:
k, = 0.45
where
D..
''-co.
*u
1.64
10
(4.3.21)
= chemical molecular diffusivity in water
(cm2/s)
I co
molecular diffusivity in water (cm2/s)
uw = wind velocity measured at 10 m above
water surface (m/s)
The Schwarzenbach formulation for kg with correction
of PCB molecular diffusivity in reference to water
vapor molecular diffusivity across the air-water
interface is given as:
p. r
= (0.2*U10 + 0.3)*
Dg_H2o
where
Da - chemical molecular diffusivity in air
(cm2/m)
Dg H O = water vapor molecular diffusivity in gas
phase (cm2/m)
Another recommendation from the LMMBP
Atmospheric Workgroup was to use the equation of
Henry's Law Constant for PCB congeners updated
by Bamford using recently developed data (Bamford
et a/., 2000). The equation originated from the
Gibbs-Helmholtz equation as:
InH' = -•
AH,
H
AS
H
(4.3.23)
Afer rearranging the equation, the temperature
dependent Henry's Law constant is given as:
241
-------
H' = e
where
fl*r
(4.3.24)
H' = temperature dependent Henry's Law
Constant (dimensionless)
AHH = the enthalpy of phase change for a PCB
congener (kJ/mol)
ASH = the entropy of phase change for a PCB
congener (kJ/mol)
R - the ideal gas constant, 8.315 x 10"3
kJ/(mol)(°K)
T = interfacial temperature (°K)
These volatilization formulas were coded into
subroutines of LM2-Toxic, and the parameters H', k,,
and kg were calculated at every time step for each
LM2-Toxic segment.
4.3.5.3 PCB-Specific Parameterization
PCB congener-specific parameters input into LM2-
Toxic model included PCB congener partitioning
coefficients, molecular weight, enthalpy, and entropy.
Chapter 4 of this part provides detailed information
on the values used in the LM2-Toxic for these
parameters, their data sources, and parameterization
procedures.
4.3.6 Modification
LM2-Toxic features some important updates from
both the IPX and GBTOX models. The project-
specific modifications were:
• Incorporation of organic carbon sorbent dynamics
used in GBMBP (Bierman etal., 1992).
• Organic carbon kinetic processes and
biotransformation.
• Incorporation of the quasi-Lagrangian framework
for sediment transport scheme used in the IPX
model.
• Added the Gibbs-Helmholtz formulation (Bamford
etal., 2000) to compute the Henry's Law Constant
for PCB congeners.
• Added the Wanninkhoff formulation for liquid film
mass transfer rate coefficient and the
Schwarzenbach formulation for gas film transfer
rate coefficient in the PCB overall volatilization
equations.
• Parameterization of segment-specific
resuspension velocity of carbon sorbent (PDC) as
a function of daily average wave height.
• Added subroutines for performing segment-
specific and lake-wide mass budget calculations,
a very useful tool to identify programming errors,
to identify artificial gain/loss of solid/chemical
mass to a water system, and to construct overall
mass budget for the system simulated.
• Capability for specifying segment-specific and
daily water temperature time functions.
• Capability for specifying segment-specific and
daily air temperature time functions.
• Capability for specifying segment-specific and
daily wind speed time functions.
• Capability for specifying segment-specific, PCB
congener-specific, and daily atmospheric PCB
concentration time functions.
• Reorganization/rewrite of many subroutines and
minor bug fixes to Euler integration. Use of utility
libraries and organized error handling (UT library)
in the codes. Modification to input format - using
FIREAD.
LM2-Toxic was developed and tested on both the
Unix and the Linux platforms and used both
FORTRAN 77 and FORTRAN 90 compilers.
242
-------
References
Ambrose, R.B., T.A. Wool, J.P. Connolly, and R.W.
Shanz. 1988. WASP4, A Hydrodynamic and
Water Quality Model - Model Theory, User's
Manual and Programmer's Guide. U.S.
Environmental Protection Agency, Office of
Research and Development, Environmental
Research Laboratory, Athens, Georgia.
EPA/600/3-87/039, 297 pp.
Baker, J.E. and S.J. Eisenreich. 1986. Influence of
Colloids on Sediment-Water Partition Coefficients
of Polychlorobiphenyl Congeners in Natural
Waters. Environ. Sci. Technol., 20(11):1136-
1143.
Bamford, H.A., J.H. Offenberg, R.K. Larsen, F.C. Ko,
J.E. Baker. 1999. Diffusive Exchange of
Polycyclic Aromatic Hydrocarbons Across the Air-
Water Interface of the Patapsco River, An
Urbanized Subestuary of the Chesapeake Bay.
Environ. Scr. Technol., 33(13):2138-2144.
Bamford, H.A., D.L. Poster, and J.E. Baker. 2000.
Henry's Law Constant of Polychlorinated
Biphenyl Congeners and Their Variation With
Temperature. J. Chem. Engin., 45:1069-1074.
Bierman, V.J., Jr., J.V. DePinto, T.C. Young, P.W.
Rodgers, S.C. Martin, and R. Raghunathan.
1992. Development and Validation of an
Integrated Exposure Model for Toxic Chemicals
in Green Bay, Lake Michigan. U.S.
Environmental Protection Agency, Office of
Research and Development, ERL-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
381 pp.
Blumberg, A.F. andG.L. Mellor. 1987. A Description
of a Three-Dimensional Coastal Ocean
Circulation Model. In: N.S. Neaps (Ed.), Three
Dimensional Coastal Ocean Models, Coastal and
Estuarine Sciences, pp. 1-16. American
Geophysical Union, Washington, D.C.
Brannon, J.M., T.E. Myers, D. Gunnison, and C.B.
Price. 1991. Nonconstant Polychlorinated
Biphenyl Partitioning in New Bedford Harbor
Sediment During Sequential Batch Leaching.
Environ. Sci. Technol., 25(6): 1082-1087.
Cahill, R.A. 1981. Geochemistry of Recent Lake
Michigan Sediments. Illinois Geological Survey,
Champaign, Illinois. Circular 517, 94 pp.
Carter, C.W. and I. H. Suffett. 1982. Binding of DDT
to Dissolved Humic Materials. Environ. Sci.
Technol., 16(11):735-740.
Chapra, S.C. and K.H. Reckhow (Eds.). 1983.
Engineering Approaches for Lake Management,
Volume 2: Mechanistic Modeling. Ann Arbor
Science Publishers, Ann Arbor, Michigan. 492
pp.
Chiou, C.T., R.L. Malcolm, T.I. Brinton, and D.E. Kile.
1986. Water Solubility Enhancement of Some
Organic Pollutants and Pesticides by Dissolved
Humic and Fulvic Acids. Environ. Sci. Technol.,
20(5):502-508.
Chiou, C.T., D.E. Kile, E.I. Brinton, R.L Malcolm, and
J.A. Leenheer. 1987. A Comparison of Water
Solubility Enhancements of Organic Solutes by
Aquatic Humic Materials and Commercial Humic
Acids. Environ. Sci. Technol., 21 (1 2): 1 231 -1 234.
DePinto, J.V. 1 994. Role of Mass Balance Modeling
in Research and Management of Toxic
Chemicals in the Great Lakes. The Green Bay
Mass Balance Study. Great Lakes Res. Rev.,
DePinto, J.V., W.M. Larson, J. Kaur, and J. Atkinson.
2003. LOTOX2 Model Documentation In
Support of Development of Load Reduction
Strategies and a TMDL for PCBs in Lake Ontario.
Submitted to New England Interstate Water
Pollution Control Commission, Boott Mills South,
Lowell, Massachusetts. 122 pp.
Eadie, B.J., N.R. Morehead, and P.P. Landrum.
1990. Three-Phase Partitioning of Hydrophobic
Organic Compounds in Great Lakes Waters.
Chemosphere, 20(1/2):161-178.
Eadie, B.J., N.R. Morehead, J.V. Klump, and P.P.
Landrum. 1992. Distribution of Hydrophobic
Organic Compounds Between Dissolved and
Particulate Organic Matter in Green Bay Waters.
J. Great Lakes Res., 18(1):91-97.
243
-------
Eadie, B.J. and S. Lozano. 1999. Grain Size
Distribution of the Surface Sediments Collected
During the Lake Michigan Mass Balance and
Environmental Mapping and Assessment
Programs. National Oceanic and Atmospheric
Administration, Great Lakes Environmental
Research Laboratory, Ann Arbor, Michigan.
NOAATechnical Memorandum ERLGLERL-111,
42pp.
Endicott, D.D. 2005. 2002 Lake Michigan Mass
Balance Project: Modeling Total PCBs Using the
MICHTOX Model. In: R. Rossmann (Ed.),
MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan, Part 2. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
EPA/600/R-05/158, 140 pp.
Endicott, D.D., W.L. Richardson, and D.J. Kandt.
2005. 1992 MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan. In: R. Rossmann (Ed.),
MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan, Part 1. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse He, Michigan.
EPA/600/R-05/158, 140 pp.
Hassett, J.P. and E. Milicic. 1985. Determination of
Equilibrium and Rate Constants for Binding of a
Polychlorinated Biphenyl Congener by Dissolved
Humic Substances. Environ. Sci. Technol.,
19(7):638-643.
Hawley, N. and B.M. Lesht. 1995. Does Local
Resuspension Maintain the Benthic Boundary
Layer in Lake Michigan? J. Sediment. Res.,
65:69-76.
Hawley, N. and C.R. Murthy. 1995. The Response
of the Benthic Nepheloid Layer to a Downwelling
Event. J. Great Lakes Res., 21(4):641-651.
Hawley, N. 1999. Sediment Resuspension and
Transport in Lake Michigan. Final Report. U.S.
Environmental Protection Agency, Office of
Research and Development, ERL-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
240 pp.
Hawley, N. 2001. Critical Wave Heights. National
Oceanic and Atmospheric Administration, Great
Lakes Environmental Research Laboratory, Ann
Arbor, Michigan. E-mail sent to Xiaomi Zhang on
February 1, 2001 (Appendices 4.3.1 and 4.3.2).
Karickhoff, S.W., D.S. Brown, and T.A. Scott. 1979.
Sorption of Hydrophobic Pollutants on Natural
Sediments. Water Res., 13(3):241-248.
Karickhoff, S.W. 1981. Semi-Empirical Estimation of
Hydrophobic Pollutants on Natural Sediments
and Soils. Chemosphere, 10(8):833-846.
Landrum, P.F., S.R. Nihart, B.J. Eadie, and W.S.
Gardner. 1984. Reverse-Phase Separation
Method for Determining Pollutant Binding to
Aldrich Humic Acid and Dissolved Organic
Carbon of Natural Waters. Environ. Sci.
Technol., 18(3):187-192.
Landrum, P.F., S.R. Nihart, B.J. Eadie, and LR.
Herche. 1987. Reduction in Bioavailability of
Organic Contaminants to the Amphipod
Pontoporeia hoyi by Dissolved Organic Matter of
Sediment Interstitial Waters. Environ. Toxicol.
Chem., 6(1 ):11-20.
Lesht, B.M. and N. Hawley. 1987. Near-Bottom
Currents and Suspended Sediment
Concentrations in Southeastern Lake Michigan.
J. Great Lakes Res., 13(3):375-386.
Quinn, F.H. 1977. Annual and Seasonal Flow
Variations Through the Straits of Mackinac.
Water Resources Res., 13(1):137-144.
244
-------
Bobbins, J.A., N.R. Morehead, R.W. Rood, D.N.
Edgington, and S. Meyer. 1999. Accumulation
and Near-Surface Mixing of Sediments in Lake
Michigan as Determined for the Lake Michigan
Mass Balance Program, Volumes 1 and 2.
Project Report. U.S. Environmental Protection
Agency, Office of Research and Development,
ERL-Duluth, Large Lakes Research Station,
Grosse He, Michigan. 503 pp.
Schwab, D.J. and D. Beletsky. 1997. Modeling
Thermal Structure and Circulation in Lake
Michigan. In: Estuarine and Coastal Modeling,
pp. 511-522. Proceedings of the 5th International
Conference of the American Society of Civil
Engineers, Alexandria, Virginia. October 22-24,
1997.
Schwab, D. and D. Beletsky. 1998. Lake Michigan
Mass Balance Study: Hydrodynamic Modeling
Project. National Oceanic and Atmospheric
Administration, Great Lakes Environmental
Research Laboratory, Ann Arbor, Michigan.
NOAATechnical Memorandum ERLGLERL-108,
53pp.
Schwarzenbach, R.P., P.M. Gschwend, and D.M.
Imboden. 1993. Environmental Organic
Chemistry. John Wiley and Sons, Incorporated,
New York, New York. 681 pp.
Settles, M. 1997. IPX: A Review of Planning and
Development. U.S. Environmental Protection
Agency, Office of Research and Development,
ERL-Duluth, Large Lakes Research Station,
Grosse lie, Michigan. Technical Note TEC-16,11
PP.
Swackhamer, D.L. and D.E. Armstrong. 1987.
Distribution and Characterization of PCBs in Lake
Michigan Water. J. Great Lakes Res., 13(1):24-
36.
Thomann, R.V. and D.M. Di Toro. 1983. Physico-
Chemical Model of Toxic Substances in the Great
Lakes. J. Great Lakes Res., 9(4):474-496.
Thomann, R.V. and J.V.Mueller. 1987. Principles of
Water Quality Modeling and Control. Harper and
Row Publishers, New York, New York. 644 pp.
U.S. Environmental Protection Agency. 1997. Lake
Michigan Mass Balance Study (LMMB) Methods
Compendium, Volume 1: Sample Collection
Techniques. U.S. Environmental Protection
Agency, Great Lakes National Program Office,
Chicago, Illinois. EPA/905/R-97/012a, 1,440pp.
Velleux, M., S. Westenbroek, J. Ruppel, M. Settles,
and D.D. Endicott. 2000. A User's Guide to
IPX, The In-Place Pollutant Export Water Quality
Modeling Framework, Version 2.7.4. U.S.
Environmental Protection Agency, Office of
Research and Development, National Health and
Environmental Effects Research Laboratory,
MED-Duluth, Large Lakes Research Station,
Grosse He, Michigan. 179 pp.
Wanninkhoff, R., J.R. Ledwell, and J. Crusius. 1991.
Gas Transfer Velocities on Lakes Measured with
Sulfur Hexafluoride. In: S.C. Wilhelm and J.S.
Gulliver (Eds.), Air-Water Mass Transfer, pp.441-
458. American Society of Civil Engineers, New
York, New York.
Wanninkhoff, R.J. 1992. Relationship Between Gas
Exchange and Wind Speed Over the Ocean. J.
Geophys. Res., 97:7373-7381.
Whitman, W.G. 1923. A Preliminary Experimental
Confirmation of the Two-Film Theory of Gas
Absorption. Chem. Metall. Eng., 29:146-148.
Zhang, X., D. Endicott, and W. Richardson. 1998.
Transport Calibration Model With Level 2 Model
Segmentation Scheme. First Lake Michigan
Mass Balance Project Science Panel Review,
Southgate, Michigan. June 23,1998. 12pp.
Zhang, X., W. Richardson, and K. Rygwelski. 2000.
Preparation and Verification Transport Field for
LMMBP Level 2 Contaminant: Transport and
Fate Models. Second Lake Michigan Mass
Balance Project Science Panel Review,
Southgate, Michigan. September 27, 2000. 15
pp.
245
-------
PART 4
LM2-TOXIC
Appendix 4.3.1. Lake Michigan Resuspension Field Data Set (One of the Two
Attachments in Nathan Hawley's E-Mail on February 2, 2001)
Station
Musk Wl
M24
M27
M19
Musk Wl
Musk Wl
M24
M27
M19
Musk Wl
M24
M27
M19
Leshtl
Lesht2
LeshtS
MWI
M24
M27
W1
W2
W3
S Haven
MO4
MO9
M11
MWI
MWI
M25
M55
Deployed
01.11.1994
01.11.1994
01.11.1994
01.11.1994
25.05.1995
12.07.1995
12.07.1995
12.07.1995
12.07.1995
31.08.1995
31.08.1995
31.08.1995
31.08.1995
02.04.1998
23.07.1998
28.10.1998
24.07.1998
24.07.1998
24.07.1998
15.10.1998
15.10.1998
27.10.1998
15.10.1999
03.03.2000
03.03.2000
03.03.2000
07.04.2000
13.09.2000
13.09.2000
13.09.2000
Retrieved
24.05.1995
24.05.1995
25.05.1995
24.05.1995
12.07.1995
21.08.1995
21.08.1995
21.08.1995
21.08.1995
17.11.1995
17.11.1995
12.10.1995
12.10.1995
30.04.1998
24.08.1998
01.12.1998
13.08.1998
13.08.1998
13.08.1998
11.11.1999
20.04.1999
10.05.1999
17.11.1999
22.05.2000
22.05.2000
22.05.2000
29.05.2000
30.10.2000
30.10.2000
27.11.2000
Latitude
4312.30'N
4313.75'N
43 09.50'N
42 02.93'N
4312.30'N
43 12.30'N
4313.75'N
43 09.50'N
43 02.93'N
43 12.30'N
43 13.75'N
43 09.50'N
43 02.93'N
42 39.90'N
42 52.22'N
4252.18'N
43 12.32'N
4311.33'N
43 10.04'N
42 08.09'N
41 44.1 4'N
42 57.50'N
42 24.23'N
41 55.58'N
42 14.87'N
42 17.36'N
43 12.21'N
43 12.23'N
43 12.24'N
43 12/73'N
Longitude
86 20.83'W
86 25.46'W
85 25.87'W
86 38.57'W
86 20.83'W
86 20.83'W
86 25.46'W
86 25.87'W
86 38.57'W
86 20.83'W
86 25.46'W
86 25.87'W
86 38.57'W
87 44.89'W
87 42.41 'W
87 42.41 'W
86 20.44'W
86 22.76'W
86 25.87'W
86 29.50'W
86 54.45'W
87 48.79'W
86 19.68'W
86 39.92'W
86 24.74'W
86 30.60'W
86 21 .OO'W
8621.32'W
86 22.90'W
86 28.65'W
Depth
13
28
58
100
13
13
28
58
100
13
28
58
100
15
25
25
14
30
60
10
10
16
18
20
18
38
15
17
26
55
Wave Height
0.7
1
3.1
>4.5
1
>1.5
">1,5"
>1.5
>1.6
0.8
1.7
>1.7
>3.3
1.5
>1.7
2.6
>1.4
>1.4
>1.4
1
0.8
2
.4
.1
2
.6
2
2.7
Comments
No resuspension
No resuspension
No resuspension
No resuspension
No resuspension
No resuspension
No resuspension
No resuspension
No resuspension
No resuspension
No resuspension
246
-------
PART 4
LM2-TOXIC
Appendix 4.3.2. Notes (One of the Two Attachments in Nathan Hawley's E-Mail on
February 2, 2001) From Nathan Hawley on the Data Set in Appendix 4.3.1
Dear Xiaomi:
As you requested, I am attaching the spreadsheet with the critical wave heights required for resuspension.
These were determined for the thirty deployments by plotting the sediment concentration near the bed against
the wave heights from the GLERL wave model. This method was used by Barry and I in our paper analyzing
a data set from 1981 (Lesht and Hawley, 1987, Journal of Great Lakes Research, v. 13, 375-386, see fig 6
for an example). In the present case this method assumes that a) high sediment concentrations are caused
only by local resuspension, and b) that local resuspension is caused mainly by wave action. If both these
assumptions are true then high sediment concentrations will occur only when the wave height exceeds a
certain value. For the thirty deployments listed in the spreadsheet, these assumptions appear to hold in about
1/3 of the cases. In another 1/3 of the cases no resuspension occurred at all, in these cases the maximum
waves during the deployment can be used as a lower bound for the critical height (the height required for
resuspension must exceed the height listed). These deployments were either at a very large depth (M19,
100m) or occurred during the stratified period. In the remaining cases there was no clear critical wave height
but resuspension did occur. In these cases, I determined the wave height by visually examining the time
series observations of concentration and wave height and then estimating the critical height as best I could.
In most cases the results aren't totally consistent (there are instances where waves larger than the critical
height do not correlate with increased sediment concentrations), but I did the best I could.
We might do a bit better if we used the combined (waves plus currents) bottom stress as the forcing
parameter, but this depends upon the wave period as well, and the wave model doesn't do a real good job of
calculating the wave period.
If you look at the data carefully, there are indications that for similar water depths larger waves are required
to resuspend sediment on the western side of the lake than on the eastern side. There is also some indication
that the sediment properties at a given location vary somewhat throughout the year, but I don't think that there
is enough data to say anything more. I did a rough plot of the data and fitted a straight line by eye. My line
suggests that a wave height of about 4.8 m would be required to resuspend sediment at 100m.
247
-------
PART 4
LM2-TOXIC
Chapter 4. Model and Field Data
The inputs necessary for running LM2-Toxic came
from two major sources. One was the Lake Michigan
Mass Balance Project (LMMBP)-generated data
collected between 1994 and 1995. Another was
historical data. These were primarily physical,
chemical, and biotransformation data. Most of the
parameters initially selected for the LM2-Toxic input
were from the Green Bay Mass Balance Project
(GBMBP) (DePinto et al., 1993). All project-
generated data were subjected to quality assurance
(QA)/quality control (QC) coordinated by the United
States Environmental Protection Agency (USEPA),
Great Lakes National Program Office (GLNPO)
(McCarty et al., 2004). Detailed descriptions of
sampling collection and sample analysis techniques
can be found in the LMMB Methods Compendium
(U.S. Environmental Protection Agency, 1997).
Further analysis and preprocessing of these data
followed the guidance detailed in the LMMBP Quality
Assurance Project Plan (QAPP) for Mathematical
Modeling (Richardson et al., 2004). Descriptions of
model input, data analysis procedures to generate
the model inputs, and sources of data are described
in the following sections.
4.4.1 Water Transport
Advective flows and bulk dispersion between water
segments drive the transport of all constituents
among water column segments of Lake Michigan.
Water transport data was among the major critical
inputs for the LM2-Toxic. Water transport fields and
associated data sources were provided in Chapter 3.
The following subsections provide additional
information on the transport field.
4.4.1.1 Circulation
The advective components of water transport in the
LM2-Toxic input consisted of bi-directional horizontal
flows, net vertical flows, balancing flows, and
tributary and boundary flows. Other than the
tributary and boundary flows, all of the flows were
expressed as daily time series for the model. Bi-
direction horizontal flows and net vertical flows were
aggregated and provided by Dr. David Schwab
(National Oceanic and Atmospheric Administration
[NOAA], Great Lakes Environmental Research
Laboratory [GLERL]) based on the 5 x 5 km2 grid
results from the Princeton Ocean Model for Great
Lakes (POMGL, Schwab and Beletsky, 1998).
Balancing flows were computed at each interface
based on the residual flows for the adjacent
segments. The horizontal and vertical flows provided
by Dr. David Schwab were 10 days short of the
complete two-year LMMBP period. For consistency
with other model input time series and convenience
to conduct long-term management scenario
simulations, a complete two-year time function for
advective flows was preferred and necessary. To
avoid unnecessary complication, the flows of the last
10 days of 1994 were duplicated at the end of 1995.
Ten tributary flows (Figure 4.4.1) were input into the
model input as annual averaged flows. Tributary
flows are listed in Table 4.4.1. The boundary flows
across the Straits of Mackinac were input into the
LM2-Toxic as monthly average flows. The lower
returning flows at the boundary entered from
segments 15 and 16 and were routed out from
248
-------
Manistique
Menominee River
Milwaukee
River
uskegon River
rand River
Sheboygan
Rive
alamazoo River
Joseph River
Figure 4.4.1. Locations of 10 tributaries whose flows were considered a part of the water transport
used in the LM2-Toxic.
Table 4.4.1. Average Annual Flows of the 10 Monitored Tributaries
Tributary Name
Milwaukee River
Sheboygan River
Fox River
Menominee River
Manistique River
Pere Marquette River
Muskegon River
Flow (m3/s)
19
11
199
166
84
30
98
249
-------
(Figures 4.3.1 and 4.3.3). Table 4.4.2 presents the
monthly boundary flows (i.e., the lower returning
flows). Both the tributary and boundary flows were
assembled based on the flows used in MICHTOX
(Endicottefa/., 2005).
4.4.1.2 Vertical Dispersion
Another major component of the water transport field
defined/used in the LM2-Toxic input is the vertical
dispersion. In order to reasonably estimate the
vertical exchange coefficients for the interfaces
between vertically adjacent segments, a thermal
balance model with the same Level 2 spatial
segmentation was used to simulate temperature
using LMMBP-generated temperature field
measurements. Instead of chloride, temperature was
the state variable being simulated. The same data
reduction and interpolation procedures used for
chloride were used for temperature. Table 4.4.3 lists
the initial temperature for 41 segments in the thermal
balance model inputs. Table 4.4.4 contains cruise-
segment mean temperatures generated using
inverse distance weighted interpolated matrix (IDW)
and volume-weighted average (VWA). The values
listed in Table 4.4.4 were used for comparison with
the thermal balance model output. Heat fluxes
provided by Dr. David Schwab (Schwab and
Beletsky, 1998) were input as loads in the model.
The initial estimation of vertical exchange coefficients
was computed using a simple equation (Chapra and
Reckhow, 1983) presented below.
V*(dT,/dt)
/f= At*(Tu-T,)
Et=vt*zt
where
(4.4.1)
(4.4.2)
v, = vertical heat exchange coefficient across the
interface between upper segment and lower
segment (m/d)
V = average volume of the upper and lower
segments (m3)
7", = temperature in lower segment (°C)
7"0 = temperature in upper segment (°C)
At = surface area of the interface between upper
segment and lower segment (m2)
t = time, d - days
£, = vertical exchange coefficient (m2/d)
z, = mixing length for the upper segment and
lower segment (m)
Following a substantial calibration effort, the final set
of vertical dispersion coefficients was determined.
The calibration results of the thermal balance model
were very good. Further discussion on the results
will be provided in Chapter 5. The complete
calibration results from the thermal balance model
and the final set of vertical dispersion coefficient plots
are presented in Appendix 4.5.1.
4.4.1.3 Verification of Water Transport Fields
The inputs for the water transport field used in the
LM2-Toxic were composed of advective flows and
vertical dispersion. The water transport field was
verified by simulating chloride, a conservative
constituent, in Lake Michigan for the LMMBP period.
In addition to the data related to water transport field,
chloride tributary loads and segment-specific initial
chloride concentrations were the only necessary
input data to run the chloride model. Daily chloride
tributary loads for the complete two-year period from
11 monitored tributaries (Figure 4.4.2) were
calculated using the stratified Beale ratio model (Hall
and Robertson, 1998; Beale, 1962; Baun, 1982;
Cohn et aL, 1989; Richards, 1998). The 18
unmonitored tributary loads were not input into the
chloride model because they were not available by
the time (June 1999) model executions were finished.
Eight field sampling cruises occurred between April
1994 and October 1995 at 41 water survey stations
throughout Lake Michigan, Green Bay, and Lake
Huron (Figure 4.4.3). The eight field cruise sampling
periods are listed in Table 4.4.5. Multiple field
samples were collected at different depths at most of
the water survey stations. Some of the samples
were duplicates. The segment-specific chloride initial
conditions (Table 4.4.6) and eight cruise-segment
mean chloride concentrations (Table 4.4.7) were
generated using an interpolation procedure that
combined the IDW interpolation formulation and a
250
-------
Table 4.4.2. Monthly Average Flows Across the Straits of Mackinac
Lower Return Flow (m3/s) From Lower Return Flow (m3/s) From
Month Segment 16 Segment 15
January
February
March
April
May
June
July
August
September
October
November
December
40
452
0
0
0
1500
3000
2000
1000
0
0
0
36
400
0
0
0
1034
2096
1726
989
0
0
0
Table 4.4.3. Initial Temperatures in Water Column Segments for the Thermal Balance Model
Segment Temperature (°C) Segment Temperature (°C)
1 2.01
2 2.01
3 1.99
4 1.99
5 2.00
6 2.00
7 1.99
8 1.99
9 2.01
10 2.05
11 2.01
12 2.01
13 2.00
14 2.00
15 2.01
16 2.01
17 2.02
18 2.02
19 2.03
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
2.01
2.01
2.00
2.00
2.01
2.01
2.02
2.03
2.04
2.01
2.01
2.01
2.01
2.01
2.01
2.01
2.50
2.50
2.50
2.50
2.00
2.00
251
-------
Table 4.4.4.
Segment
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
Cruise-Segment Mean Temperatures
Cruise 1
3.93
4.51
2.54
2.64
1.99
1.93
2.70
2.78
4.74
4.65
3.56
3.77
2.49
2.57
2.01
1.90
2.71
2.83
4.54
3.23
3.41
2.54
2.59
2.10
1.93
2.66
2.79
4.30
2.94
2.93
2.20
2.33
2.06
1.86
2.52
2.68
2.72
2.55
2.62
2.14
1.96
Cruise 2
18.95
19.02
16.29
15.28
18.02
17.86
15.61
16.06
17.07
17.03
16.28
16.26
13.50
12.25
16.47
16.17
12.41
13.19
12.20
8.23
8.89
7.63
7.13
11.08
10.05
8.15
9.00
8.82
5.40
5.64
5.41
5.04
7.60
6.39
8.65
4.13
4.05
4.04
4.03
5.31
4.51
Cruise 3
21.04
23.22
21.16
21.27
21.20
21.17
21.83
22.12
21.53
21.48
10.54
13.70
14.71
13.95
14.47
14.36
14.89
14.79
15.00
6.34
6.36
8.03
7.31
8.00
7.80
9.49
9.56
9.43
5.44
5.20
6.16
5.67
5.69
5.58
6.75
4.11
4.12
4.32
4.23
4.25
4.19
for the LMMBP Project Period
Cruise 4
2.82
3.14
2.96
2.93
2.94
2.94
0.00
0.00
0.00
0.00
2.86
3.15
2.96
2.93
2.94
2.94
0.00
0.00
0.00
2.92
3.17
2.96
2.93
2.95
2.94
0.00
0.00
0.00
3.43
3.80
3.75
3.81
3.68
3.70
0.00
4.13
4.12
4.13
4.13
4.12
4.12
Cruise 6
2.37
2.36
2.30
2.50
2.05
2.01
1.56
1.48
1.26
1.26
2.38
2.38
2.31
2.51
2.06
2.00
1.57
1.48
1.27
2.35
2.34
2.33
2.53
2.06
1.98
1.59
1.49
1.29
2.34
2.24
2.40
2.57
2.08
1.88
1.39
2.42
2.40
2.58
2.59
2.01
1.80
Cruise 7
11.32
12.13
11.01
12.18
12.59
12.92
12.45
12.84
13.39
13.33
10.76
11.91
10.36
11.75
12.40
12.67
11.57
12.16
12.87
9.79
11.27
8.70
10.49
11.78
11.80
9.01
9.64
11.60
6.65
7.13
5.77
5.95
9.17
7.21
7.32
4.18
4.14
4.19
4.15
5.62
4.89
Cruise 8
15.22
15.25
15.57
14.90
17.07
16.80
16.58
16.68
16.72
16.72
13.74
14.77
14.78
14.34
16.99
16.62
16.40
16.57
16.63
8.90
10.37
11.77
11.80
14.49
13.61
13.02
13.46
14.64
5.14
5.08
6.93
6.91
9.18
7.44
8.59
4.07
4.05
4.18
4.10
5.12
4.78
252
-------
\A/U> *• u Sturgeon
Whitefishx a t
Millecoquins
Escanaba
Menominee River
Peshtiga
Oconto "~
Pensaukee;
Manitowac
Sheboygan
River
Milwaukee
River
Manistee
Pere Marquette River
uskegon River
Grand River
alamazoo River
\
Black
St. Joseph River
• Monitored tributaries
O Unmonitored tributaries
Figure 4.4.2. Locations of monitored and unmonitored tributaries during the LMMBP.
253
-------
Q
Menominee River
Milwaukee
River
Pere Marquette River
Sheboyga
River
Kalamazoo River
St. Joseph River
Figure 4.4.3. Lake Michigan water sampling sites during the LMMBP.
254
-------
Table 4.4.5. The LMMBP Sampling Cruises
Cruise Number Start Date End Date
Cruise 1 April 24, 1994 May 11, 1994
Cruise 2 June 17, 1994 June 26, 1994
Cruise 3 August 3, 1994 August 26, 1994
Cruise 4 October 14, 1994 November 7, 1994
Cruise 5 January 16, 1995 January 26, 1995
Cruise 6 March 23, 1995 April 18, 1995
Cruise 7 August 3, 1995 August 16,1995
Cruise 8 September 16, 1995 October 13, 1995
Table 4.4.6. Initial Chloride Concentrations in Water Column Segments for the Chloride Model
Segment Chloride (mg/L) Segment Chloride (mg/L)
1 10.39
2 10.78
3 10.25
4 10.33
5 9.93
6 9.94
7 10.55
8 10.65
9 12.66
10 12.55
11 10.38
12 10.68
13 10.24
14 10.32
15 9.82
16 9.80
17 10.56
18 10.67
19 12.51
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
10.17
10.18
10.21
10.21
10.08
9.82
10.45
10.49
10.40
10.15
10.14
10.21
10.19
10.26
10.23
10.51
10.17
10.18
10.17
10.16
10.12
10.09
255
-------
Table 4.4.7. Cruise-Segment Mean Chloride Concentrations for the LMMBP Period
Segment
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
Cruise 1
10.39
10.78
10.25
10.33
9.93
9.94
10.55
10.65
12.66
12.55
10.38
10.68
10.24
10.32
9.82
9.80
10.56
10.67
12.51
10.17
10.18
10.21
10.21
10.08
9.82
10.45
10.49
10.40
10.15
10.14
10.21
10.19
10.26
10.23
10.51
10.17
10.18
10.17
10.16
10.12
10.09
Cruise 2
10.25
10.46
10.19
10.20
10.22
10.23
10.20
10.20
10.19
10.19
9.98
10.24
10.19
10.20
10.23
10.24
10.20
10.20
10.19
10.13
10.16
10.16
10.14
10.17
10.16
10.17
10.17
10.18
10.12
10.16
10.13
10.12
10.13
10.13
10.13
9.99
10.00
10.12
10.11
10.13
10.12
Cruise 3
10.47
10.57
10.36
10.44
10.12
10.28
10.63
10.74
11.61
11.56
10.36
10.46
10.36
10.36
10.42
10.44
10.53
10.53
11.18
10.27
10.41
10.24
10.25
9.91
9.45
10.40
10.40
10.40
10.24
10.31
10.23
10.23
10.08
10.17
10.14
10.19
10.24
10.25
10.25
10.10
10.15
Cruise 4
10.45
10.55
10.23
10.30
9.79
9.86
10.45
10.51
11.09
11.05
10.43
10.58
10.26
10.32
9.74
9.80
10.43
10.49
11.01
10.39
10.58
10.23
10.31
9.76
9.41
10.39
10.48
10.90
10.29
10.28
10.18
10.18
10.05
10.14
10.22
10.19
10.24
10.19
10.20
10.14
10.17
Cruise 6
10.30
10.61
10.24
10.27
9.99
10.01
10.44
10.49
11.54
11.52
10.51
11.11
10.30
10.37
9.95
10.02
10.47
10.54
11.51
10.56
11.17
10.37
10.42
10.16
10.11
10.34
10.39
10.30
10.14
10.13
10.24
10.25
10.14
10.17
10.41
10.09
10.05
10.16
10.13
10.08
10.04
Cruise 7
10.46
10.38
10.24
10.27
10.16
10.16
10.47
10.56
10.38
10.36
10.15
10.14
10.07
9.98
8.34
8.33
10.17
10.18
10.15
10.17
10.14
10.03
10.00
8.61
8.27
9.99
10.02
10.12
10.20
10.19
10.40
10.35
10.50
10.45
10.49
10.22
10.19
10.37
10.35
10.45
10.42
Cruise 8
10.40
10.47
10.36
10.39
10.22
10.27
10.53
10.58
10.94
10.92
10.34
10.46
10.33
10.36
10.16
10.20
10.53
10.59
10.91
10.23
10.32
10.26
10.32
10.03
9.90
10.36
10.45
10.87
10.21
10.24
10.26
10.28
10.31
10.29
10.36
10.15
10.17
10.26
10.25
10.31
10.31
256
-------
VWA algorithm (Appendix 4.4.1). The major
advantages of this interpolation approach were its
convenience and effectiveness. The subroutines/-
programs of IDW and VWA are widely available in
many major software packages. On the other hand,
the conventional approach of calculating the
arithmetic mean and standard deviation of actual field
measurements is constrained by a data theory that
is only valid for a randomly sampled data set with
normal distribution. Most environmental field
samples, including water quality samples, were
collected based on biased sampling designs.
Appendix 4.4.1 and publications (Lesht, 1988a, b;
Bierman et al., 1992) provided further details on the
interpolation approach. Lake Michigan chloride
distribution was simulated once for the LMMBP
period without adjustment on any parameter or
coefficient. The results were very good. Further
details are presented in Chapter 5. Appendix 4.5.4
provides the complete simulation results from the
chloride model.
4.4.2 Organic Carbon
Three carbon sorbents were simulated as principal
state variables in the LM2-Toxic. They were
dissolved organic carbon (DOC), biotic carbon (BIC),
and particulate detrital carbon (PDC). Measured
data used for directly generating carbon sorbent
concentrations and estimating their loadings were
DOC and particulate organic carbon (POC). POC
was then divided into BIC and PDC by first applying
a fixed carbon/chlorophyll a ratio (C:Chl) to estimate
BIC (BIC is equal to the product of the ratio and
measured chlorophyll a). PDC was calculated by
subtracting BIC from POC. Negative values were not
allowed during the derivation. A final value (see Part
2, LM3-Eutro) Pauer et al., 2005) was selected for
splitting BIC and PDC from POC. DOC was
characterized as colloidal-sized particles that did not
settle. It was assumed that BIC did not accumulate
in sediment. Instead, BIC settling to the sediment
surface would instantly convert to PDC. DOC had
very little variation spatially and temporally
throughout the lake. The concentration of DOC was
about five times higher than POC (BIC + PDC). The
typical concentration of DOC in the lake was 1.6
mg/L BIC was strongly controlled by seasonal and
local phytoplankton blooms and associated
dynamics. The typical range of BIC concentrations
for the lake was between 0.05 and 0.25 mg/L. PDC
in the lake was highly controlled by phytoplankton
decomposition and settling and resuspension
processes. Therefore, the variation of PDC in the
lake was also seasonally and spatially characterized.
The typical range of PDC concentrations varied from
0.1 to 0.25 mg/L in Lake Michigan. The
concentration of PDC was doubled or tripled in Green
Bay where carbon loads from tributaries and
resuspension fluxes were more dominant
components than in the main lake. The average
PDC concentration range in sediments was from
2,000 to 7,000 mg/L.
4.4.2.1 Loads
Loads of organic carbon sorbents were classified as
external and internal. External loads entered the lake
from tributaries. Internal loads were generated from
gross primary production of phytoplankton within
Lake Michigan.
Segment-specific primary production was
aggregated from the high-resolution eutrophication
model LM3-Eutro. The gross primary production had
strong seasonal variations. Figure 4.4.4 shows the
seasonal fluctuation of gross primary production
generated from the LM3-Eutro for the lake during the
LMMBP period. By assuming 20% of gross primary
production released from phytoplankton was DOC,
the internal loads were then further divided into BIC
and DOC (Bierman era/., 1992).
5
^1
o
'•8
1
Q.
to
1.60E+07
1.40E+07
1.20E+07
1.00E+07
8.00E+06
6.00E+06
4.00E+06
2.00E+06
0
Lake Michigan Primary Production
Jan94
Jun94
Dec94
Jun95
Nov95
Figure 4.4.4. Primary production generated from
the LM3-Eutro for Lake Michigan, including Green
Bay.
External loads provided by the United States
Geological Survey (USGS) included loads from 11
monitored tributaries and 18 unmonitored tributaries
(Hall and Robertson, 1998). Tributary loads were
257
-------
received as DOC, POC, and chlorophyll loads. A
carbon/chlorophyll ratio of 40 was then used to split
POC loads into BIC and PDC loads.
Both external and internal loads were input into the
LM2-Toxic as daily time functions. Summaries of the
annual average organic carbon external and internal
loads are presented in Tables 4.4.8, 4.4.9, and
4.4.10. During the two-year LMMBP period, virtually
all of BIC load (98.6%) came from internal primary
production, virtually all of the POC load (POC = BIC
+ PDC, 97.3%) came from internal primary
production, and 88% of total organic carbon load
(TOC = DOC + BIC + PDC) came from internal
primary production. Therefore, the internal primary
production dictates the level of all three organic
carbon concentrations in the lake, and thus, it also
influences observed polychlorinated biphenyl (PCB)
concentrations in the system, including sediments.
4.4.2.2 Field Data and Initial Conditions
By including some field data (sediment data)
collected during the GBMBP, the LMMBP-generated
data were used to generate segment-specific water
column initial concentrations, sediment
concentrations, and cruise-segment mean
concentrations for the three organic carbon sorbents
simulated in the LM2-Toxic. Samples for water
column DOC and POC were collected at 41 water
survey stations (Figure 4.4.3) between April 1994
and October 1995. The original POC cruise data
were split into BIC and PDC by applying a carbon
and chlorophyll a ratio of 40. Both the inverse
distance and natural-neighbor algorithms (Appendix
4.4.1) were used for interpolation of all three organic
carbon concentrations in the water column. In order
to choose the best interpolation for the water column
carbon data set, the results from both interpolation
methods were compared. The two interpolation
methods gave quite different results when there were
not enough samples and poor spatial resolution. The
inverse distance algorithm was selected for the
interpolation of the organic carbon sorbents because
the interpolation results using the inverse distance
algorithm showed more realistic organic carbon
distributions in the water column than the ones from
the natural-neighbor algorithm. Table 4.4.11
presents the water column initial concentrations of all
three organic sorbent state variables used in the
LM2-Toxic. The cruise-segment mean
concentrations for DOC, BIC, and PDC are listed in
Tables 4.4.12, 4.4.13, and 4.4.14, respectively.
Sediments were collected during the LMMBP
sampling period of 1994 and 1995, and sediment
organic carbon data were collected and analyzed
(Eadie and Bobbins, 2004). A total of 116 sediment
samples (Figure 4.4.5) were used to generate
sediment segment-specific organic carbon
concentrations, including 53 box core samples, 60
Ponar grab samples, and three gravity core samples.
Among these sediment samples, only four stations
(15B1, 41B1, 86B1, and 112B1) had complete
vertical profile analysis done for organic carbon; the
rest of the stations had only the top 1 cm of sediment
analyzed. The reported sediment organic carbon
(POC) concentrations were reported as mg carbon
per g dry sediment (mg/gdw). In order to convert the
unit (mg/gdw) into the standard unit (mg/L, i.e., bulk
concentration) used in LM2-Toxic, the following
equation was used for the calculation:
FD *
103
(4.4.3)
where
sPOC
sPOC -
POC concentration in surficial sediments
(mg/L), the unit required for the LM2-
Toxic
POC concentration in surficial sediments
(mg/gdw), the unit of field measurements
bulk density of surficial sediments, wet
weight (g/cm3). To convert cm3 to L
(liter), a factor of 103 is needed in the
equation.
= fraction of organic carbon (gdv/g)
The sediment POC data from the GBMBP were
incorporated into the data set to generate the
sediment POC contour maps for Lake Michigan,
including Green Bay. Figures 4.4.6 and 4.4.7 show
the distribution of POC in Lake Michigan sediments
in units of mg/gdw and mg/L, respectively. The
natural-neighbor algorithm (Appendix 4.4.1) was
used for the interpolation and the segment-specific
258
-------
Table 4.4.8. Annual Average Organic Carbon Loads From 11 Monitored Tributaries to Lake Michigan
During the LMMBP
Tributary (Monitored)
Milwaukee River
Sheboygan River
Calumet River
St. Joseph River
Kalamazoo River
Grand River
Muskegon River
Pere Marquette River
Manistique River
Menominee River
Fox River
Total
DOC (ton/year)
2508.50
1549.45
1974.49
18202.45
10205.13
31922.79
13191.20
3990.09
11502.03
25727.66
33972.50
154746.30
BIC (ton/year)
284.13
237.09
61.85
2688.78
1585.62
4646.54
394.89
104.17
95.80
621.73
7059.93
17780.53
PDC (ton/year)
225.91
169.43
426.65
2801.15
1 827.60
3496.82
881 .68
738.60
1232.81
2477.21
4743.01
19020.87
Table 4.4.9. Annual Average Organic Carbon Loads From 18 Unmonitored Tributaries to Lake
Michigan During the LMMBP
Tributary (Unmonitored)
Manitowac
Root
Galien
Black
Kalamazoo-Minor
Kewaunee
Pere Marquette-Minor
Manistee
Betsie
Millecoquins
Jordan
Sturgeon
Whitefish
Escanaba
Cedar
Peshtigo
Oconto
Pensaukee
DOC (ton/year)
4507.49
1188.18
2530.82
31 32.23
268.82
4106.73
6841.12
11095.82
3909.74
4406.39
8012.93
4385.19
3997.10
5815.29
6544.03
7236.21
5568.69
2702.66
BIC (ton/year)
689.71
134.58
373.84
486.67
41.77
853.43
204.79
289.68
117.04
36.70
239.87
36.52
96.59
140.53
158.14
174.87
1157.25
561 .65
PDC (ton/year)
492.89
107.01
389.47
560.94
48.14
573.35
457.25
2053.92
261.32
472.28
535.57
470.01
384.86
559.93
630.10
696.74
777.46
377.33
Total
86249.43
5793.65
3896.44
259
-------
Table 4.4.10. Annual Average Organic Carbon Internal Loads Generated From the LM3-Eutro for Lake
Michigan During the LMMBP
Segment
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
DOC (ton/year)
700929.10
689189.92
518083.37
403259.87
189956.41
170745.86
68920.37
69189.01
124169.03
78248.13
250457.73
246500.23
188306.18
145633.10
53584.35
56110.59
15072.21
14053.69
13520.81
50927.44
49818.55
39893.25
31062.96
6879.84
10523.70
1388.52
816.87
213.29
9827.70
9681 .05
8149.94
6621.80
1017.37
1648.21
19.20
293.42
314.62
276.19
244.65
18.65
33.69
BIG (ton/year)
2803716.40
2756759.68
2072333.48
1613039.48
759825.64
682983.44
275681.48
276756.05
496676.11
312992.51
1001830.93
986000.92
753224.70
582532.39
214337.41
224442.38
60288.86
56214.78
54083.22
203709.77
199274.20
159573.00
124251.84
27519.37
42094.79
5554.07
3267.48
853.14
39310.79
38724.21
32599.76
26487.18
4069.47
6592.86
76.80
1173.68
1258.48
1104.74
978.61
74.59
134.75
Total
422560.09
1690240.34
260
-------
Table 4.4.11. Initial Concentrations of Organic Carbon Sorbents in Water Column Segments for the
LM2-Toxic
Segment
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
DOC (mg/L)
1 .5694
1 .7768
.5311
1.6138
1 .6464
1.6113
1 .7583
1.968
3.6044
3.6154
1 .5338
1.6015
1.532
1.6247
1 .5429
1 .4676
1.7187
2.0062
3.4877
1.5442
1 .5581
1 .6276
1 .6385
1 .5571
1.763
.8679
.8811
.8553
.5242
.5286
.7129
.6961
.8431
1.87
1.8
1.415
1 .3966
1 .6693
1 .6572
1 .7343
1 .7286
BIC (mg/L)
0.038332
0.038024
0.030031
0.029434
0.033097
0.030774
0.034256
0.036219
0.035715
0.07318
0.036831
0.036452
0.029746
0.029355
0.032192
0.030419
0.033151
0.03439
0.034217
0.035098
0.034508
0.029078
0.02881 1
0.030666
0.029705
0.031902
0.03355
0.034123
0.031241
0.03062
0.026917
0.026645
0.027635
0.027417
0.031769
0.024359
0.024124
0.019674
0.020367
0.023314
0.024052
PDC (mg/L)
0.1484
0.1383
0.1818
0.168
0.1763
0.1768
0.2616
0.2616
0.2616
0.2616
0.1544
0.1441
0.1802
0.1651
0.1761
0.1753
0.2616
0.2616
0.2616
0.1294
0.1218
0.1598
0.151
0.2015
0.1699
0.1865
0.1959
0.1777
0.1246
0.1222
0.1411
0.146
0.1538
0.1575
0.1388
0.0833
0.0811
0.1144
0.1169
0.1183
0.1182
261
-------
Table 4.4.12. Cruise-Segment Mean Concentrations of DOC (mg/L) for the LMMBP Period
Segment Cruise 1 Cruise 2 Cruise 3 Cruise 4 Cruise 5 Cruise 6 Cruise 7 Cruise 8
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
1.569
1.777
1.531
1.614
1.646
1.611
1.758
1.968
3.604
3.615
1.534
1.602
1.532
1.625
1.543
1.468
1.719
2.006
3.488
1.544
1.558
1.628
1.639
1.557
1.763
1.868
1.881
1.855
1.524
1.529
1.713
1.696
1.843
1.870
1.800
1.415
1.397
1.669
1.657
1.734
1.729
1.574
1.593
1.565
1.594
1.569
1.573
1.519
1.575
1.558
1.587
1.559
1.563
1.440
1.417
1.379
1.388
1.388
1.391
1.438
1.404
1.620
1.572
1.582
1.567
1.433
1.422
1.446
1.433
1.305
1.312
1.623
1.577
1.707
1.637
1.608
1.629
2.285
2.378
2.885
2.858
1.720
1.615
1.742
1.716
1.632
1.688
2.408
2.468
2.755
1.461
1.470
1.618
1.587
1.514
1.461
1.698
1.662
1.733
1.459
1.473
1.633
1.606
.566
.591
.661
.540
.475
.451
1.419
1.406
1.387
1.611
1.635
1.596
1.609
1.349
1.401
1.930
1.948
2.432
2.427
1.622
1.635
1.593
1.613
1.363
1.412
1.883
1.924
2.331
1.650
1.619
1.517
1.518
1.338
1.349
1.496
1.457
1.534
1.555
1.547
1.592
1.585
1.557
1.568
1.448
1.584
1.539
1.565
1.554
1.388
1.317
1 .503 .559
1 .482 1 .637
1.496 1.515
1 .499 1 .51 1
1.497 1.484
1.497 1.483
1.672
1.797
2.770
2.776
1.561
1.592
1.569
1.535
1.505
1.504
1.649
1.820
2.700
1.535
1.487
1.605
1.530
1.524
1.543
1.730
1.730
1.730
1.531
.550
.453
.474
.382
.371
1.476
1.480
1.457
1.440
1.383
1.371
1.594
1.611
1.676
1.639
1.646
1.624
2.108
2.248
1.965
1.939
1.604
1.620
1.683
1.647
1.663
1.630
2.111
2.244
1.976
1.385
1.391
1.348
1.326
1.284
1.253
1.430
1.430
1.430
1.354
1.375
1.377
1.345
1.318
1.318
1.397
1.433
1.400
1.433
1.409
1.311
1.320
1.709
1.681
1.691
1.724
1.701
1.821
1.789
1.852
2.149
2.135
1.668
1.690
1.633
1.683
1.606
1.738
1.691
1.780
2.037
1.540
1.578
1.483
1.551
1.436
1.466
1.419
1.444
1.395
1.471
1.481
1.392
1.416
1.488
1.687
1.445
1.443
1.439
1.464
1.489
1.542
1.713
262
-------
Table 4.4.13. Cruise-Segment Mean Concentrations of BIC (mg/L) for the LMMBP Period
Segment
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
Cruise 1
0.092
0.093
0.090
0.090
0.098
0.095
0.100
0.100
0.100
0.100
0.093
0.093
0.088
0.083
0.097
0.089
0.100
0.100
0.100
0.093
0.095
0.091
0.084
0.127
0.105
0.101
0.101
0.102
0.095
0.093
0.091
0.083
0.098
0.099
0.102
0.064
0.067
0.046
0.042
0.049
0.049
Cruise 2
0.087
0.094
0.106
0.107
0.104
0.104
0.181
0.151
0.129
0.137
0.133
0.135
0.104
0.090
0.096
0.096
0.094
0.094
0.078
0.069
0.079
0.079
0.077
0.077
0.049
0.043
0.063
0.058
0.066
0.065
Cruise 3
0.041
0.044
0.046
0.050
0.045
0.048
0.071
0.077
0.184
0.179
0.037
0.044
0.053
0.059
0.050
0.055
0.076
0.082
0.158
0.077
0.088
0.111
0.130
0.088
0.092
0.082
0.078
0.086
0.079
0.092
0.108
0.128
0.090
0.096
0.078
0.023
0.020
0.027
0.023
0.029
0.034
Cruise 4 Cruise 5
0.065 0.046
0.064 0.048
0.052 0.046
0.053 0.046
0.055 0.047
0.053 0.047
0.097
0.103
0.141
0.141
0.067
0.066
0.048
0.052
0.054
0.052
0.084
0.095
0.128
0.070
0.065
0.041
0.047
0.051
0.049
0.045
0.047
0.043
0.015
0.014
0.017
0.018
0.026
0.023
0.043
0.009
0.009
0.005
0.006
0.004
0.004
Cruise 6
0.058
0.063
0.056
0.047
0.069
0.062
0.088
0.092
0.122
0.122
0.054
0.055
0.065
0.051
0.070
0.064
0.081
0.082
0.115
0.053
0.053
0.059
0.045
0.076
0.064
0.075
0.075
0.075
0.051
0.049
0.063
0.059
0.058
0.057
0.041
0.039
0.030
0.031
0.045
0.051
Cruise 7
0.026
0.024
0.038
0.037
0.040
0.038
0.080
0.090
0.069
0.068
0.026
0.023
0.040
0.038
0.042
0.039
0.080
0.089
0.070
0.052
0.046
0.044
0.046
0.052
0.057
0.037
0.037
0.037
0.047
0.044
0.048
0.047
0.047
0.047
0.040
0.021
0.018
0.025
0.023
0.027
0.024
Cruise 8
0.050
0.047
0.041
0.044
0.043
0.044
0.087
0.094
0.217
0.214
0.049
0.048
0.043
0.044
0.044
0.045
0.073
0.085
0.187
0.021
0.019
0.029
0.031
0.026
0.027
0.028
0.031
0.026
0.010
0.009
0.018
0.017
0.015
0.010
0.031
0.007
0.006
0.012
0.010
0.013
0.009
263
-------
Table 4.4.14. Cruise-Segment Mean Concentrations of PDC (mg/L) for the LMMBP Period
Segment
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
Cruise 1
0.148
0.138
0.182
0.168
0.176
0.177
0.262
0.262
0.262
0.262
0.154
0.144
0.180
0.165
0.176
0.175
0.262
0.262
0.262
0.129
0.122
0.160
0.151
0.202
0.170
0.187
0.196
0.178
0.125
0.122
0.141
0.146
0.154
0.158
0.139
0.083
0.081
0.114
0.117
0.118
0.118
Cruise 2
0.202
0.203
0.224
0.221
0.220
0.219
0.165
0.187
0.215
0.210
0.210
0.208
0.114
0.126
0.147
0.137
0.144
0.141
0.091
0.108
0.086
0.087
0.089
0.090
0.078
0.077
0.080
0.079
0.079
0.079
Cruise 3
0.221
0.231
0.206
0.204
0.159
0.175
0.283
0.306
0.525
0.514
0.210
0.222
0.204
0.206
0.141
0.162
0.295
0.317
0.439
0.145
0.102
0.144
0.128
0.146
0.145
0.167
0.185
0.149
0.149
0.120
0.141
0.130
0.148
0.148
0.186
0.122
0.105
0.079
0.083
0.059
0.057
Cruise 4 Cruise 5
0.222 0.113
0.207 0.113
0.210 0.118
0.202 0.119
0.239 0.119
0.227 0.120
0.357
0.356
0.397
0.397
0.222
0.209
0.197
0.198
0.242
0.225
0.321
0.333
0.374
0.185
0.185
0.152
0.168
0.223
0.220
0.146
0.142
0.149
0.177
0.196
0.126
0.134
0.142
0.142
0.140
0.103
0.097
0.072
0.076
0.064
0.065
Cruise 6
0.139
0.152
0.178
0.142
0.202
0.188
0.237
0.248
0.332
0.333
0.127
0.135
0.177
0.144
0.203
0.189
0.255
0.266
0.320
0.118
0.120
0.163
0.128
0.196
0.176
0.270
0.270
0.270
0.122
0.124
0.159
0.155
0.193
0.198
0.096
0.095
0.088
0.093
0.149
0.176
Cruise 7
0.135
0.130
0.187
0.178
0.195
0.186
0.372
0.438
0.306
0.293
0.133
0.129
0.174
0.170
0.200
0.186
0.374
0.436
0.311
0.158
0.140
0.177
0.162
0.170
0.168
0.183
0.183
0.183
0.113
0.127
0.135
0.134
0.148
0.145
0.173
0.106
0.108
0.091
0.089
0.094
0.091
Cruise 8
0.144
0.139
0.162
0.151
0.150
0.147
0.272
0.295
0.480
0.474
0.138
0.130
0.166
0.152
0.144
0.143
0.239
0.272
0.427
0.114
0.111
0.144
0.141
0.149
0.144
0.180
0.196
0.165
0.080
0.077
0.094
0.083
0.096
0.069
0.196
0.074
0.074
0.062
0.059
0.079
0.061
264
-------
Muskegon River
Grand River
St. Joseph River
Sediment Participate
Organic Carbon
(mg/g dry weight)
1001
90
80*
-
Figure 4.4.5. Lake Michigan sediment sampling Figure 4.4.6. Distribution of POC in Lake
sites for organic carbon during the LMMBP. Michigan surficial sediments (mg/gdw).
265
-------
Sediment Particulate
Organic Carbon
bulk concentration
(mgC/L)
Figure 4.4.7. Distribution of POC in Lake
Michigan surficial sediments (mg/L).
averaged sediment organic carbon concentrations
are presented in Table 4.4.15.
4.4.2.3 Parameterization
There were a number of organic carbon sorbent-
specific parameters that had to be specified as input
for the LM2-Toxic. These parameters included
carbon biotransformation parameters (carbon decay
and yield), carbon sorbent vertical transport
parameters (settling velocities, resuspension rates,
burial rates), and parameters derived during
empirical wave- induced resuspension (critical wave
height and empirical wave coefficient). Most of these
parameters were not project-generated parameters.
Initial values of these parameters were specified
based on values from the literature or were derived
from empirical relationships. Their values could have
a large variation, and results from the LM2-Toxic are
very sensitive to some of these parameters.
Therefore, final values of these parameters were
defined by calibration. Tables 4.4.16 through 4.4.19
present final lists of carbon sorbent-specific
parameters, definitions, units, their final values, and
data sources used in the LM2-Toxic.
The carbon biotransformation parameters are listed
in Table 4.4.16. Initial values of the parameters in
this table were from the GBMBP (Bierman et a/.,
1992; DePinto et at., 1993). Table 4.4.17 presents
segment-specific settling velocities for both BIG and
PDC, respectively. All of the settling velocities were
calibration parameters. More detailed discussion of
the LM2-Toxic model calibration can be found in
Chapter 5. The water column segment-specific
effective PDC concentrations (CJ used in the
steady-state resuspension calculation (Equation
4.3.8) were computed by averaging the eight cruise
segment-mean concentrations derived from samples
collected during the LMMBP. The PDC
concentrations (Cw) are presented in Table 4.4.18.
An accurate estimation of the burial rate was
essential in computing the steady-state resuspension
rate and calibrating organic carbon sorbent
dynamics. A reliable burial rate could reduce one
degree of freedom in Equation 4.3.8, making the
carbon settling velocity the only calibration parameter
in the equation. Table 4.4.19 provides the segment-
specific burial rates along with the thickness of
mixing layer for the surficial sediment segments
based on the analyzed results of the LMMBP-
generated sediment core samples (Robbins et a/.,
1999). Table 4.4.20 lists the segment-specific critical
wave heights and empirical wave coefficients for
calculating wave-induced resuspension based on the
LMMBP-generated data using linear-regression (see
Section 3.4.2 for discussion and details). The final
segment-specific resuspension rates were computed
as daily time functions and input into the LM2-Toxic.
266
-------
Table 4.4.15. Concentration of Organic Carbon Sorbents in Surficial Segments for the LM2-Toxic
Segment
42
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
DOC (mg/L)
1.6
1.6
1.6
1.6
1.6
1.6
1.6
1.6
1.6
1.6
1.6
1.6
1.6
1.6
1.7
1.7
1.7
1.7
1.7
1.7
1.7
1.7
1.7
1.7
1.7
1.7
1.7
PDC (mg/L)
2981.3
3342.3
4190.0
3782.3
6384.2
6056.2
7040.6
4561 .8
4688.9
5354.8
4058.7
1942.4
3009.7
1709.3
1778.6
2906.6
4447.1
5557.6
4768.6
4968.2
4770.0
3519.9
3024.3
6175.6
5771 .8
4533.5
3970.2
Segment
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
DOC (mg/L)
1.7
1.7
1.7
1.7
1.7
1.7
1.7
1.7
1.7
2
2
2
2
2
1.9
1.9
1.9
1.9
2
3.4
3.4
3.4
3.4
3.6
3.6
3.6
PDC (mg/L)
3634.1
4120.2
4340.0
2712.5
2642.3
2626.8
2172.5
2433.5
4322.5
6147.7
5817.2
5705. 1
6422.5
5699.2
5831.2
5987.4
6665.9
6354.6
5672.5
5040. 1
5877.2
6611.8
5917.6
6399.6
5460.0
5293.0
267
-------
Table 4.4.16. Organic Carbon Sorbent Biotransformation Parameters Specified for the LM2-Toxic
Parameter
Definition and Units
Value
Source
d(DOC)
Substrate saturated decay rate of DOC in water
column at temperatures equal to 20°C, d'1
0.005 DePinto era/., 1993;
Calibration
Michaelis-Menten half-saturation constantfor DOC 3.4
decay in water column, mg/L
DePinto etal., 1993;
Calibration
ds(DOC)
Decay rate for DOC in surficial sediments at 0.015
temperature equal to 20°C, d'1
DePinto et ai, 1993;
Calibration
d(BIC)
Substrate saturated decay rate of BIG in water 0.56
column at temperature equal to 20°C, d1
DePinto et al., 1993;
Calibration
Michaelis-Menten half-saturation constant for BIG
decay in water column, mg/L
0.4 DePinto era/., 1993;
Calibration
'(BIC PDC)
Yield coefficient (percentage) of PDC during BIC
decay in water column, dimensionless
90% DePinto et al., 1993;
Calibration
d(PDC)
Substrate saturated decay rate of PDC in water
column at temperature equal to 20°C, d"1
0.04 DePinto etal., 1993;
Calibration
Michaelis-Menten half-saturation constantfor PDC 0.55
decay in water column, mg/L
DePinto etal., 1993;
Calibration
Y(PDC DOC) Yield coefficient (percentage) of DOC during PDC 60% DePinto ef al., 1993;
decay in water column, dimensionless Calibration
^ds(PDC)
Decay rate for PDC in surficial sediments at 0.0001
temperature equal to 20°C, d"1
DePinto etal., 1993;
Calibration
YS(PDC DOC) Yield coefficient (percentage) of PDC decay in
surficial sediments, dimensionless
60% DePinto et al., 1993;
Calibration
268
-------
Table 4.4.17. Segment-Specific Settling Rates (m/d) for Organic Carbon Sorbents (BIC and PDC)
Specified for LM2-Toxic
Segment
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
BIC Settling Velocity
0.15
0.15
0.15
0.15
0.15
0.15
0.2
0.2
0.2
0.2
0.15
0.15
0.15
0.15
0.15
0.15
0.2
0.2
0.2
0.06
0.06
0.06
0.06
0.06
0.06
0.2
0.2
0.2
0.06
0.06
0.06
0.06
0.06
0.06
0.08
0.06
0.06
0.06
0.06
0.06
0.06
PDC Settling Velocity
0.15
0.15
0.15
0.15
0.15
0.15
0.2
0.2
0.2
0.2
0.15
0.15
0.15
0.15
0.15
0.15
0.2
0.2
0.2
0.25
0.25
0.25
0.25
0.25
0.25
0.8
0.8
0.8
0.4
0.4
0.4
0.4
0.4
0.4
0.75
0.75
0.75
0.75
0.75
0.75
0.75
Source
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
Calibration
269
-------
Table 4.4.18. Segment-Specific Effective PDC Concentrations (Cw) Used in the Steady-State
Resuspension Calculation Based on the LMMBP Data
Segment Cw (mg/L)
1 223.63
2 223.74
3 242.84
4 232.14
5 245.25
6 240.86
7 384.32
8 410.02
9 522.45
10 515.67
1 1 236.60
12 233.70
13 253.99
14 243.83
15 257.86
16 252.19
17 373.12
18 402.88
19 386.85
Segment
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
Cw(mg/L)
153.59
145.00
171.26
161.49
193.38
182.99
204.06
210.65
197.85
130.39
132.76
135.10
133.21
147.27
144.40
190.47
97.10
93.47
86.09
87.14
94.54
95.34
Table 4.4.19. Segment-Specific Sediment Accumulation Rates (vb) and Thickness of Mixing Layer (z)
Used in the Steady-State Resuspension Calculation (Original Data Source: Bobbins etal., 1999)
Segment
42
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
vb (m/d)
0
0
6.12E-09
1 .59E-07
4.97E-07
1 .36E-06
3.11E-06
9.94E-06
7.29E-06
3.96E-06
6.37E-06
1 .36E-06
1 .28E-06
1 .87E-07
0
9.87E-08
7.56E-07
2.90E-06
6.06E-06
7.47E-06
7.58E-06
5.06E-06
3.76E-06
3.65E-08
8.78E-08
9.83E-09
4.83E-07
z(m)
0.001
0.001
0.001
0.001
0.013
0.018
0.025
0.031
0.029
0.019
0.012
0.001
0.001
0.001
0.001
0.001
0.013
0.019
0.028
0.026
0.031
0.016
0.001
0.001
0.001
0.001
0.001
Segment
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
vb (m/d)
1.49E-06
3.19E-06
5.31 E-06
0
4.18E-07
9.05E-08
1 .47E-07
3.36E-07
4.94E-07
1 .36E-06
1 .36E-06
1 .36E-06
1 .36E-06
1 .36E-06
1 .36E-06
1 .36E-06
1 .36E-06
1 .36E-06
1 .36E-06
2.72E-06
2.72E-06
2.72E-06
2.72E-06
2.60E-06
2.60E-06
2.60E-06
z(m)
0.001
0.024
0.03
0.001
0.011
0.01
0.001
0.001
0.001
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
270
-------
Table 4.4.20. Segment-Specific Critical Wave Heights (Wcr) and Empirical Wave Coefficients (a) Used
in the Wave-Induced Resuspension Calculation Based on the LMMBP Data
Segment
42
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
Wcr(m)
0.93
1.20
1.63
2.24
3.75
4.01
5.52
4.83
4.50
4.15
2.28
2.22
1.62
1.29
1.74
2.29
3.84
5.02
7.89
7.66
5.47
3.81
2.40
0.95
1.22
1.63
2.19
a
52
70
131
565
0
0
0
0
0
0
61
257
56
88
0
0
0
0
0
0
0
0
0
96
143
520
0
Segment
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
Wcr(m)
3.49
3.33
3.90
3.57
3.76
3.35
2.21
1.65
1.24
0.90
1.22
1.57
1.60
1.69
2.07
1.94
1.65
1.25
0.85
0.92
1.22
1.31
1.33
0.74
0.82
0.82
a
0
0
0
0
0
0
0
0
214
99
207
0
0
0
0
0
0
142
175
273
0
0
0
142
157
146
4.4.3 PCBs
Thirty-six PCB state variables representing a total of
54 PCB congeners were modeled in the LM2-Toxic
as either individual PCB congeners or co-eluting PCB
congeners. The mass for the sum of all 54 PCB
congeners (ZPCB) accounted for approximately 70%
of total PCB mass in Lake Michigan. Table 4.4.21
provides a list of these PCB congeners by IUPAC
numbers. Because of the huge amount of PCB
congener data generated during the LMMBP, it was
impossible to provide summaries of field data for
each modeled PCB congener. Thus, PCB loadings,
field data, and related information in the following
subsections are reported only as the sum of the 54
congeners (IPCBs). Organic carbon sorbent
dynamics, PCB partitioning, and PCB air-water
exchange dictated the variation of PCB
concentrations in Lake Michigan. The lake-wide
average concentrations for vapor phase IPCBs
varied from 0.28 to 0.42 ng/m3. The average
concentrations in the main lake for dissolved and
particulate IPCBs were 0.15 and 0.07 ng/L,
respectively. In Green Bay, the average
concentrations of dissolved and particulate IPCBs
were almost double or triple, and their values were
0.3 and 0.36 ng/L, respectively. The average IPCB
concentrations in the surficial sediments ranged from
650 to 25,000 ng/L. Data requirements for PCB input
to the LM2-Toxic included loads (tributary and
271
-------
Table 4.4.21. List of PCB State Variables Modeled in the LM2-Toxic
PCB Congeners
PCB Congeners
PCB Congeners
PCB Congeners
PCB8+5
PCB15+17
PCB16+32
PCB18
PCB26
PCB28+31
PCB33
PCB37+42
PCB44
PCB49
PCB52
PCB56+60
PCB66
PCB70+76
PCB74
PCB77+110
PCB81
PCB87
PCB92+84
PCB89
PCB99
PCB101
PCB118
PCB123+149
PCB1 05+1 32+1 53
PCB151
PCB163+138
PCB170+190
PCB172+197
PCB180
PCB187+182
PCB1 95+208
PCB1 96+203
PCB201
PCB85
PCB146
atmospheric), boundary conditions (Straits of
Mackinac and atmospheric gas phase
concentrations), initial conditions in both the water
column and surficial sediments, process related
parameters, and kinetic time functions such as
temperature and wind speed. The segment-mean
concentration for each cruise, generated by IDWdata
interpolation with VWA, was an essential part of field
data analysis used to calibrate the LM2-Toxic. See
Part 4, Appendix 4.4.1.
4.4.3.1 Loading
Tributary and atmospheric sources were the two
major external loads of PCBs to Lake Michigan.
Tributary loads included loads from 11 monitored
tributaries and 18 unmonitored tributaries (Hall and
Robertson, 1998). Atmospheric loads were derived
from monthly atmospheric dry and wet deposition
fluxes (McCarty et al., 2004; Miller et al., 2001).
Tributary loads were input as daily time functions into
the LM2-Toxic. The seasonal variation of the IPCB
tributary load represents the sum of the tributary
loads including 11 monitored and 18 unmonitored
tributaries (Figure 4.4.8). There were 18 months of
atmospheric dry deposition measurements during the
LMMBP period. The data set provided for the
modeling was organized for the period from April
1994 to March 1995. For consistency with other
input loading time series in the model, the loads for
the first three months of 1995 were used as the loads
for the same period of 1994. The loads for April-
December of 1995 were assumed to be identical to
the loads for the same period of 1994. Eighteen
3.0
2.5
2.0
Lake Michigan IPCB tributary load
(11 monitored, 18 unmonitored)
i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—r .
Jan94 Jun94 Dec94 Jun95 Nov95
Figure 4.4.8. IPCB tributary (11 monitored and
18 unmonitored tributaries) loads to Lake
Michigan during the LMMBP period.
months of monthly atmospheric wet deposition were
available starting in April 1994 and ending in
September 1995. Among the 18 months, the loads
in April and May of 1994 were significantly higher
(about 10 times higher) than the monthly loads of the
remaining 16 months. Therefore, the approach used
in the first three months of 1994 for dry loads was not
applicable for wet deposition. In order to construct a
complete two-year atmospheric wet load time series,
monthly wet loads for January, February, and March
of 1994 were assumed to be the same as that for
April 1994, and the loads for the last three months in
1995 were assumed to be the same as the loads for
the last three months of 1994. Figure 4.4.9 provides
the information on lake-wide temporal variation of
total atmospheric deposition (dry + wet) for the
IPCBs. Based on the loads generated for the two-
year LMMBP period, the annual average tributary
272
-------
1 g 1 ato IMi^hinan VDOR
-D ' °
CD 1 K
O
., 14.
"a> 1.2
a. -a 1 0
§ ^ 0.8
ro 0.6
S 0.4-
a. 0.2 j
w o-
Ja
atmosphRrir, (dry
/ \
and wet) load
I
I
I
I — /^, / \ XA
17 \^J ^ ^ \^
n94 Jun94 Dec94
Jun95 Nov95
used to generate segment-specific sediment
concentrations of individual PCB congeners.
Equation 4.4.4 was used to convert sediment PCB
measurements from the unit of ng/gdw to the standard
unit (ng/L, i.e., bulk concentration) used in the LM2-
Toxic.
C* = Csb * p * (1 -<(>) * 103
where
(4.4.4)
Figure 4.4.9. Estimated ZPCB atmospheric loads
including dry and wet deposition into Lake
Michigan during the LMMBP period.
and atmospheric loads of the ZPCBs are presented
in Tables 4.4.22 and 4.4.23, respectively. For the
two-year LMMBP period, the ZPCB external loads to
Lake Michigan from tributary and atmospheric loads
were roughly equal.
Cs" =
C," =
PCB concentrations in surficial sediments
(ng/L), the unit required for LM2-Toxic
PCB concentrations in surficial sediments
(ng/gdw), the unit of field measurements
= bulk density of surficial sediments
constant (2.45 gdw/cm3). To convert cm3 to L
(liter), a factor
equation.
of 103 is needed in the
4.4.3.2 Field Data,
Boundary Conditions
Initial Conditions, and (p = porosity (dimensionless)
PCB congener level data were collected at eight
shoreline sites (Figure 4.4.10), 41 water column
stations (Figure 4.4.3), and about 120 sediment
locations (Figure 4.4.5) during the LMMBP. The
LMMB Methods Compendium provides detailed
descriptions of sampling methods and analysis
procedures (U.S. Environmental Protection Agency,
1997). The LM2-Toxic required inputs of PCB
congener concentrations in air, water, and
sediments. Daily vapor phase PCB boundary
concentrations were interpolated and calculated at a
spatial resolution of 5 x 5 km2 grid cells (Green et a/.,
2000; Miller et al., 2001) based on air samples
collected at eight shoreline sites and 14 over-water
sites (Figure 4.4.10). These results were aggregated
to the LM2-Toxic segmentation for input into the
model. Strong seasonal variation was observed for
PCB vapor phase concentrations during the study
period (Figure 4.4.11). Table 4.4.24 presents the
segment-specific annual average ZPCB vapor phase
concentrations above the 10 surface segments of the
LM2-Toxic. The same interpolation methods and
averaging procedures used for organic carbon
species were applied to individual PCB congeners in
both the water column and sediments. One hundred
and sixteen sediment samples (Figure 4.4.12) were
PCB concentrations for the sediment segments in
Green Bay were estimated based on only four
sediment samples (SD89, 95G1, SD106P, and
113G1) collected during the two-year LMMBP period.
Figures 4.4.13 and 4.4.14 show the distribution of
ZPCBs in Lake Michigan sediments in units of ng/gdw
and ng/L, respectively. Table 4.4.25 presents the
water column initial concentrations expressed as the
sum (ZPCBs) of the individual PCB congener initial
concentrations used in the LM2-Toxic. The cruise-
segment mean concentrations for ZPCBs are listed
in Table 4.4.26. The segment-specific sediment
concentrations for ZPCBs are provided in Table
4.4.27. The boundary conditions for individual PCB
congeners used the measurements taken at a
sampling location in Lake Huron, LM54M.
4.4.3.3 Parameterization
The PCB congener-specific parameters that must be
specified for input to the LM2-Toxic include partition
coefficients, molecular weights, enthalpy, and
entropy. The POC partition coefficients, K'POc,a.
were computed initially using the two-phase
partitioning model (Equation 4.3.14) for each
selected PCBs computed initially using the two-phase
273
-------
Table 4.4.22. Annual Average IPCB Loads From 11 Monitored and 18 Unmonitored Tributaries to Lake
Michigan During the LMMBP
Tributary (Monitored) IPCB Loads (kg/year) Tributary (Unmonitored) ZPCB Loads kg/year)
Milwaukee River
Sheboygan River
Calumet River
St. Joseph River
Kalamazoo River
Grand River
Muskegon River
Pere Marquette River
Manistique River
Menominee River
Fox River
Total
9.00
9.40
29.86
8.79
30.31
8.60
1.71
0.51
1.17
4.01
174.62
277.98
Manitowac
Root
Galien
Black
Kalamazoo-Minor
Kewaunee
Pere Marquette-Minor
Manistee
Betsie
Millecoquins
Jordan
Sturgeon
Whitefish
Escanaba
Cedar
Peshtigo
Oconto
Pensaukee
1.20
4.26
1.22
9.30
0.80
0.75
0.89
1.41
0.51
0.45
1.04
0.45
0.62
0.91
1.02
1.13
1.02
0.50
Total
21.84
Table 4.4.23. Annual Average IPCB Atmospheric Dry and Wet Loads in the 10 Surface Water Column
Segments of Lake Michigan During the LMMBP
Segment
Total Loads
IPCB Atmospheric Dry Loads IPCB Atmospheric Wet Loads
(kg/year) (kg/year)
1 18.13
2 13.90
3 12.27
4 9.20
5 4.38
6 4.12
7 1.36
8 1.26
9 1.47
10 0.37
35.67
34.69
22.29
16.89
9.04
8.46
2.54
2.34
2.79
0.65
66.46
135.35
274
-------
Manitowoc
Kenosha
IIT-Chic
Sleeping Bear
Dunes
based sampling
water based sampling
South Haven
Figure 4.4.10. Lake Michigan atmospheric sampling sites during the LMMBP.
275
-------
SPCB vapor concentratior
(ng/m3)
1 .U
0 9
0.8
0 7
0 6
0.5
0.4
0.3
0.2
0.1
n
Lake Michigan 2.PCB
vapor concentration A
\
/ \ / \
/ \ / \
/ \ / \
/ \ J \
Jan94 Jun94 Dec94 Jun95 Nov95
Figure 4.4.11. Seasonal variation of IPCB vapor phase concentrations observed during the LMMBP.
Table 4.4.24. Annual Average Boundary Conditions of IPCB Vapor Phase Concentrations for Lake
Michigan During the LMMBP
Segment
ZPCBs (ng/m3)
1
2
3
4
5
6
7
8
9
10
0.418367
0.374108
0.297252
0.284844
0.299123
0.282733
0.30676
0.307972
0.312705
0.300599
276
-------
Menominee River
Pere Marquette River
Milwaukee
Sheboygan
River
Muskegon River
Grand River
Kalamazoo River
w o *-*
D
' 0 °
0 D
N o
*
\ (-»
D D / \
an /
9] /
• °A
x . l^ . _.
Core types
D box
O ponar
• gravity
St. Joseph River
Figure 4.4.12. Lake Michigan sediment sampling sites for PCBs during the LMMBP.
277
-------
Total PCS
(ng/g dry weight)
Figure 4.4.13. Distribution of IPCBs in Lake
Michigan surficial sediments (ng/gdw).
Figure 4.4.14. Distribution of IPCBs in Lake
Michigan surficial sediments (ng/L).
Table 4.4.25. Initial Concentrations of IPCBs in Water Column Segments for Lake Michigan
Segment
IPCBs (ng/L)
Segment
ZPCBs (ng/L)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
0.1903
0.2414
0.2264
0.2680
0.1597
0.2350
0.5910
0.6349
2.4938
2.5051
0.1141
0.1888
0.2106
0.1682
0.1859
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
0.1401
0.1872
0.2053
0.1438
0.1621
0.1604
0.1388
0.1639
0.1643
0.1492
0.2039
0.1663
0.1148
0.1560
0.1531
278
-------
Table 4.4.26. Cruise-Segment Mean Concentration of ZPCBs (ng/L) for the LMMBP Period
Segment
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
Cruise 1
0.192
0.372
0.227
0.269
0.214
0.236
0.591
0.807
2.494
2.505
0.154
0.194
0.212
0.227
0.196
0.199
0.550
0.847
2.373
0.184
0.188
0.212
0.181
0.170
0.179
0.550
0.847
2.373
0.210
0.204
0.167
0.162
0.162
0.161
0.847
0.170
0.178
0.111
0.117
0.117
0.118
Cruise 2
0.124
0.172
0.136
0.138
0.148
0.150
0.133
0.180
0.138
0.140
0.150
0.152
0.146
0.180
0.145
0.152
0.154
0.157
0.166
0.206
0.144
0.150
0.154
0.157
0.137
0.146
0.112
0.113
0.108
0.108
Cruise 3
0.136
0.136
0.112
0.103
0.078
0.083
0.195
0.186
0.486
0.475
0.129
0.145
0.113
0.107
0.084
0.085
0.293
0.321
1.327
0.179
0.173
0.142
0.128
0.122
0.122
0.293
0.321
1.327
0.178
0.174
0.143
0.130
0.126
0.127
0.293
0.200
0.201
0.116
0.125
0.103
0.102
Cruise 4 Cruise 5
0.199 0.241
0.241 0.230
0.158 0.232
0.157 0.232
0.116 0.233
0.123 0.233
0.211
0.221
0.839
0.833
0.190
0.240
0.168
0.160
0.119
0.125
0.196
0.221
0.733
0.136
0.170
0.168
0.155
0.111
0.116
0.196
0.221
0.733
0.359
0.410
0.213
0.224
0.189
0.197
0.196
0.226
0.229
0.201
0.202
0.147
0.132
Cruise 6
0.214
0.249
0.181
0.206
0.130
0.155
0.376
0.452
1.039
1.043
0.217
0.236
0.171
0.199
0.132
0.155
0.362
0.465
0.997
0.219
0.216
0.194
0.218
0.185
0.196
0.362
0.465
0.997
0.171
0.177
0.145
0.151
0.136
0.135
0.362
0.156
0.161
0.126
0.126
0.128
0.130
Cruise 7
0.183
0.211
0.169
0.171
0.166
0.166
0.199
0.199
0.199
0.199
0.184
0.211
0.169
0.171
0.167
0.166
0.199
0.199
0.199
0.296
0.281
0.235
0.221
0.239
0.217
0.199
0.199
0.199
0.251
0.259
0.241
0.228
0.269
0.260
0.199
0.189
0.200
0.146
0.149
0.151
0.149
Cruise 8
0.233
0.154
0.143
0.109
0.117
0.114
0.170
0.185
0.660
0.648
0.240
0.138
0.091
0.088
0.092
0.094
0.154
0.179
0.564
0.350
0.296
0.179
0.286
0.197
0.217
0.154
0.179
0.564
0.341
0.243
0.178
0.194
0.162
0.161
0.154
0.289
0.194
0.140
0.143
0.151
0.156
279
-------
Table 4.4.27. Initial Concentration of IPCBs in Sediment Segments for Lake Michigan
Segment
42
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
ZPCBs (ng/L)
11238.3
10438.8
8680.7
9765.4
8451 .6
8349.3
12396.8
14131.3
9141.6
9960.8
13205.0
2528.2
8925.6
3730.8
626.7
1447.1
3814.2
4599.3
6756.3
8685.5
5617.1
4717.8
3818.2
2183.8
3071.1
2872.5
2148.1
Segment
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
IPCBs (ng/L)
2178.5
2453.6
1740.4
1711.2
1877.1
1501.6
1556.9
1802.7
2855.3
12509.3
13727.2
7267.3
10515.1
9084.9
7084.6
9136.2
10282.1
11444.5
11787.7
21459.7
15602.5
25096.6
20717.4
7957.1
9250.4
8975.4
partitioning model (Equation 4.3.14) for each
selected PCB congener. A number of studies have
shown a strong relationship between measured
organic carbon partition coefficients and the octanol-
water partition coefficient (Kow) for PCBs in natural
water systems (Karickhoff, 1981; Hassett et a/.,
1980; Di Toro, 1985; Thomann and Mueller, 1987).
For the purpose of comparison in general variation
trend, the set of estimated POC partition coefficients,
K'POc,a> were Plotted (Figure 4.4.15) against the Kow
calculated by Hawker and Connell (1988) for 209
PCB congeners. Figure 4.4.15 illustrates the strong
relationship between the POC partition coefficients
estimated from the two-phase partitioning model and
Kow computed by Hawker and Connell (1988). This
result supported the conclusion of Eadie etal. (1990)
that "ambient Great Lakes POC is similar to octanol
as a substrate for binding" and indicated the
reasonable accuracy of the initially estimated K'POc,a
coefficients from the two-phase partitioning model.
There were some exceptions for less-chlorinated and
high-chlorinated congeners. Assuming the DOC
partition coefficients (KDOC) has a similar variation
trend as K'POc,a and that it is about two orders
magnitude less than K'POc a (Carter and Suffet, 1982;
Landrum et a/., 1984,1987; Hassetand Milicic, 1985;
Chiou etal., 1986, 1987; Eadie etal., 1990, 1992;
Bierman et a/., 1992), the KDOC for each PCB
congener were estimated by multiplying the value of
KVoc,a by a factor of 10'2. When necessary, the
values of the pair of partition coefficients for a PCB
congener were adjusted during the course of the
LM2-Toxic PCB calibration. Table 4.4.28 presents
the final values of POC and DOC partition
coefficients used in the LM2-Toxic.
280
-------
Hawker's log Kow vs Two Phase log K'poc.a Partition Coefficients
8.0-
7.5'
• log Kow
Dlog K'poc.a
5.0
0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170 180 190 200
PCB Congener
Figure 4.4.15. Comparison between the estimated log K'POc,afor tne LMMBP selected PCB congeners
based on the two-phase partitioning model and Kow calculated by Hawker and Connell (1988) for all
209 PCB congeners.
Air-water exchange was another crucial process
affecting PCB concentrations in Lake Michigan.
Parameters related to volatilization were molecular
weight, enthalpy, and entropy of each PCB congener.
Table 4.4.29 provides a listing of molecular weights,
enthalpy, and entropy, including their sources for
each PCB congener or co-eluting PCB congeners
required for the LM2-Toxic. During the LM2-Toxic
PCB calibration, the values of enthalpy for some PCB
state variables were adjusted within their allowed
ranges (i.e., one standard error). It is worth noting
that the enthalpy and entropy for co-eluting PCB
congeners were estimated by simple arithmetic
averaging of the values for the individual congeners.
Because quantitative information for the mass of
each PCB congener in a co-eluting PCB mixture was
lacking, a certain amount of uncertainty was
associated with their estimated enthalpy and entropy.
Thus, some adjustments on these parameters were
done during the LM2-Toxic PCB calibration.
4.4.3.4 Kinetic Time Functions
The kinetic time functions for the LM2-Toxic included
reaeration, segment-specific wind speed time
functions, segment-specific water temperature time
functions, and segment-specific air temperature time
functions (Schwab and Beletsky, 1998).
References
Bamford, H.A., D.L Poster, R.E. Huie, and J.E.
Baker. 2002. Using Extrathermodynamic
Relationships to Model the Temperature
Dependence of Henry's Law Constants of 209
PCB Congeners. Environ. Sci. Technol.,
36(20):4395-4402.
Baun, K. 1982. Alternative Methods of Estimating
Pollutant Loads in Flowing Water. Wisconsin
Department of Natural Resources, Madison,
Wisconsin. Technical Bulletin 133, 11 pp.
Beale, E.M.L. 1962. Some Uses of Computers in
Operational Research. Industrielle Organisation,
31:51-52.
Bierman, V.J., Jr., J.V. DePinto, T.C. Young, P.W.
Rodgers, S.C. Martin, and R. Raghunathan.
1992. Development and Validation of an
Integrated Exposure Model for Toxic Chemicals
in Green Bay, Lake Michigan. Final Report. U.S.
Environmental Protection Agency, Office of
Research and Development, ERL-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
381 pp.
281
-------
Table 4.4.28. Final Partition Coefficients for the LMMBP Selected PCBs Used in the LM2-Toxic
PCB Congeners
5+8
15+17
16+32
18
26
28+31
33
37+42
44
49
52
56+60
66
70+76
74
77+110
81
92+84
85
87
89
99
101
105+132+153
118
123+149
163+138
146
151
170+190
172+197
180
1 87+1 82
1 95+208
1 96+203
201
log K'POc.a1(M
-------
Table 4.4.29. Values of Parameters Used for Air-Water Exchange in the LM2-Toxic for the LMMBP
Selected PCB Congeners (Bamford et al., 2002)
PCB Congeners Molecular Weight (g/mol)
5+8
15+17
16+32
18
26
28+31
33
37+42
44
49
52
56+60
66
70+76
74
77+110
81
92+84
85
87
89
99
101
105+132+153
118
123+149
163+138
146
151
170+190
172+197
180
187+182
195+208
196+203
201
223.10
240.30
257.54
257.54
257.54
257.54
257.54
274.77
291 .99
291 .99
291 .99
291 .99
291.99
291 .99
291.99
309.21
291.99
326.43
326.43
326.43
326.43
326.43
326.43
349.40
326.43
343.66
360.88
360.88
360.88
395.32
412.55
395.32
395.32
446.99
446.99
429.77
Enthalpy (kJ/mol)
46.0
46.5
44.5
35.0
41.0
34.0
43.0
41.0
26.0
25.0
32.5
28.5
27.5
27.0
25.0
35.0
33.0
25.0
21.0
33.0
27.0
22.0
30.0
61.0
44.0
47.5
80.0
61.0
32.0
145.0
137.0
137.0
97.0
146.0
143.0
138.0
Entropy (kJ/mol °K)
0.115
0.120
0.110
0.080
0.100
0.085
0.110
0.095
0.050
0.050
0.070
0.065
0.060
0.060
0.050
0.090
0.070
0.040
0.050
0.070
0.040
0.020
0.070
0.190
0.130
0.135
0.250
0.170
0.100
0.485
0.460
0.450
0.295
0.250
0.490
0.460
283
-------
Carter, C.W. and I.H. Suffet. 1982. Binding of DDT
to Dissolved Humic Materials. Environ. Sci.
Technol., 16(11):735-740.
Chapra, S.C. and K.H. Reckhow (Eds.). 1983.
Engineering Approaches for Lake Management,
Volume 2: Mechanistic Modeling. Ann Arbor
Science Publishers, Ann Arbor, Michigan. 492
pp.
Chiou, C.T., R.L Malcolm, T.I. Brinton, and D.E. Kile.
1986. Water Solubility Enhancement of Some
Organic Pollutants and Pesticides by Dissolved
Humic and Fulvic Acids. Environ. Sci. Technol.,
20(5):502-508.
Chiou, C.T., D.E. Kile, E.I. Brinton, R.L. Malcolm, and
J.A. Leenheer. 1987. A Comparison of Water
Solubility Enhancements of Organic Solutes by
Aquatic Humic Materials and Commercial Humic
Acids. Environ. Sci. Technol., 21 (12):1231-1234.
Conn, T.A., L.L. DeLong, E.J. Gilroy, R.M. Hirsch,
and O.K. Wells. 1989. Estimating Constituent
Loads. Water Resources Res., 25:937-942.
DePinto, J.V., R. Raghunathan, P Sierzenga, X.
Zhang, V.J. Bierman, Jr., P.W. Rodgers, and T.C.
Young. 1993. Recalibration of GBTOX: An
Integrated Exposure Model for Toxic Chemicals
in Green Bay, Lake Michigan. Final Report. U.S.
Environmental Protection Agency, Office of
Research and Development, ERL-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
132pp.
Di Toro, D.M. 1985. A Particle Interaction Model of
Reversible Organic Chemical Sorption.
Chemosphere, 14(10):1503-1538.
Eadie, B.J., N.R. Morehead, and P.P. Landrum.
1990. Three-Phase Partitioning of Hydrophobic
Organic Compounds in Great Lakes Water.
Chemosphere, 20(1/2):161-178.
Eadie, B.J., N.R. Morehead, J.V. Klump, and P.P.
Landrum. 1992. Distribution of Hydrophobic
Organic Compounds Between Dissolved and
Particulate Organic Matter in Green Bay Waters.
J. Great Lakes Res., 18(1):91-97.
Eadie, B.J. and J.A. Robbins. 2004. Composition
and Accumulation of Lake Michigan Sediments.
In: M. Munawar andT. Edsall (Eds.), The State of
Lake Michigan. In press.
Endicott, D.D., W.L. Richardson, and D.J. Kandt.
2005. 1992 MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan. In: R. Rossmann (Ed.),
MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan, Part 1. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
EPA/600/R-05/158, 140 pp.
Green, M.L., J.V. DePinto, C.W. Sweet, and K.C.
Hornbuckle. 2000. Regional Spatial and
Temporal Interpolation of Atmospheric PCBs:
Interpretation of Lake Michigan Mass Balance
Data. Environ. Sci. Technol., 34(9):1833-1841.
Hall, D. and D. Robertson. 1998. Estimation of
Contaminant Loading from Monitored and
Unmonitored Tributaries to Lake Michigan for the
USEPA Lake Michigan Mass Balance Study.
Quality Systems and Implementation Plan.
Submitted October23,1998. U.S. Environmental
Protection Agency, Great Lakes National
Program Office, Chicago, Illinois. 19 pp.
Hassett, J.P., J.C. Means, W.L. Barnwart, and S.G.
Wood. 1980. Sorption Properties of Sediments
and Energy-Related Pollutants. U.S.
Environmental Protection Agency, Environmental
Research Laboratory, Athens, Georgia.
EPA/600/3-80/041, 133 pp.
Hasset, J.P. and E. Milicic. 1985. Determination of
Equilibrium and Rate Constants for Binding of a
Polychlorinated Biphenyl Congener by Dissolved
Humic Substances. Environ. Sci. Technol.,
19(7):638-643.
Hawker, D.W. and D.W. Connell. 1988. Octanol-
Water Partition Coefficients of Polychlorinated
Biphenyl Congeners. Environ. Sci. Technol.,
22(4):382-387.
284
-------
Karickhoff, S.W. 1981. Semi-Empirical Estimation of
Hydrophobic Pollutants on Natural Sediments
and Soils. Chemosphere, 10(8):833-846.
Landrum, P.P., S.R. Nihart, B.J. Eadie, and W.S.
Gardner. 1984. Reverse-Phase Separation
Method for Determining Pollutant Binding to
Aldrich Humic Acid and Dissolved Organic
Carbon of Natural Waters. Environ. Sci.
Technol., 18(3): 187-192.
Landrum, P.P., S.R. Nihart, B.J. Eadie, and L.R.
Herche. 1987. Reduction in Bioavailability of
Organic Contaminants to the Amphipod
Pontoporeia hoyiby Dissolved Organic Matter of
Sediment Interstitial Waters. Environ. Toxicol.
Chem., 6(1 ):11-20.
Lesht, B.M. 1988a. Comparison of Two Computer
Programs for Volume-Weighted Averaging of
Limnological Data. Final Report. U.S.
Environmental Protection Agency, Great Lakes
National Program Office, Chicago, Illinois.
Lesht, B.M. 1988b. Nonparametric Evaluation of the
Size of Limnological Sampling Networks:
Application to the Design of a Survey of Green
Bay. J. Great Lakes Res., 14(3):325-337.
McCarty, H.B., J. Schofield, K. Miller, R.N. Brent, P.
Van Hoff, and B. Eadie. 2004. Results of the
Lake Michigan Mass Balance Study:
Polychlorinated Biphenyls and frans-Nonachlor
Data Report. U.S. Environmental Protection
Agency, Great Lakes National Program Office,
Chicago, Illinois. EPA/905/R-01/011, 289 pp.
Miller, S.M., M.L. Green, J.V. DePinto, and K.C.
Hornbuckle. 2001. Results from the Lake
Michigan Mass Balance Study: Concentrations
and Fluxes of Atmospheric Polychlorinated
Biphenyls and frans-Nonachlor. Environ. Sci.
Technol., 35(2):278-285.
Richards, *R.P. 1998. Estimation of Pollutant Loads
in Rivers and Stream: A Guidance Document for
NPS Programs. U.S. Environmental Protection
Agency, Region VIII, Denver, Colorado. 108 pp.
Richardson, W.L., D.D. Endicott, R.G. Kreis, Jr., and
K.R. Rygwelski (Eds.). 2004. The Lake Michigan
Mass Balance Project Quality Assurance Plan for
Mathematical Modeling. Prepared by the
Modeling Workgroup. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
EPA/600/R-04/018, 233 pp.
Robbins, J.A., N.R. Morehead, R.W. Rood, D.N.
Edgington, and S. Meyer. 1999. Accumulation
and Near-Surface Mixing of Sediments in Lake
Michigan as Determined for the Lake Michigan
Mass Balance Program, Volumes 1 and 2.
Project Report. U.S. Environmental Protection
Agency, Office of Research and Development,
ERL-Duluth, Large Lakes Research Station,
Grosse lie, Michigan. 503 pp.
Schwab, D. and D. Beletsky. 1998. Lake Michigan
Mass Balance Study: Hydrodynamic Modeling
Project. National Oceanic and Atmospheric
Administration, Great Lakes Environmental
Research Laboratory, Ann Arbor, Michigan.
NOAATechnical Memorandum ERLGLERL-108,
53pp.
Thomann, R.V. and J.A.Mueller. 1987. Principles of
Water Quality Modeling and Control. Harper and
Row Publishers, New York, New York. 644 pp.
U.S. Environmental Protection Agency. 1997. Lake
Michigan Mass Balance Study (LMMB) Methods
Compendium, Volume 1: Sample Collection
Techniques. U.S. Environmental Protection
Agency, Great Lakes National Program Office,
Chicago, Illinois. EPA/905/R-97/012a, 1,440pp.
285
-------
PART 4
LM2-TOXIC
Appendix 4.4.1 Sample Data Interpolation
for the LMMBP
Xiangsheng Xia
Computer Sciences Corporation
Large Lakes Research Station
9311 Groh Road
Grosse lie, Michigan 48138
Many sample data sets of physical and chemical
parameters collected for the Lake Michigan Mass
Balance Project (LMMBP) were often sparse and
occurred on irregular grids. For modeling purposes,
values of these parameters were needed on a 5 x 5
km grid. This presented a problem of using sample
data to estimate or predict values in areas which
were not sampled. Thus, some interpolation
mechanisms based on "insufficient" samples were
needed to bridge the gap between the desired and
the reality world of data collection. Distance square
inverse and natural-neighbor interpolation methods
were carefully studied and applied to sample data
analysis for this project.
The distance-weighted-averaging approach
determines the estimated values at grid points as
the sum of weighted values of the individual sample
datum. In general, the closer a datum point to the
grid point to be estimated, the greater influence the
datum at that point exerts. It is a method
characterized as a global approach. The distance-
weighted-averaging method is well understood and
widely accepted by scientists in various fields. It is
also easy to implement. The major disadvantage of
this method has been its tendency to smooth out all
small variations in the relatively small local area.
Therefore, it is not very well suited to find the trend of
samples in small local areas. The distance-weighted-
averaging interpolation is compromised by its
essentially one-dimensional nature. Although the
interpolating surface is smooth, it cannot, for
instance, form ridges or domes from sparse data.
Furthermore, distance-weighted-averaging is unable
to infer (or extrapolate) a surface lying above or
below the range of sample values. In general, the
estimation computed by distance-weighted-averaging
lies between the maximum and minimum of the
sample data.
Neighborhood-based interpolation, on the other
hand, is a local approach which utilizes all the
(natural) neighbors of the sample points. The natural-
neighbor method can infer values outside the known
range. It is unique for a given data configuration and
choice of blending function parameters. If used
properly on dense data sets, neighborhood-based
interpolation can provide much richer information
such as rapid changes, ridges, or dams in smaller
areas. However, neighborhood-based interpolation,
in contrast to distance-based methods, is much more
complicated to implement and harder to understand.
In case an ambiguity or unexpected phenomena
arise from a neighborhood-based interpolation, it may
require a knowledgeable person to make reasonable
interpretation of results.
During the course of the LMMBP data analysis
process, distance square inverse interpolation
combined with application codes written in
Interactive Data Language (IDL) were used
intensively to find the interpolated values of a 5 x 5
km grid of Lake Michigan for various parameters,
286
-------
such as polychlorinated biphenyls (PCBs), atrazine,
nutrients, etc. On the other hand, natural-neighbor
interpolation was primarily used for sediment data
analysis where sample locations were relatively
dense.
More details of distance square inverse and natural-
neighbor interpolations are presented in the next two
sections. Many applications of interpolation have
been developed, including contour plots, volume-
weighted averages, and others. These are
discussed in Section 4.4.1.4. Some problems
applying natural-neighbor interpolation are discussed
in Section 4.4.1.5.
A4.4.1.1 The Distance Square Inverse
Method
The inverse distance to a power method is a
weighted-average interpolation. Data are weighted
during interpolation such that the influence of one
sample point relative to another declines with
increasing distance from the grid node. Weighting is
assigned to data using a weighting power that
controls how the weighting factor drops off as
distance from a grid node increases. As the power
increases, the grid node value approaches the value
of the neighboring data points. The weighting power
determines how quickly weighting falls off with
distance from the grid node. As the power
approaches zero, the generated surface approaches
a horizontal planar surface through the average of all
observations from the data file. As the power
increases, the generated surface is a "nearest
neighbor" interpolation, and the resultant surface
becomes polygons which represent the nearest
observation to the interpolated node. Power values
are usually between one and three to avoid extreme
results. Distance square inverse is the distance-
weighted method with the power chosen as two.
The smoothing factor parameter allows one to
incorporate an uncertainty factor associated with
sample data. The larger the smoothing factor
parameter, the less influence a particular
observation has in computing a neighboring grid.
The smoothing factor for this study was 2.5 (miles).
The equation in the inverse distance square method
is:
(A4.4.1.1)
where
/' = runs for all samples
v, - interpolated value at grid point i,
Cj = value of sample j,
dy = distance between grid point i and sample
location j,
r0 = smoothing factor
n = total number of samples being considered in
the interpolation
An IDL code which implements the inverse distance
square interpolation scheme was received from
David Schwab (National Oceanic and Atmospheric
Administration, Great Lakes Environmental
Research Laboratory). This was further developed for
the LMMBP data analysis.
A4.4.1.2 The Natural-Neighborhood
Method
Natural-neighbor interpolation offers a different
approach to spatial interpolation and extrapolation.
It has good mathematical properties and offers more
flexibility than the distance square inverse method.
All interpolation methods involve, to some extent, the
idea that the value of the interpolated point should
depend more on data values at nearby data sites
than at distant ones. In natural-neighbor
interpolation, the idea of neighbors in a spatial
configuration is formalized in a natural way and made
quantitative, and the properties of the method
depend on an apparently new geometrical identity
relating this quantitative measure of neighbors to
position.
287
-------
Any two data are natural-neighbors if there is a
location or region that is equally close to each of the
pair, and no other datum is closer. Any three or
more data on the plane are natural-neighbors if no
other datum lies within their circum-circle. The
spatial relationships determined by a set of natural-
neighbors circles have two common and widely
known graphical representations. These are Voronoi
tessellation (of Voronoi polygons) and Delaunay
triangulations. The Voronoi tessellation illustrates
that each datum has a unique natural-neighbor
region associated with it and is bounded by halfway
interfaces of that datum with its natural-neighbor.
Neighborhood coordinates are local coordinates
relating the position of the interpolation point to reach
a datum in the neighborhood subset. These
coordinates (weights for the interpolation), ranging
between zero and one, are proportional to areas
defined on natural-neighbor regions for each of the
data. Such coordinates are superior to distance-
based coordinates. Distance-based coordinates
make no allowance for the distances to the other
data; that is, distance-based interpolation is not
sensitive to a changing spatial context. Finally,
natural-neighbor interpolation is a linear-weighted
average of natural-neighbor coordinates. The basic
equation used in natural-neighbor interpolation can
be defined as follows:
/=1
where
V, =
k =
Sj =
(A4.4.1.2)
interpolated value at grid point i,
number of samples inside the natural-
neighborhood of V,
value of sample j,
weight associated with S,
The c-code nngridr, a complete commercial package
of the natural-neighbor algorithm, from David Watson
(Watson, 1994), with some modification by in-house
developers, was used for developing applications of
natural-neighbor interpolation at the Large Lakes
Research Station (LLRS).
A4.4.1.3 Application
Interpolations, distance square inverse or natural-
neighbor, were needed to build two-dimensional
estimates of a 5 x 5 km grid of Lake Michigan from a
limited number of samples of various parameters.
Other applications could then be applied based on
the interpolated grid data.
A4.4.1.3.1 Contouring Plots
Interpolated grid data is a list of numbers
representing the estimates of physical parameters on
the grid for each grid point. Contour plots connect
points in the grid having the same value with lines.
An incredible amount of information about the data
can be revealed by contour plots. These include
plateaus and canyons, trends, the existence and
location of high and low concentrations, etc.
Contour plots are very effective visualization tools for
analyzing data. Contours in this study were created
by using IDL and other tools. There were a
surprising variety of approaches used to generate
contours. The various techniques that were applied
possess their own advantages and disadvantages.
IDL's standard CONTOUR procedure uses grid
contouring which is the most widely used contouring
technique (Research Systems, Inc., 1995, see
Section 15-1). CONTOUR generates plots from data
stored in a rectangular array (grid data) which usually
is generated by interpolation and extrapolation.
Some other information such as the boundary of
Lake Michigan, sample locations, and the maximum
and minimum values for samples were also
produced.
A4.4.1.3.2 Volume-Weighted Averaging With
Formulations
One way to evaluate and validate the performance of
mathematical models is to compare the model output
and the measured data at the same time (cruises)
and same location (segments). Volume-weighted
average (VWA) is a method to compute the
estimated field data associated with a segment and
a cruise. Depending on the model and segmentation
scheme used, a segment consists of cells of 5 x 5 km
at certain depth range called a layer. The locations of
cells associated with segments are normally
288
-------
provided by segmentation files. The volume
concentration for one cell can be computed by
multiplying interpolated concentration of this cell by
its volume. The volume concentration for a segment
is the sum of volume concentration of all cells in this
segment. And finally, VWA can be computed by
dividing the volume concentration of the segment by
the total segment volume. The equation for
computing VWA is :
(A4.4.1.3)
where
V, =
C, =
n =
volume-weighted average of segment n,
volume of cell i,
concentration associated with cell i
total number of cells
Besides the VWAs, statistical information (mean,
variance, standard error) is also generated for the
users. VWAs were generated by IDL programs
developed in-house. The interpolated grid field data
were generated by either distance square inverse or
natural-neighbor from samples collected for the
LMMBP project.
A4.4.1.4 Discussion
It has been observed and documented that
extrapolations generated by using the natural-
neighbor c-code nngridr could cause problems.
Extrapolation sometimes is necessary to estimate
values for grid points located outside the convex hull,
which is a polygon bounded by the outermost
sample data points. At the beginning of the
interpolation process of running nngridr, a very large
triangle is established which encloses all data being
used for interpolation. Then, a pseudo datum is
assigned to each of the three vertices of the triangle.
Extrapolation, if needed, is performed based on the
pseudo data. This process is doomed to be
unreliable due to the unpredictable nature of the
pseudo numbers and the large triangle used in this
process.
This problem can be remedied by adding some extra
reasonable pseudo samples at the corners outside
the gridding data area so all interpolated grid points
will be inside the expanded convex hull. By doing
this, nngridr is forced to use interpolation, rather than
extrapolation, to calculate estimations based on the
original and pseudo samples. This is a more reliable
estimation process. The choice of pseudo samples,
if necessary, should be based on experience and
nearby samples.
Another limitation of nngridr is that it can only handle
two-dimensional interpolation. There are occasions
when three-dimensional interpolation is needed. One
example is the sediment PCB concentration
estimations to be used for fish uptake. This is much
better represented if the depth of samples could be
utilized to define the neighborhood. The
neighborhood becomes a three-dimensional ball
instead of a two-dimensional circle. Because nngridr
is a relatively large program, there was no easy way
to add a three-dimensional interpolation.
A4.4.1.5 Steps to Run nngridr
The c-code from David Watson, nngridr, was used to
generate the natural-neighbor interpolation. Often,
nngridr was called within an IDL program to generate
the interpolation. Sample data were reformatted to
the required IDL format. The interpolation on the 5
km grid was then used for data analysis and
visualization (post-process) applications.
Details about how to initialize and run nngridr
together with IDL application programs at LLRS
follow.
1. Change the c-code
In nngridr.c, comment out the statement
'InstringO' right after the statement 'printf
("Change parameters or Make the grid? C or M)",
to prevent the read option from a terminal.
Therefore, the c-code nngridrto generate the grid
is run by using the default option 'M'.
289
-------
2 Initialization
A. Change the make file and then use the
command: %make -f makefile to generate
the executable code.
B. The nngridr is run first to generate the initial
file and setup parameters (file names, grid
configurations, etc.). The result will be in
nngridr.ini, which can be used for successive
runs without changing parameters again.
The most important aspect of initialization is
to generate a two-dimensional grid. The
southwest corner with longitude-87.9721 and
latitude 41.5845 is used as the origin of the
grid coordinates. The northeast corner is at
longitude -84.7206 and latitude 46.1069
which is the grid coordinate of (53, 102).
There is an option to output the grid south-
north or north-south. The orientation of
output should be set from south-to-north.
Otherwise, the image will be upside down.
See Watson (1994) for more details.
3. Raw data files are pre-processed to prepare the
input data for IDL code. The formats of data files
should be the same.
4. The configuration of segmentation should be
stored in a file for the segmentation classification.
5. IDL programs are coded to generate data files
similar to jdavis.dat by reading the pre-processed
data. Coordinates are converted from longitude
and latitude coordinates to 5 km grid coordinates.
6. Once the jdavis.dat is established, nngridr is run
by the following commands within IDL programs:
'SPAWN, 'nngridr', Results, /NOSHELU.
This creates a child process under the Unix
operating system and stores all messages
generated by this code into the character array
Results.
7 After a successful run (need error checking if run
fails), the grid data should be generated and
named as nngridr.grd. This file is called in IDL
programs to generate contour plots, VWA results,
and statistics.
8. For the Unix system, both jdavis.dat and
nngridr.grd will be destroyed automatically when
new ones are created. For other operating
systems, Microsoft Windows, for example, these
files need to be deleted.
9. Green Bay data need to be processed separately
from open lake data.
References
Research Systems, Incorporated. 1995. IDL User's
Guide: Interactive Data Language, Version 4.
Research Systems, Incorporated, Boulder,
Colorado. 544 pp.
Watson, D. 1994. nngridr An Implementation of
Natural Neighbor Interpolation. Claremont,
Australia. 170 pp.
290
-------
PART 4
LM2-TOXIC
Chapter 5. LM2-Toxic Calibration and
Confirmation
For diagnosis of system behavior and reasonable
prediction of long-term reactions of a water system,
it is very important that a water quality model is well-
calibrated and confirmed under a conceptually well-
constructed solid and toxic chemical dynamics. Two
of the main objectives in developing the Level 2
contaminant transport and fate model (LM2-Toxic)
were to calibrate and confirm the model using the
Lake Michigan Mass Balance Project (LMMBP)-
generated data and to apply the model for long-term
forecasts of the polychlorinated biphenyl (PCB)
concentrations in water and sediment of the lake for
different forcing functions and loading scenarios.
The advantages of calibrating a coarse spatial grid
water quality model like LM2-Toxic are its efficiency
and quickness.
LM2-Toxic is a coupled mass balance of organic
carbon solids and toxic chemical (PCBs) dynamics.
Because of no feedback mechanisms from PCB
behavior to organic carbon behavior in the model, the
calibration of the LM2-Toxic was done using two
separate calibration stages. The first stage was the
calibration of organic carbon dynamics without
considering any behavior of PCBs in the system and
the second stage was the calibration of PCB
dynamics without adjusting any parameters
associated with organic carbon dynamics.
Prior to the organic carbon and PCB dynamic
calibrations, the LM2-Toxic was used as a thermal
balance model to calibrate the vertical dispersion
coefficients between the water column layers. A
detailed description of inputs used in the thermal
balance model can be found in Part 4, Chapter 4,
Section 4.4.1.2.
The data collected for the LMMBP were spatially
averaged using a volume-weighted averaging
algorithm to generate segment-specific cruise mean
concentrations (Appendix 4.4.1). The averaging
algorithm also computed the statistical standard error
that was expressed as error bars related to the
mean. The cruise mean concentrations were
compared with the predicted concentrations from the
LM2-Toxic for the vertical dispersion coefficients
calibration, the organic carbon dynamics calibration,
and the PCB dynamics calibration.
4.5.1 Vertical Dispersion Coefficients
Calibration
Vertical dispersion coefficients at the interfaces
between LM2-Toxic water column layers were very
important to overall water transport in Lake Michigan.
The vertical mixing defined by the vertical dispersion
coefficients moves large amounts of mass vertically
in the lake and is strongly influenced by water
temperature that has a distinct seasonal variation. A
thermal balance model was applied to calibrate the
predefined vertical dispersion coefficients. The
results in Appendix 4.5.1 demonstrate excellent
agreement between observed temperature and the
model simulation for the LMMBP period. The results
from the thermal balance model were also compared
to the results from the Princeton Ocean
Hydrodynamic Model (POM) (Schwab and Beletsky,
1998) for selected segments (Figure 4.5.1). Both
291
-------
30
o 20-
10
segment 1
10-
30
20-
10-
segment 20
Mar-94 Aug-94 Jan-95 Jun-95 Nov-95 Mar-94 Aug-94 Jan-95 Jun-95 Nov-95 Mar-94 Aug-94 Jan-95 Jun-95 Nov-95
30
O 20-
o
O)
10
segment 25
30
20-
10
segment 36
Temperature (degrees C)
LM2-Toxic
— Model Output
• Cruise Mean
error bars = standard error
30
20-
10-
segment 41
Mar-94 Aug-94 Jan-95 Jun-95 Nov-95 Mar-94 Aug-94 Jan-95 Jun-95 Nov-95 Mar-94 Aug-94 Jan-95 Jun-95 Nov-95
Figure 4.5.1 a. Comparison between the temporal profiles for temperature results from the LM2-Toxic.
30
O 20-
"> -in
•a 1U
30
20
10-
segment 6
30
20-
10-
segment 20
UH , , , r1 0-1 1 r-^ , rJ OH 1 1 1 H
Mar-94 Aug-94 Jan-95 Jun-95 Nov-95 Mar-94 Aug-94 Jan-95 Jun-95 Nov-95 Mar-94 Aug-94 Jan-95 Jun-95 Nov-95
30
o
w
-------
models did a good job of simulating temperature in
the lake.
4.5.2 Organic Carbon Dynamics
Calibration
Because the movement of particulate organic carbon
(POC) (POC = PDC + BIG, where PDC = particulate
detrital carbon, BIC = biotic carbon) dominates the
transport and fate of the PCBs in a natural water
system, the quality of the calibration of the organic
carbon sorbent dynamics was very crucial for a
successful subsequent calibration of PCB dynamics.
The calibration strategy for the organic carbon
dynamics was to 1) fix as many independent
parameters used in the model as possible based on
literature, field measurements, and analytical results
from the field samples, and 2) adjust the parameters
considered to be the most uncertain and site-specific
without using values exceeding their range
constrained by literature.
4.5.2.1 Calibration Process/Procedure
The procedures taken in the organic carbon
dynamics calibration for the LM2-Toxic were the
following:
1. Fixing segment-specific sediment burial rates (vb)
and thickness of surficial sediment mixing layer
based on analyzed results from the LMMBP
sediment core measurements (Bobbins et al.,
1999).
2. Fixing the dissolved organic carbon (DOC)
diffusion coefficient between the surficial
sediment layer and the overlying water column at
1.73 x 10-4 m2/d (DePinto et al., 1993).
3. Fixing segment-specific critical wave heights
using Equation 4.3.12.
4. Assigning initial segment-specific settling
velocities for BIC and PDC based on values from
the literature (DePinto et al., 1993; Eadie et al.,
1990; Baker et al., 1991; Eadie, 1997, Eadie et
al., 1984; Thomann and Di Toro, 1983).
5. Estimating segment-specific empirical wave
coefficients (a) using Equation 4.3.11.
6. Computing segment-specific daily resuspension
rates using procedures described in Sections
4.3.4.2.1 (Steady-State Resuspension
Calculation) and 4.3.4.2.2 (Empirical Wave-
Induced Resuspension Calculation).
7. Executing the LM2-Toxic, examining model
outputs (carbon concentrations), and adjusting
segment-specific settling velocities and
resuspension rates accordingly by repeating
steps 4-6 as necessary.
8. Adjusting biochemical organic carbon
transformation rates and yield coefficients listed
in Table 4.4.16.
Most of the organic carbon decay rates
(transformation rates) and yield coefficients were
initially given the same values used in the Green Bay
Mass Balance Project (GBMBP) and were adjusted
during the LM2-Toxic organic carbon dynamics
calibration. The principal parameters adjusted during
the organic carbon dynamics calibration were rates
related to carbon vertical transport such as settling
and resuspension velocities, carbon decay rates, and
yield coefficients. The final values for these rates
had to be consistent with available literature data and
limnological theory.
4.5.2.2 Results and Discussion
The final values for the biochemical transformation
rates and yield coefficients are presented in Table
4.4.16. The carbon decay rates and yield coefficients
shown in this table are quite consistent with the
limnological theory of organic carbon cycling in a
natural water system. It was expected that the final
carbon decay rates (left-hand side of Equations 4.3.3
4.3.5) in pelagic freshwater systems would
decrease in the order of BIC, PDC, and DOC.
Therefore, the values of substrate saturated decay
rates and Michaelis-Menten half-saturation constants
for the organic carbon would have the following
sequences:
'Wo- respectively.
The lake-wide concentrations (including Green Bay)
in surface water layers (epilimnion) for DOC, PDC,
293
-------
and BIG could be as high as 1.8, 0.30, and 0.20
mg/L, respectively. It is favorable and reasonable to
have'the value of Michaelis-Menten half-saturation
constant for each organic carbon state variable close
to double that of its concentration. The final
sediment PDC decay rate used in the LM2-Toxic was
within the range of the value (5.7 x 10'5 d'1 at 20°C)
used in GBMBP (DePinto et ai, 1993) and the value
(0.001 d'1) from other literature (Gardiner et al.,
1984). The values of yield coefficients (Y(BIC^PDC),
Y(PDC-DOO in Table 4.4.16 indicated less loss to
carbon dioxide (CO2) during the conversion from BIG
to PDC than during the conversion from PDC to
DOC.
The model results of organic carbon dynamics
calibration for the LM2-Toxic are presented in
Figures 4.5.2 to 4.5.4 for temporal profiles of all 41
water column segments. The complete set of
calibration plots for the organic carbon including
temporal profiles in the sediments are provided in
Appendix 4.5.2.
Figure 4.5.2 shows slight temporal variation of DOC
in the main lake and outer Green Bay (see Figure
4.3.1 for Level 2 segmentation). This temporal
change of DOC concentration was closely related to
the strong seasonal variation of PDC concentration
and was more evident in the epilimnion of the lake.
Other than the inner Green Bay area, there was
almost no horizontal or vertical spatial variation for
DOC.
The temporal profiles based on the model outputs for
both BIG and PDC (Figures 4.5.3 and 4.5.4) showed
very strong seasonal variation throughout the lake,
especially in the epilimnion segments. Excluding the
inner shallow water of Green Bay segments where
BIG concentration was controlled by its load from the
Fox River and very localized algal growth, there was
not much horizontal spatial variation for BIG in the
lake. The peaks of the BIG temporal profiles for the
epilimnion segments resulted from an algal bloom in
late spring and early summer. Compared with the
epilimnion segments, having a similar temporal
variation, the concentration of BIG decreased
dramatically in the hypolimnion segments.
The temporal profiles of PDC (Figure 4.5.4) indicated
that the main lake and Green Bay were two very
different systems. It appeared that the PDC
concentration in Green Bay was more dominated by
sediment-water interactions (i.e., resuspension
events evidenced by the spikes on the plots) than
other components or processes such as tributary
loads, decomposition of BIC, and its own decay. In
the main lake where water segments were larger and
thicker, there was little evidence of resuspension
events from the PDC temporal profiles, with the
exception of a couple of main lake segments at the
end of 1995. Algal primary production and BIG-to-
PDC decay were obvious control processes for PDC
concentration in the main lake, especially in the
epilimnion. An almost constant concentration of PDC
(0.12 mg/L) in the large bottom water column
segments indicated that, without much contribution
from BIC decay to PDC in these segments, there
were persistent and substantial PDC fluxes entering
these segments by settling from the upper water
column layer, resuspension from the surficial
sediments, and/or focusing process from the
adjacent shallower area (segments) in order to
maintain this concentration of PDC.
Figures 4.5.5a to 4.5.5c show a comparison between
the model predictions and observed data (cruise-
segment mean concentration) of DOC, BIC, and PDC
for the entire calibration (i.e., the LMMBP study
period). In general, there was good agreement
between the data and model predictions for all
organic carbons. There was a slight overprediction
of BIC concentration indicated by the low slope of the
regression line. Though the results from the
regression (Figure 4.5.5b) indicated a poorer quality
of fit to the observed data and overprediction of BIC,
visual examination of temporal profiles of BIC
concentration (Figure 4.4.3) suggested the fit was
good for most of the segments.
4.5.3 PCB Dynamics Calibration
Because of no feedback mechanisms from PCB
dynamics to organic carbon dynamics, only a limited
number of parameters associated with PCB
dynamics needed to be slightly adjusted. Due to the
significant influence of organic carbon movement on
the transport and fate of PCBs in Lake Michigan, the
PCB partitioning coefficient became one of the very
crucial parameters for the LM2-Toxic PCB dynamics
calibration. Mass budget analysis for ZPCBs (see
Chapter 6 for details) based on the outputs of the
LM-2 Toxic for the two-year LMMBP period indicated
294
-------
4
3-
2-
O
segment 1
segment 2
segment 3
segment 4
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
)
~
o
3-
2-
1 -
n.
segment 5
• v r^ * *
*
-
.
segment 6
segment 1 1
i *-*-*- t * *
segment 12
• i • •» » « *
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
~ 3-
~ 2-
o
o
0 1.
0-
Jar
4-,
~ 3-
I 2-
8
0-
segment 1 3
t
segment 14
segment 15
segment 1 6
.
i94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec
segment 20
segment 21
segment 22
segment 23
— * T » t t
Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
DOC (mg/L)
O ->• K) CO J
segment 24
* » »
segment 25
* * » - *-•,•".
•*. Model Output
• Cruise Mean
error bars = standard error
Dissolved Organic Carbon (mg/L)
Lake Michigan layers 1 , 2, 3
upper 30 meter
water segments
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
Figure 4.5.2. Temporal profiles of DOC in the Lake Michigan water column segments for the organic
carbon dynamics calibration of the LM2-Toxic and the LMMBP cruise mean data.
295
-------
4-,
^ 3-
Q"
75i
•§• 2-
o
O
0 1-
segment 29
segment 30
.
segment 31
— *~* * * ' i i
-
segment 32
1 — *"• * f * _+
Jan94 Jul94 Dec94 Jui95 Dec95 Jan94 JuJ94 Dec94 JUJ95 Dec95 Jan94 Jul94 Dec94 JuJ95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec!
^ 3-
^j"
-§- 2-
O
O
° 1-
Q
segment 33
•
segment 34
»
segment 36
A t * *
segment 37
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec£
•~v Model Output
^ » Cruise Mean
_ 3-
e 2
8
Q .
0-
segment 38
error bars = standard error
ocbjmc, ,1 Djss0|Ved organic Carbon (mg/L) s-egmenl 40
Lake Michigan layers 4 & 5
30 meters to bottom
water segments
*T«. * i
^ »
segment 41
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
Figure 4.5.2. Temporal profiles of DOC in the Lake Michigan water column segments for the organic
carbon dynamics calibration of the LM2-Toxic and the LMMBP cruise mean data (Continued).
296
-------
^ 3-
cn
-§- 2-
O
O
0-
segment 7
frv^s***-* *
T ^** «n^_ i.J>*^1" C^^fc
.
segment 8
*
Pv^*4^~>-ij_
i ^ "^. £s*i^**~>*^**
1
I \ I
^^^N-t f — .
^N^^^_/ X/VN
segment 9
I ?\A A
vyVv ^nA
^rX. Jw W
segment 10
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9!
/]
•3-
15)
£ 2-
0
O
u 1-
0-
segment 17
f
^ ' m -lai ___ iii L ^^^ A.
L * "
•
segment 18
vwl \
^~" 'iNi i ^' ^^^ | r^^. A^..^
1 segment 19
V^sv
^v/ *^~ - Nj"
segment 26
^I^SI*!-.^ r^~»*fc *
* ~~^-*urJft^JnJ"*t ^*«f**~
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9i
4
3-
^j
e o
o
0
1 -
0-
segment 27
J^^v^^^^^^
^~^-^~»»^*~vy^^»
segment 28
v
^^^v_ /^^__ /^*^
* »«
segment 35
i...
-% Model Output
» Cruise Mean
error bars = standard error
Dissolved Organic Carbon
(mg/L)
Green Bay
water segments
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
Figure 4.5.2. Temporal profiles of DOC in the Lake Michigan water column segments for the organic
carbon dynamics calibration of the LM2-Toxic and the LMMBP cruise mean data (Continued).
297
-------
0.5-
0.4-
i 0.3-
0 0.2-
CO
0.1-
0-
Jan
0.5-
0.4
5 n-a
01 0.3-
E,
0 0.2-
CQ
0.1-
o-
Jar
0.5-
0.4
=5, 0.3 J
0 0.2-
00
0.1
0
Ja
0.5
0.4
'S. 0.3
E
o 0.2
CO
0.1
0
Ja
0.5
0.4
Bi 0.3-
o 0.2-
CQ
0.1-
o-
segment 1
_y VK±> /*\
segment 2
^
segment 3
^/^^/^r^
•
segment 4
^^^\
94 Jui94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec
^
segment 5
x*'*-—^ /*s*— s.
^/ » *v±y » * v
segment 6
__y » *\^A » *\«
segment 11
\
segment 12
9\
194 Jui94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 JuJ95 Dec95 Jan94 Jui94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 JuJ95 Dec^
segment 13
f
segment 14
*
segment 15
«
^_^^^^
segment 16
»
n94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec£
segment 20
^
segment 21
^^^^^
segment 22
***
segment 23
T
..*
n94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec'
segment 24
f
« « t
segment 25
* * «
-«. Model Output
* Cruise Mean
error bars = standard error
Blotlc Carbon (mg/L)
Lake Michigan layers 1 , 2, 3
upper 30 meter
water segments
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
Figure 4.5.3. Temporal profiles of BIC in the Lake Michigan water column segments for the organic
carbon dynamics calibration of the LM2-Toxic and the LMMBP cruise mean data.
298
-------
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
i1
o
en
e
o
CD
0.4-
0.3
0.2'
0.1-
0
Jar
0.5,
0.4-
0.3
0.2-
0.1-
0-
segment 29
, _^i 4
i94 Jul94 Dec94 Jul95 Dec!
segment 33
^1^--^
segment 30
segment 31
~r
segment 32
segment 34
• *
» ^
segment 36
^» •
segment 37
— *•«-,. * i
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Ju!94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
"01
_§,
O
00
0.4
0.3
0.2
0.1-
0.
segment 38
*- A , .
segment 39
«*^
-x Model Output
» Cruise Mean
error bars = standard error
Biotic Carbon (mg/L)
Lake Michigan layers 4 & 5
30 meters to bottom
water segments
^ i
u
segment 40
_i A
•^
segment 41
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
Figure 4.5.3. Temporal profiles of BIC in the Lake Michigan water column segments for the organic
carbon dynamics calibration of the LM2-Toxic and the LMMBP cruise mean data (Continued).
299
-------
O
m
0.5
0.4-
0.3-
0.2-
0.1
segment 7
segment 8
segment 9
2.0
1.6-
1.2-
0.8-
0.4-
segmentlO
Jan94 Jui94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
0.5,
O
ca
0.4.
0.3.
0.2-
0.1
segment 17
segment 18
segment 19
segment 26
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
_1
_§
O
m
0.4.
0.3-
0.2-
0.1-
0
segment 27
»
y%~v*A«5xj»-vv^wSv
segment 28
segment 35
—, Model Output
» Cruise Mean
error bars = standard error
Biotic Carbon (mg/L)
Green Bay
water segments
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
Figure 4.5.3. Temporal profiles of BIC in the Lake Michigan water column segments for the organic
carbon dynamics calibration of the LM2-Toxic and the LMMBP cruise mean data (Continued).
300
-------
o>
o
Q
Q_
0.5
0.4
0.3
0.2
0.1
segment 1
segment 2
segment 3
segment 4
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
0.5 n
O
Q
CL
0.4.
0.3
0.2
0.1
segment 5
segment 6
»/*
segment 11
segment 12
0
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
0.5,
O
Q
0.
0.4-
0.3-
0.2-
0.1
segment 13
iff-
segment 14
segment 16
V *
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
0.51
0.4-
0.3-
o
Q
CL
segment 20
segment 21
segment 22
segment 23
o
Q
CL
0.1-
0
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
0.5
0.4-
0.3-
0.2-
segment 25
-^ Model Output
» Cruise Mean
error bars = standard error
Participate Detrital Carbon (mg/L)
Lake Michigan layers 1, 2, 3
upper 30 meter
water segments
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
Figure 4.5.4. Temporal profiles of PDC in the Lake Michigan water column segments for the organic
carbon dynamics calibration of the LM2-Toxic and the LMMBP cruise mean data.
301
-------
0.5-
0.4-
|> 0.3-
Q 0-2-
CL
0.1-
0
Jar
0.5 1
0.4-
15> 0.3
0 0.2-
Q.
0.1-
0
Ja
0.5
0.4
IT
"a 0.3
0 0.2
Q.
0.1
0
segment 29
jju J^l
segment 30
* r*
'S^>f"r>O^/'^
segment 31
i
'
segment 32
* -Urv f^^
Vs^t^V^^T
i94 Jul94 Dec94 Ju!95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec,
segment 33
» *
segment 34
*
segment 36
<_^^^
**n ~" *
segment 37
^+— --
194 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Ded
segment 38
~.M
» u
segment 39 error bars
odeIC
xiise f
= star
)utput
/ean
idard error segment 40
Participate Detrital Carbon
(mg/L)
Lake Michigan layers 4 & 5
30 meters to bottom
water segments
--rrr-^r
*»*
segment 41
***
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
Figure 4.5.4. Temporal profiles of PDC In the Lake Michigan water column segments for the organic
carbon dynamics calibration of the LM2-Toxic and the LMMBP cruise mean data (Continued).
302
-------
O)
O
Q
0.
1.0
0.8-
0.6-
0.4
0.2-
CD
O
Q
0.
D)
O
Q
Q.
segment 7
1.0
0.8-
0.6-
0.4-
0.2-
segment 8
1.0
0.8-
0.6-
0.4-
0.2-
segment 9
0
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
1.0
0.8-
0.6.
0.4
0.2-
o.
segment 17
1.0
0.8-
0.6-
0.4
0.2
0
segment 18
1.0
0.8-
0.6-
0.4
0.2-
n
segment 19
1.0
0.8-
0.6-
0.4-
0.2-
n
segment 26
Jan94 Jul94 Dec94 Jul95 Dec95
1.0
0.8-
0.6-
segment 27
0
1.0
0.8-
0.6-
0.4-
0.2-
0
Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
1-0-,
segment 28
0.8-
0.6.
0.4-
0.2-
0
segment 35
«•% Model Output
» Cruise Mean
error bars = standard error
Participate
Detrital Carbon
(mg/L)
Green Bay
water segments
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
Figure 4.5.4. Temporal profiles of PDC in the Lake Michigan water column segments for the organic
carbon dynamics calibration of the LM2-Toxic and the LMMBP cruise mean data (Continued).
303
-------
3.0
-2.64
D)
to
T3
JJ1.8-
o>
I 1.4
1.0
1.1548x-0.1961
R2 = 0.55
DOC
(figure a)
1.0 1.4 1.8 2.2 2.6
Model outputs mg/L
y = 0.9113x +0.0153
R2 = 0.5347
y = 0.5792X + 0.0281
R2 = 0.2297
0.2
Model outputs mg/L
0.2 0.4 0.6
Model outputs mg/L
Figure 4.5.5. Observed data versus the LM2-Toxic predictions for DOC, BIC, and PDC for the LMMBP
period.
Michigan. Enthalpy in Equation 4.3.24 was one of
the volatilization parameters to which the PCB
concentrations in the main lake were very sensitive
to, and it was slightly adjusted during the PCB
dynamics calibration. In addition to adjustments of
these two parameters, the initial conditions in some
of the Green Bay sediment segments were adjusted
for some of the PCB congeners. This was
considered acceptable because PCB concentrations
initially used in the LM2-Toxic for both the water
column and the sediment segments of Green Bay
were very questionable. Previous studies (Bierman
et al., 1992, DePinto et al, 1993) of the Green Bay
system have shown that strong PCB gradients exist
in both the water column and the sediments. Small
number of samples of that system would not,
therefore, be able to capture the details of these
gradients. Also, the system is vulnerable to
resuspension events that can contribute to high
variability in observed PCB concentrations in the
water column. The initial PCB concentrations for the
Green Bay water column segments were derived
from interpolations based on samples from only two
water stations (GB17 and GB24M). The initial PCB
concentrations for the Green Bay sediment segments
were arithmetic averages of samples collected at
only four sediment stations (two Ponar sampling
locations: SD89P and SD106P; two gravity core
sampling locations: 95GI and 113GI) during the
LMMBP period.
The segment-specific cruise mean concentrations for
each PCB congener were generated using the same
inverse distance and volume-weighted averaging
algorithm (Appendix 4.4.1) based on the 1994-1995
LMMBP samples collected during the eight cruises.
These mean concentrations were then used for the
LM2-Toxic PCB dynamics calibration and comparison
with the LM2-Toxic outputs.
4.5.3.1 Calibration Procedures
The specific procedures taken in PCB congener
dynamics in the lake for the LM2-Toxic were the
following:
1. Fixing the PCB diffusion coefficient between the
surficial sediment layer and the overlying water
column at 1.73 x 10/4 nf/d (DePinto et al., 1993).
2. Assigning estimated initial partition coefficients
for each PCB congener computed by using a
two-phase partitioning model based on data from
the LMMBP collected field samples.
3. Assigning volatilization-related parameters such
as enthalpy and entropy of each PCB congener
as derived by Bamford et al. (2002).
4. Slightly adjusting, within the bounds of literature
values and/or acceptable variation in field
observations, sediment PCB initial conditions for
the Green Bay sediment segments, partition
304
-------
coefficients, and enthalpy for the PCB congeners
as necessary to improve the fit between
observed data and model outputs.
4.5.3.2 Results and Discussion
The values for the final set of partition coefficients
(log KVoc,a and Io9 KDOC in Table 4.4.28) for the
LMMBP selected individual and co-eluting PCB
congeners were close to the octanol-water partition
coefficient (Kow) calculated by Hawker and Connell
(1988). As shown in Figure 4.4.15, the partition
coefficients for some of the low and high chlorinated
PCB congeners were adjusted the most during the
LM2-Toxic PCB calibration. Based on the PCB
dynamics calibration procedure outlined in the
previous section along with the parameters listed in
Tables 4.4.28 and 4.4.29, the PCB congener
dynamics in the LM2-Toxic were calibrated for the
two-year LMMBP period. Due to the extremely large
amount of output from the model on a congener
basis, the results and discussion in this section will
be focused on only one PCB congener (PCB28+31)
and the sum of all the LMMBP modeled PCB
congeners (IPCBs) for the purpose of demonstrating
the calibration outcomes. PCB28+31 had the highest
external loads, and its concentration in the Lake
Michigan system was easily double that of the next
closest PCB congener. The mass of IPCBs
accounted for approximately 70-75% of total PCB
mass.
Figures 4.5.6 and 4.5.7 show that temporal
calibration profiles of all 41 water column segments
for PCB28+31 and IPCBs (dissolved phase +
paniculate phase). Appendix 4.5.3 provides a
complete set of calibration plots for PCB28+31 and
IPCBs, including temporal profiles in sediments, total
dissolved phase (unbound and DOC bound), and
particulate phase (sorbed to PDC and BIG).
Although complete sets of calibration plots for each
PCB congener are available, it is impossible to put all
of them in this report.
Based on cruise mean data and model outputs in
Appendix 4.5.3, PCB concentration in the dissolved
phase was approximately double its concentration in
the particulate phase in the main lake. The temporal
profiles (Figures 4.5.6 and 4.5.7) show some degree
of temporal variation controlled by a combination of
seasonal variation of external loads, atmospheric
concentration, and sediment resuspension events.
The separate dissolved and particulate PCB temporal
plots (Appendix 4.5.3) also illustrate that the
seasonal variation of particulate PCBs was much
more prominent than the seasonal variation of
dissolved PCBs.
The temporal profiles (Figures 4.5.6 and 4.5.7)
indicate that there was a slight longitudinal
concentration gradient throughout the main lake.
The highest concentrations were found in the
southern segments due to higher PCB atmospheric
deposition (dry + wet) and much higher PCB
atmospheric concentrations observed in the area
close to Chicago. There was little vertical gradient
found based on main lake cruise mean data. This
indicated that vertical transport such as vertical
advective flows, vertical mixing, and resuspension
strongly influenced the PCB concentrations in Lake
Michigan water column. On the other hand, the
model output predicted higher concentrations in the
bottom layer of the main lake. Possible explanations
of this difference between predicted and observed
include: 1) segment-specific resuspension rates for
the depositional area were not setup properly for the
model, 2) the initial average PCB concentrations in
surficial sediments were too high, or 3) there should
be a bottom layer with a large pool of resuspended
materials (benthic nepheloid layer, BNL) containing
PCBs added into the model segmentation (Baker et
al, 1991; Eadie et al., 1984; Eadie et al., 1990;
Eadie, 1997). The BNL, with much higher carbon
and PCB concentrations, could serve as a buffer
between the hypolimnion layer and the surficial
sediment. The LM2-Toxic PCB dynamics calibration
was conducted on the PCB congener level.
Therefore, the calibration was a tedious and very
time-consuming task. The PCB calibration strategy
was to adjust only those parameters that definitely
needed to be adjusted. The PCB concentrations in
the water column, especially the hypolimnion, were
very sensitive to the PCB initial conditions in the
sediments. Sediment samples collected by box
coring may be biased on the high side due to the
selection of sites that could be cored (fine-grained
sediment). In the deep depositional areas of the
lake, it is unlikely that resuspension occurs. To avoid
controversy and excessive effort on the model
calibration, the segment-specific resuspension rates
were not adjusted independently during the
305
-------
O.OS-i-
g- 0.04-
" 0.03-
CO
8 0.02
CO
CL 0.01
segment 1
-
segment 2
segment 3
I ._
segment 4
I
J°an94 Jui94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 JuJ94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec!
5 0.04-
c
~ 0.03
CO
8 0.02
m
CL 0.01
f\
segment 5
•
segment 6
*
segment 11
NMTff-v^~T]~
segment 12
i I
Jan94 Jui94 Dec94 Jul95 Dec95 Jan94 Jul'94 Dec94 Jul95 Dec95 Jan94 Jui94 Dec94 Jul'95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec£
5 0.04-
cn
c
^ 0.03-
co
8 0.02
a 0.01
0
Ja
0.05
IT 0.04
0)
" 0.03
CO
-f
8 0.02
CO
CL 0.01
0
Ja
0.05
S 0.04
r 0.03
8 0.02
03
" 0.01-
0
segment 13
**
segment 14
* +
segment 15
segment 16
*
» ^^^
n94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 DecE
segment 20
w^^McN— 4*^
segment 21
segment 22
t
segment 23
* * T
n94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 DecE
segment 24
segment 25
~. Model Output
» Cruise Mean
error bars = standard error
PCB 28+31 (ng/L)
Lake Michigan layers 1 , 2, 3
upper 30 meter
water segments
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
Figure 4.5.6. Temporal profiles of PCB28+31 (dissolved phase + particulate phase) in Lake Michigan
water column segments for PCB dynamics calibration of the LM2-Toxic and the LMMBP cruise mean
data.
306
-------
u.uo-
j 0.04-
~ 0.03-
CO
S 0.02
m
CL 0.01
segment 29
, f
-
segment 30
I
segment 31
* «
segment 32
* «
Jan94 Jul94 Dec94 Jul95 Deo95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec
005
j 0.04-
"81
^ 0.03-
CO
a 0.02.
CD
CL 0.01
o
segment 33
segment 34
segment 36
*»» + « » ,
segment 37
,H . if
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Ju!94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec!
005
5" 0.04
D>
^ 0.03 -
CO
8 0.02
m
o- 0.01-
0
segment 38
•wM
» C
segment 39 error bars
odeIC
uise t
= star
)utput
/lean
idard error segment 40
PCB 28+31 (ng/L)
Lake Michigan layers 4 & 5
30 meters to bottom
water segments
^^^T
* »
segment 41
* * *
»
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
Figure 4.5.6. Temporal profiles of PCB28+31 (dissolved phase + particulate phase) in Lake Michigan
water column segments for PCB dynamics calibration of the LM2-Toxic and the LMMBP cruise mean
data (Continued).
307
-------
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
0.10,
0.08 J
segment 18
segment 19
segment 26
CO
CM
CD
o
Q.
0.02-
0
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
0.10,
0.08 J
0.06J
0.04-
segment 35
-s Model Output
» Cruise Mean
error bars = standard error
PCB 28+31 (ng/L)
Green Bay
water segments
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
Figure 4.5.6. Temporal profiles of PCB28+31 (dissolved phase + particulate phase) In Lake Michigan
water column segments for PCB dynamics calibration of the LM2-Toxic and the LMMBP cruise mean
data (Continued).
308
-------
u.o-
0.4-
|> 0.3-
8 0.2-
Q.
W
0.1-
0-
Jar
0.5,
0.4-
|> 0.3-
o 0.2-
Q_
w
0.1-
0-
Jar
0.5n
0.4-
°> 0.3-
0 0.2-
a.
0.1-
0-
Jar
0.5-
0.4-
o> 0.3-
0 0.2-
Q_
W
0.1-
Jar
0.5-
0.4-
g> 0.3-
CD
0 0.2-
w
0.1-
0-
segment 1
. * *
-
segment 2
t »
•
segment 3
^^^-
•
segment 4
^-^S,
94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
segment 5
»
•
segment 6
» * *
segment 1 1
v*^>-
segment 1 2
^^~
i94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
segment 13
Or^s,
.
segment 14
*
segment 15
^-— ^
-
segment 16
O^-^rJ^-
» *
194 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
segment 20
•
*~^>_
segment 21
»»
^J__
segment 22
* *
segment 23
»
^^— ^
194 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
segment 24
* * 1
4
segment 25
•~> Model Output
» Cruise Mean
error bars = standard error
IPCB (ng/L)
Lake Michigan layers 1 , 2, 3
upper 30 meter
water segments
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
Figure 4.5.7. Temporal profiles of ZPCBs (dissolved phase + particulate phase) in Lake Michigan water
column segments for PCB dynamics calibration of the LM2-Toxic and the LMMBP cruise mean data.
309
-------
CO
O
CL
1.0-
0.8-
0.6
0.4
0.2
segment 7
segment 9
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
CO
O
Q.
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
1.0
O
0.
segment 35
-^ Model Output
• Cruise Mean
error bars = standard error
SPCB (ng/L)
Green Bay
water segments
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
Figure 4.5.7. Temporal profiles of ZPCBs (dissolved phase + particulate phase) in Lake Michigan water
column segments for PCB dynamics calibration of the LM2-Toxic and the LMMBP cruise mean data
(Continued).
310
-------
U.3-
0.4-
I 0.3-
ca
0 0.2-
0.
w
0.1-
segment 29
^
. i . *
jULjij* *^wL ii -frtL
^^nf
segment 30
i ^*^f — I,
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94
0 5
0.4.
I1 0.3-
co
0 0.2-
a
w
0.1-
Q
segment 33
»
^^mf" *^N ^^f^ \/L
^^^ ir+ V^^dV _|X^^L *V™
*
*
. ___J__..i
k^~ ^"W
segment 31
»
*
^— ~^** * ^ ^^
segment 32
+ »
^^•i"^» ^-^^sJ*— ^"*-"^ ^V
Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
segment 34
_--*_
»^-<~>
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94
0 5
0.4-
ra 0.3-
8 0.2-
Q.
0.1-
0-
segment 38
^f*T*-^ ^^-^
^^""*^"^ ^ ^^^
>*** » * *
-
segment 39
-^^rx^».
f~**~^v~
segment 36
,
j*W^^%-
r*~*~^ * ^^ r**^~'m~~* ""
* ^^™^^ *
segment 37
rf-"^*7^-*
fn — ^4, V. j^+ ~"^~~**
+ ^"^^^
+
Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
•N, Model Output
t Cruise Moan
error bars = standard error
1PCB (ng/L) ^yincm-ru
Lake Michigan layers 4 & 5
30 meters to bottom
water segments
_-^
^"TT* • **
^^—
_^^^ ^^*b^^^^~^^^>**
*~~ ^ * *
segment 41
^^^-.
^_/*"^ \^ -***^' ^.
» . * * * *
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
Figure4.5.7. Temporal profiles of ZPCBs (dissolved phase + participate phase) in Lake Michigan water
column segments for PCB dynamics calibration of the LM2-Toxic and the LMMBP cruise mean data
(Continued).
311
-------
calibration, and the PCB initial sediment
concentrations in the main lake were kept as the
original results of the interpolation based on surf icial
sediment field data.
Temporal variations of PCBs in Green Bay were
much different from those in the main lake. PCB
concentrations in Green Bay were dominated by the
particulate phase (Appendix 4.5.3), especially for the
water column segments in the inner bay. Dissolved
PCBs were about 1.5 to 2.0 times higher than the
particulate PCBs for the water column segments in
the main lake. From visual examination of the
temporal profiles (Figures 4.5.6 and 4.5.7),
calibration results for Green Bay segments were
better than those for the main lake. This was
unexpected because observational data for the
Green Bay water segments were based on only two
water stations (GB17 and GB24M). The PCB
concentrations in Green Bay were about 5 to 10
times higher than that in the main lake. It appears
that it was much easier for a water quality model to
simulate and get a good fit with observed data for a
chemical constituent with a much higher
concentration than for cases when the concentration
was close to the detection limit.
The distribution of PCBs between dissolved and
particulate phases during the PCB calibration was
very sensitive to the adjustment of the POC partition
coefficients (KPOCa). This distribution was not
sensitive to the DOC partition coefficient due to its
value being two orders of magnitude lower, even
though the DOC concentration was about 5 to 10
times higher than the POC concentration in the main
lake. Figures 4.5.8a and 4.5.8b show a comparison
between the cruise mean concentrations and model
simulation results for PCB28+31 and ZPCBs for the
calibration period. The plots illustrate that there was
PCB 28+31
y = 0.9569x + 0.001
R2 = 0.4847
CD
CO
O
0.40-
-i r
0.01 0.02 0.03 0.04 0.05
Model outputs ng/L
0.00
1.60
I 1.20-
(0
"O
1 0.80-
E
SPCB
sum of 54 PCB congeners
y = 0.9471x +0.0119
R2 = 0.6165
0 0.4 0.8 1.2
Model outputs ng/L
1.6
F«r!od 4"5"8'
data VerSUS the LM2-Tox'c predictions for PCB28+31 and IPCB for the LMMBP
312
-------
a good correspondence between model simulation
results and cruise mean data. Most of the
overpredictions were for segments in the bottom
layer of the main lake.
Table 4.5.1 presents the coefficients (slope,
intercept, and R squares) of the regression equations
based on the results of a comparison between cruise
mean concentration and the LM2-Toxic simulated
concentration for all the LMMBP selected PCS
congeners. In general, the results from the
regression for the PCS congeners were reasonably
good. The results would be much better if the
original initial conditions in the Lake Michigan
sediment segments for the PCB congeners were
slightly adjusted (initial conditions were. adjusted
down to maximum of 20%). The calibration strategy
was to devote only limited effort to fine-tune the
calibration and to adjust as few parameters as
possible during the calibration.
4.5.4 The LM-2 Toxic Confirmation
As for any mass balance model, the results of the
LM2-Toxic calibration and the outcomes of long-term
model prediction have some degree of uncertainty.
As the framework of a water quality model becomes
more complex, the credibility of the model does not
necessarily increase. When the model complexity
increases, the dimensions of the associated
uncertainties will increase, and the ability to describe
model performance will also become increasingly
difficult. Other than the uncertainties associated with
the numerical algorithms and equations used to
describe physical and biochemical processes in the
model, data entered into the model are subject to a
certain error, and this error propagates into the model
results. Due to time constraints, qualitative
evaluation of the model was the focus for the LM2-
Toxic verification. The following modeling effort was
made for the LM2-Toxic confirmation in order to
reduce uncertainties associated with water transport,
settling and resuspension, and sedimentation.
These processes are crucial to long-term projections
using any water quality model.
4.5.4.1 Mass Balance Checking
It is important to check the mass balance on an
individual segment basis for a state variable
simulated in a model. This can ensure that there is
no excessive mass increase or decrease in a
segment due to programming error(s) and error(s)
associated with the numerical method used in the
model. The task was completed very successfully.
Three organic carbon solids (BIG, PDC, and DOC)
and a conservative tracer (any assumed conservative
tracer) were designated as the only state variables.
With initial and boundary concentrations of the
conservative tracer set equal to 1 mg/L in both the
water column and sediment segments and with no
external loads, gas exchange, partitioning processes,
or any kinetic processes, LM2-Toxic was run for
short-term (two years) and long-term (62 years)
simulations. The results from the short-term run
showed almost no change in all media. For the long-
term run, an extremely small change was found in
water segments, and a maximum change of 1 % was
found in a couple of the sediment segments. Table
4.5.2 shows the mass balance checking results for
the long-term model run in each segment.
Considering the small volumes of surficial sediment
segments (average thickness about 1-2 cm), a one
percent change in concentrations over 62 years is an
acceptable variation in terms of conserving mass.
4.5.4.2 Chloride Model
A chloride model was applied as another confirmation
step to ensure that the overall water transport
components, including both advection and
dispersion, used in the LM2-Toxic were reasonably
accurate. With the same model input structure as
the temperature balance model, chloride was
simulated as a state variable without adjusting any
parameter. The resulting temporal profiles in
Appendix 4.3.2 confirmed the credibility of the water
transport used in LM2-Toxic.
4.5.4.3 137Cs and 23»-M0pu Simulation and Results
Based on particle (PDC) net burial in sediments using
Pb-210 core dating results (Robbins era/., 1999), a
set of settling and resuspension velocities were
adjusted simultaneously to maintain the net burial
during the LM2-Toxic carbon dynamics calibration.
The best-fit values were selected for BIG and PDC.
These rates yielded a rate of cycling of particulate
matter between the sediments and overlying water.
Errors in specifying the settling and resuspension
rates can have a significant impact on the model
calibrations and the resulting long-term predictions.
313
-------
Table 4.5.1. Results of the Regression Between the LM2-Toxic Calibration Results and the Cruise
Mean Data for the LMMBP Selected PCB Congeners
Congener
PCB8+5
PCB15+17
PCB1 6+32
PCB18
PCB26
PCB28+31
PCB33
PCB37+42
PCB44
PCB49
PCB52
PCB56+60
PCB66
PCB70+76
PCB74
PCB77+110
PCB81
PCB87
PCB92+84
PCB89
PCB99
PCB101
PCB118
PCB123+149
PCB1 05+1 32+1 53
PCB151
PCB163+138
PCB170+190
PCB172+197
PCB180
PCB187+182
PCB1 95+208
PCB1 96+203
PCB201
PCB85
PCB146
IPCBs3
Slope
0.9512
0.6072
0.9255
0.6271
0.5238
0.9569
0.4931
0.7625
0.8514
1.1693
0.8729
0.5822
0.8250
0.7159
0.6659
0.5593
0.4079
0.5492
0.4801
0.0825
0.3746
0.8253
0.4403
0.5762
0.7421
0.4629
0.5649
0.5879
0.3501
0.4372
0.7757
0.1696
0.2402
0.0911
0.5099
0.3329
0.9471
Intercept
0.0022
0.0034
0.0122
0.0023
0.0009
0.001
0.0054
0.0041
0.0012
0.0006
0.0042
0.0008
- 0.002
5E-05
0.0008
0.0018
0.0013
0.0020
0.0087
0.0012
0.0055
0.0006
0.0020
0.0011
0.0002
9E-05
0.0028
4E-05
0.0001
8E-05
0.0012
9E-05
0.0007
0.0046
0.0006
0.0013
0.0119
R2
0.1845
0.3995
0.0628
0.4112
0.4188
0.4847
0.2811
0.3332
0.5210
0.3165
0.4145
0.5976
0.5896
0.5146
0.5769
0.4219
0.0160
0.4007
0.1896
0.0029
0.1153
0.4186
0.4540
0.4723
0.6205
0.5125
0.4258
0.3735
0.2309
0.3545
0.2425
0.4053
0.1035
0.0004
0.5194
0.2261
0.6165
aSum of all the LMMBP selected PCB congeners.
314
-------
Table 4.5.2. Results of the LM2-Toxic Mass Balance Checking for a 62-Year Simulation of an Assumed
Conservative Tracer (Set the Model Initial Conditions and Boundary Concentrations of the
Conservative Trace = 1)
Water Column
Segment
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
Initial Concentration (mg/L)
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1.00000
1.00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1.00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
1 .00000
0.99959
Concentration at the End of
Simulation (mg/L)
1.00003
1 .00003
1 .00002
1.00002
1.00002
1.00002
1 .00002
1 .00003
1 .00004
1 .00007
1 .00003
1 .00003
1 .00001
1 .00002
1 .00002
1 .00002
1 .00002
1 .00004
1 .00001
1 .00003
1 .00003
1 .00001
1.00001
1 .00002
1 .00002
1 .00002
1 .00002
1 .99999
1 .00003
1.00003
1 .00001
1 .00001
1.00001
1.00002
1 .00001
1 .00002
1 .00002
1 .00000
1 .00000
1 .00001
1 .00001
0.99616
Change (%)
0.003
0.003
0.002
0.002
0.002
0.002
0.002
0.003
0.004
0.007
0.003
0.003
0.001
0.002
0.002
0.002
0.002
0.004
0.001
0.003
0.003
0.001
0.001
0.002
0.002
0.002
0.002
-0.001
0.003
0.003
0.001
0.001
0.001
0.002
0.001
0.002
0.002
0.000
0.000
0.001
0.001
-0.343
315
-------
Table 4.5.2. Results of the LM2-Toxic Mass Balance Checking for a 62-Year Simulation of an Assumed
Conservative Tracer (Set the Model Initial Conditions and Boundary Concentrations of the
Conservative Trace = 1) (Continued)
Water Column
Segment
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
73
74
75
76
77
~7O
78
79
f\f\
80
81
O^"»
82
o/%
83
O A
84
Initial Concentration (mg/L)
0.99955
0.99944
0.99947
0.99939
0.99958
0.99965
0.99980
0.99979
0.99964
0.99953
0.99963
0.99958
0.99971
0.99979
0.99963
0.99955
0.99962
0.99977
0.99974
0.99979
0.99969
0.99960
0.99922
0.99926
0.99938
0.99950
0.99952
0.99977
0.99980
0.99963
0.99967
0.99964
0.99972
0.99970
0.99943
0.99980
0.99981
0.99982
0.99980
0.99982
0.99982
0.99982
Concentration at the End of
Simulation (mg/L)
0.99496
0.99362
0.99387
0.99064
0.99103
0.98954
0.99263
0.99258
0.99187
0.99359
0.99585
0.99525
0.99772
0.99737
0.99603
0.99330
0.99160
0.99252
0.99215
0.99245
0.99430
0.99566
0.99102
0.99158
0.99318
0.99447
0.99442
0.99356
0.99317
0.99561
0.99563
0.99566
0.99711
0.99688
0.99359
0.99163
0.99284
0.99136
0.99035
0.99137
0.99134
0.99112
Change (%)
-0.459
-0.582
-0.560
-0.876
-0.856
-1.011
-0.717
-0.721
-0.778
-0.594
-0.379
-0.433
-0.200
-0.242
-0.361
-0.626
-0.802
-0.725
-0.759
-0.734
-0.539
-0.394
-0.820
-0.768
-0.621
-0.504
-0.511
-0.621
-0.664
-0.402
-0.404
-0.398
-0.261
-0.282
-0.585
-0.817
-0.697
-0.846
-0.945
-0.845
-0.849
-0.870
316
-------
Table 4.5.2. Results of the LM2-Toxic Mass Balance Checking for a 62-Year Simulation of an Assumed
Conservative Tracer (Set the Model Initial Conditions and Boundary Concentrations of the
Conservative Trace = 1) (Continued)
Water Column
Segment
Initial Concentration (mg/L)
Concentration at the End of
Simulation (mg/L) Change (%)
85
86
87
88
89
90
91
92
93
94
0.99979
0.99979
0.99981
0.99983
0.99983
0.99980
0.99982
0.99979
0.99982
0.99982
0.99032
0.99072
0.99184
0.99389
0.99146
0.98917
0.99063
0.99097
0.99202
0.99233
-0.947
-0.907
-0.798
-0.595
-0.837
-1 .063
-0.920
-0.883
-0.780
0.749
To confirm the settling and resuspension rates
selected for the LM2-Toxic, a radionuclide model was
developed. The radionuclides, 137Cs and 239-24°pu,
were used as state variables in the LM2-Toxic, and
a 46-year hindcast (1950-1995) simulation was
executed. Figure 4.5.9 presents cesium and
Plutonium lake-wide water column concentrations
resulting from the radionuclide model 46-year
hindcast simulation. The profiles of both 137Cs and
239,240pu jncjjcate that the water column concentration
of the radionuclides decreased at a rate faster than
the one suggested by the data. There are quite a
few factors that can contribute to this discrepancy.
Among them, loading history of total suspended
solids, internal primary production history, partitioning
coefficients, coarse segmentation, and the settling
and resuspension rates could have a strong influence
in the level of 137Cs and 239'240pu concentration in the
water column of the lake.
To determine the representativeness of the rate of
sediment-water solids cycling used in the LM2-Toxic,
the total sediment 137Cs predicted and observed
inventories were compared. There was a very good
agreement in total sediment 137Cs inventory for those
sediment segments where the sediment core
samples were collected (Figure 4.5.10). There was
a large discrepancy in the comparison for segment
52. The model underestimated total sediment 137Cs
inventory in that segment. This underestimation may
be caused by the location of segment 52. It is a very
narrow transitional area along the southeast
shoreline of the lake with a steep slope. The LM2-
Toxic, with a very coarse spatial resolution for the
water column, was unable to predict the radionuclide
sediment inventory in such a narrow transitional
band. In general, the results from the LM2137Cs and
239.240pu mocje| jmp|y that the combination of the
settling, resuspension, and burial rates used in the
LM2-Toxic were a reasonable representation of the
rate of cycling of particulate matter between the
sediments and overlying water column of the lake.
4.5.4.4 Long-Term Organic Carbon Simulations
For the purpose of providing another dimension of
LM2-Toxic confirmation, a 40-year long-term organic
carbon simulation was executed to produce the total
sediment accumulation over the simulation period.
Using the model, total sediment thickness (cm)
accumulated over 40 years was calculated. It was
converted to sediment accumulation rate (cm/year)
by dividing by 40 years and comparing to the
measured sediment accumulation rate. The three
organic carbons (DOC, BIG, PDC) were the only
state variables in the LM2-Toxic for the simulation.
The set of sediment accumulation rates at each box
core sampling location (Robbins et a/., 1999) were
interpolated using a natural-neighbor interpolation
algorithm (Appendix 4.4.1) to generate segment-
317
-------
0.0100
o
Q.
c
o
c
(D
U
c
O
O
0.0075
0.0050 •
.3 0.0025
o
J3
Q.
A. water column average plutonium
n 1 1 1 1
1950 1955 1960 1965 1970 1975 1980 1985 1990 1995
O
CL
c
o
O
O
in
CD
O
0.75
0.5 .
0.25 .
B. water column average cesium
1950 1955 1960 1965 1970 1975 1980 1985 1990 1995
Figure 4.5.9. Lake-wide average concentrations of (a) 137Cs and (b) 239240Ru computed by the LM2
radionuclide model over 46 years (1950-1995).
50 51 52 60 61
Sediment Segment
62
63
70
Figure 4.5.10. Sediment 137Cs inventory comparison between the observed data (see Appendix 4.8 for
details) and the LM2 radionuclide model outputs.
318
-------
specific accumulation rates (cm/year). Table 4.5.3
presents comparisons between 1) the segment -
specific sediment accumulation rates generated
based on the data (Bobbins et al., 1999) and the
segment-specific accumulation rates computed by
the LM2-Toxic and 2) lake-wide average sediment
accumulation rates generated from the data (Bobbins
et al., 1999) and computed by the LM2-Toxic. The
discrepancy between the field data and model output
on a lake-wide basis was about 30%. A large portion
of this difference was contributed by sediment
segments under water columns 5 and 6 (segments
65-77) where the bottom geometry is complicated
and by Green Bay sediment segments. Because
there was only one sediment box core sample
collected in the sediment segments under water
columns 5 and 6 and no box core sample was taken
in Green Bay, the interpolation of the LMMBP-
generated accumulation rates could create
substantial uncertainty in the interpolated segment-
specific accumulation rates for the sediment
segments under water columns 5 and 6 and in Green
Bay. In general, the results in Table 4.5.3 showed
reasonably good match between the field data and
model predicted accumulation rates, especially in
areas where a large number of box cores were
collected.
4.5.4.5 PCBHindcast
A hindcast used as either a calibration or a
confirmation of a mathematical model is considered
an important and preferred approach to assess the
credibility of a model. It has been used for models
such as MICHTOX (Endicott, 2002), LOTOX
(DePinto era/., 2003), HUDTOX (U.S. Environmental
Protection Agency, 2000), and the Delaware River
model (Delaware River Basin Commission, 2003).
To gain confidence in the prediction of a toxic
chemical model, a PCB hindcast is conducted to
confirm the suitability of the processes
conceptualized in the model and of the associated
parameters used in the model.
A PCB hindcast was not one of the confirmation
components proposed in the original LMMBP
modeling work plan (U.S. Environmental Protection
Agency, 1997). This modeling task was done in
response to a suggestion from the 2004 peer review
panel (summarized by Rygwelski in Part 7, Appendix
1). Because LM2-Toxic went through a very
thorough and successful short-term (two-year project
period) calibration and a series of model confirmation
efforts detailed in the previous sections, the main
purpose of the LM2-Toxic PCB hindcast simulation
was to confirm the representativeness of the
calibration parameters determined from the short-
term calibration and to check the consistency of the
estimated historical PCB load with the available
sediment core profiles and historical inventory data.
A description of data and procedure used for the
LM2-Toxic PCB hindcast and a discussion of the
results and findings from the PCB hindcast are
presented in the following sections.
4.5.4.5.1 Data and Procedure for the PCB
Hindcast
In order to run the LM2-Toxic PCB hindcast, the
following data were required: 1) PCB loading history;
2) historical atmospheric vapor phase PCB
concentrations; 3) historical primary productivity; 4)
the date when PCBs started being loaded into Lake
Michigan; 5) physical conditions such as
temperature, wind speed, and carbon and PCB
transport and kinetic parameters; and 6) the
estimated long-term carbon solid concentrations on
which the resuspension rates should be adjusted to
keep solid cycling rates balanced (see Section
4.3.4.2.1 - Steady-State Resuspension Calibration
for details).
The minimum data needed to reasonably assess the
performance of the LM2-Toxic PCB hindcast were 1)
continuous long-term historical annual averaged
water column observed PCB concentrations; 2)
sediment PCB core profiles representative of the
depositional area, and 3) the sediment total PCB
inventory.
Among the data necessary for a successful PCB
hindcast and to assess model performance, PCB
loading history, historical PCB vapor phase
concentrations, and a reliable estimation of sediment
inventory are the most critical. PCB loading history
and historical PCB vapor phase concentrations
control in-flux and out-flux of PCB mass for the Lake
Michigan system. Historical concentrations in the
system are controlled by these fluxes. When
compared to a reliable estimate of sediment PCB
inventory, the model generated lake-wide sediment
inventory can provide the confidence that there is no
319
-------
Table 4.5.3. Comparison Between the LMMBP Field-Generated and the LM2-Toxic-Generated Sediment
Accumulation Rates (cm/year)
Segment
42
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
73
74
75
76
77
78
79
80
81
82
83
84
85
86
Field Data3 (cm/year)
0.0000
0.0000
0.0002
0.0058
0.0181
0.0495
0.1135
0.3630
0.2662
0.1444
0.2324
0.0497
0.0466
0.0068
0.0000
0.0036
0.0276
0.1059
0.2213
0.2726
0.2768
0.1847
0.1372
0.0013
0.0032
0.0004
0.0176
0.0543
0.1165
0.1938
0.0000
0.0153
0.0033
0.0054
0.0123
0.0180
0.0496
0.0496
0.0496
0.0496
0.0496
0.0496
0.0496
0.0496
0.0496
Model Output (cm/year)
0.0000
0.0000
0.0000
0.0000
0.0650
0.0900
0.1250
0.3875
0.2900
0.1900
0.1800
0.0000
0.0250
0.0000
0.0000
0.0000
0.1300
0.1900
0.2800
0.3250
0.3100
0.3200
0.0250
0.0000
0.0000
0.0000
0.0000
0.2750
0.2400
0.3000
0.3000
0.3575
0.3250
0.0000
0.0000
0.0000
0.0000
0.0000
0.1000
0.1000
0.1000
0.1000
0.0000
0.0000
0.0000
320
-------
Table 4.5.3. Comparison Between the LMMBP Field-Generated and the LM2-Toxic-Generated
Sediment Accumulation Rates (cm/year) (Continued)
Segment
Field Data3 (cm/year)
Model Output (cm/year)
87
88
89
90
91
92
93
94
Lake-Wide Average
0.0496
0.0993
0.0993
0.0993
0.0993
0.0947
0.0947
0.0947
0.1337
0.0000
0.0000
0.0000
0.6000
0.4000
0.3000
0.4000
0.4000
0.1742
"Sediment segment-specific accumulation rates computed based on sediment accumulation rates at each
sediment box core sampling location estimated by John Robbins (Bobbins etal., 1999).
significant overestimation or underestimation by the
LM2-Toxic for the total PCB mass in the lake
sediments. The major challenge for the PCB
hindcast was that these data are very limited for Lake
Michigan. The following subsections provide the
references for data sources and a brief description of
each major data set used in the LM2-Toxic PCB
hindcast.
4.5.4.5.1.1 PCB Loading Reconstruction - As is the
case for most large water bodies, there is no long-
term historical PCB load record available for the lake.
Endicott et al. (2005) and Endicott (2005) are the
only researchers who attempted to reconstruct the
long-term PCB loading history of Lake Michigan
using an approach similar to that applied to Lake
Ontario by Mackay (1988) and Gobas et al. (1995).
Endicott (2005) conducted three 55-five year total
PCB hindcast simulations using MICHTOX and
concluded that the results from Scenario B had the
best-fit with the available historical PCB data. The
year when PCB contamination began was assumed
to be 1940. The PCB loading function of Scenario B
peaked in 1961 and declined after 1963. The rates
of increase in the loads including tributary and
atmospheric loads were the same as the rates of
decrease. Based on observations of the past 25
years, the rates of increase and decrease were
defined as 0.115/year and 0.0535/year for
atmospheric loads and tributary loads, respectively.
In contrast to the procedure of reconstructing the
historical PCB total loading function used in the
MICHTOX (Endicott, 2005), the historical PCB
loading estimation used in the LM2-Toxic PCB
hindcast simulations was done by relating it to the
results obtained from the analysis of a highly
resolved sediment core (LM94-15A) collected during
the LMMBP period (Details are in Part 1, Chapter 7
of this report). In doing so, some assumptions and
professional judgments were made. One important
assumption was that the total PCB inventory and the
shape of the total PCB profile obtained from the
sediment core represented the PCB loading history
from all sources. Consequently, the following key
elements that were used in reconstructing a historical
PCB loading function were derived based on the core
LM94-15A profile: 1) the starting date of the PCB
loading time function (January 01,1949); 2) the year
in which the PCB loading function reached its peak
which was determined to be 1967; and 3) the slopes
on both the inclining and declining sides of the core
profile curve were linear. The historical PCB loading
time function was then back-projected based on the
above elements and variations of the LMMBP
estimated loads for 1994 reported by the United
321
-------
States Geological Survey (USGS) (Hall and
Robertson, 1998) for tributary loads and the LMMBP
atmospheric working group (McCarty et al, 2004;
Miller et al., 2001) for atmospheric loads. Details on
the determination of the key elements and the
procedure in reconstructing the PCB load history are
further discussed in Part 1, Chapter 7. Figure 4.5.11
shows the final reconstructed total PCB historical
loading function (January 01, 1949 - December 31,
1995) and the PCB profile from the sediment core
(LM94-15A). The monthly variation pattern in the
reconstructed long-term historical PCB load functions
used in the LM2-Toxic model for each PCB congener
followed the same monthly pattern established by the
LMMBP estimated loads.
4.5.4.5.1.2 PCB Atmospheric Vapor Phase
Concentration Reconstruction - Recent studies
(Thomann and Di Toro, 1983; Rodgers et al., 1988;
Jeremiason et al., 1994; DePinto et al., 2003; Part 3
of this report; Part 4.6.2 of this report) have shown
that the air-water exchange between dissolved PCBs
in surface water and overlaying vapor phase PCBs is
a very important and possibly the most significant
loss/gain process, especially in recent years, for
PCBs in the Great Lakes region. The gross
volatilization and absorption fluxes transport large
amounts of PCB mass in and out of the lakes due to
the gradient of PCB concentrations between the air
and surface water. These individual fluxes were
usually greater than the sum of the external loads
(tributary, atmospheric wet and dry deposition).
Therefore, a reasonable representation of a
reconstructed historical PCB atmospheric vapor
phase concentration is critical to the success of the
LM2-Toxic PCB hindcast for Lake Michigan.
Following a similar approach used in the
reconstruction of the loading time function, the
starting date of the PCB vapor phase time function
was January 01, 1949, the year for which the PCB
vapor concentration reached its peak was 1967, and
the slope on the inclining side of the time function is
the same as the one indicated from the core LM94-
15A profile. The only difference is that the vapor
concentration on the declining side of the time
function was back-projected based on the LMMBP
(1994-1995) generated PCB vapor concentration and
the decline rate that was carefully selected via a
thorough review of research studies conducted on
PCB vapor phase concentration and the data
collected within the Lake Michigan watershed for the
25 years prior to 1996.
The estimation of the declining rate in the vapor
phase PCB concentrations is very subjective to the
data set and number of data points used in the
derivation of the rate. The half-life of the declining
rate published in the literature ranges from 6 to 20
years for PCB vapor phase concentrations over Lake
Michigan (Hillery et al., 1997; Schneider et al., 2001;
and Buehler et al., 2002, 2004). The individual data
sets used by researchers to determine the PCB
vapor concentrations declining rate were usually a
subset of the historical PCB vapor concentrations
measured between 1977 and 2001. Hillery et al.
(1997) and Schneider et al. (2001) calculated a
declining rate of six-year half-life based on the
measurements collected before 1997, while a rate of
half-life as high as 20 years was estimated by
Buehler et al. (2002, 2004) using only the data from
the Integrated Atmospheric Deposition Network
(IADN), 1992-2001. In this study, a declining rate of
0.115/year, which corresponds to a six-year half-life,
was chosen because the simulation period of the
LM2-Toxic PCB hindcast was from 1949 through
1995 and overlapped the post-peak declining period
used by Hillery et al. (1997) and Schneider et al.
(2000). Similar to the PCB load reconstruction, the
reconstructed PCB vapor phase concentration time
function maintains the same monthly pattern
observed during the LMMBP period. Figure 4.5.12
shows the reconstructed historical time function of
total PCB vapor phase concentrations (January 01,
1949 - December 31, 1995) along with the
reconstructed total loading function for Lake
Michigan.
4.5.4.5.1.3 Estimation of Historical Primary
Productivity - PCBs are hydrophobic organic
chemicals and are closely associated with organic
carbon in natural water systems. The movement and
fate of organic carbon mass are very important to the
distribution of PCBs in the Lake Michigan system.
Therefore, a historical organic carbon loading
function was necessary for a LM2-Toxic PCB
hindcast. The carbon solid loads to the lake are
primarily from tributaries (external loads) and primary
production of phytoplankton (internal load). Both the
measurements and outputs from the LM3-Eutro for
the two-year project period (1994-1995) indicated
that the internally-generated organic carbon load
322
-------
tr9000-
CO
0»
"01
•Q
re
_o
CD
O
Q.
7500-
6000-
4500-
3000-
1500-
0
— Projected historical annual PCS load
o Sediment core LMMB94 15A
i^ o o o
280
•240
-200
-160
120
80
40
0
1940 1950 1960 1970 1980 1990 2000
Years
-5*
"5)
c
c
.0
13
"c
o
o
c
o
o
00
o
Q.
in
o
O
Figure 4.5.11. Reconstructed historical total PCB loading time functions and sediment core LM94-15A
total PCB concentration profiles for Lake Michigan. Note: The total PCB loading time function was
back-projected based on the estimated PCB load for 1994. The last data point of the load time function
represents the estimated PCB load for 1995.
ro
m
o
a.
3
o
— Reconstructed PCB load
o Reconstructed PCB vapor concentration
.5
9.0-
7.5-
6.0-
4.5-
3.0-
1.5-
9000
•8000
7000
heooo
5000
4000
(-3000
2000
MOOO
^
^
•a
_i
O
0
1940 1950 1960 1970 1980 1990 2000
Years
Figure 4.5.12. Reconstructed total PCB vapor phase concentrations and total PCB loading time
functions for Lake Michigan. Note: The total PCB loading time function was back-projected based on
the estimated PCB load for 1994. The last data point of the load time function represents the estimated
PCB load for 1995.
323
-------
(primary production) accounted for approximately
90% of the total carbon solid loads for Lake Michigan
(details are in Part 4, Section 4.4.2). Consequently,
it would be an adequate representation of the
historical carbon solid loading function if the primary
production history for Lake Michigan could be
reasonably reconstructed. However, there are too
few primary production data for Lake Michigan in the
literature to permit creation of an organic carbon
loading history. Due to time constraints, a simple
relationship was established by correlating primary
productivity with available historical total phosphorus
loads (Appendix 4.5.5).
By establishing a relationship between total
phosphorus loads and primary productivity, an
estimation of historical primary productivity was
constructed for Lake Michigan using an approach
similar to the one by Vollenweider et al. (1974).
Figure 4.5.5.1 in Appendix 4.5.5 shows the
relationship between total phosphorus loads and lake
area normalized primary productivity. Figure 4.5.13
presents the reconstructed total carbon solids loading
time function used for the LM2-Toxic PCB hindcast
simulations.
The reconstructed carbon solids loading history
(Figure 4.5.13) indicated that the annual average
internal carbon loads for Lake Michigan reached a
peak value of approximately 3.5 x 109 kg/year in
1980 and then decreased to the current level (1994-
1995) of 2.1 x 109 kg/year used in the LM2-Toxic
model. The annual average carbon concentrations
were expected to follow the same trend as the
historical carbon solids loading function.
4.5.4.5.1.4 Other Physical and Kinetic Parameters -
In terms of physical processes, one of the main
assumptions made for the LM2-Toxic PCB hindcast
was that the time functions for water transport, wind
speed, wave heights, and temperature constructed
for the two-year LMMBP period were representative
of annual average conditions (see Part 1, Chapter 4
for details) that existed in the lake for the entire
hindcast simulation period. Values of the kinetic
parameters used in the model calibration were kept
the same during the hindcast simulation.
Resuspension rate is a function of organic carbon
concentration in the water column, assuming the
other parameters in Equation 4.3.8 are constant (see
Part 4, Chapter 3 for details). By using the carbon
concentration data generated during the LMMBP and
the reconstructed carbon solids loading history, the
annual average organic carbon concentrations in the
water column were estimated (Appendix 4.5.5). The
resuspension rates for each year throughout the
hindcast simulation period were adjusted with the
estimated annual organic carbon concentrations (see
Section 4.3.4.2.1 - Steady-State Resuspension
Calibration for details). The adjusted resuspension
time function maintained the same monthly pattern
that was derived from the observations made during
the LMMBP period.
4.5.4.5.1.5 Historical Water Column PCB
Concentrations and Sediment Core Profiles -
Historical PCB concentrations in the water column
and sediments are essential data to evaluate the
results of a water quality model hindcast simulation.
The available Lake Michigan water column data are
listed in Table 4.5.4 together with their sources.
Measurements made before 1994 LMMBP were
taken at a very limited number of locations, within
short-time periods and at depths less than 10 m in
the water column. Thus they are biased and may not
truly represent annual average water column PCB
concentrations.
Table 4.5.5 provides the sediment PCB core profiles
with depth and temporal information at three LMMBP
sediment sampling locations (LM94-015A, LM95-
061 A, and LM95-086A). Two of the sediment cores
(LM95-061A and LM95-086A) were taken within
depositional zones as defined in Figure 4.3.2. Core
LM94-015A was collected in the transitional zone.
Figure 4.5.14 shows the locations of the three
sediment cores taken during the LMMBP period. For
more details about these three sediment cores, see
section Part 1, Chapter 7.
4.5.4.5.1.6 PCB Inventory in Lake Michigan
Sediments - An estimate of PCB inventory in the
lake sediments is the final information needed to
ultimately confirm the outcomes of the LM2-Toxic
PCB hindcast and the applicability of the model for
future predictions. Table 4.5.6 lists the available
PCB sediment inventory data for Lake Michigan. The
most recent estimate from Eadie and Van Hoof (see
Part 1, Chapter 7 for details) showed that the PCB
inventory for the lake is about 60,998 kg.
324
-------
1940 1950 1960 1970
Years
1980
1990
Figure4.5.13. Reconstructed total organic carbon load (primary production + LMMBP tributary loads)
for Lake Michigan.
Table 4.5.4. Available Historical Water Column Total PCB Concentrations for Lake Michigan
Year
1976
1979
1980
1980
1980
1981
1986
1991
1992
1993
1994-1995
Concentration
(ng/L)
7.2
2.88
5.66
6.36
1.2
0.28
1.1
0.64
0.424
0.22
0.259
Standard Error/
Deviation (ng/L)
3.1
3.37
1.12
1.3
0.5
0.2
0.43
0.058
0.04
0.172
Depth
(m)
1
1
1
5
1
8
5
Reference
Chambers and Eadie, 1980
Rice eta/., 1982
Rice etal., 1982
Rice etal., 1982
Swackhamer and Armstrong, 1 987
Filkins etal., 1983
Lefkovitz, 1 987
Pearson, 1996
Bicksler, 1996
Anderson, 1999
LMMBP
325
-------
Table 4.5.5. Sediment PCB Concentration Vertical Profiles Analyzed for Three Sediment Box Cores
Taken During the LMMBP
LM94-015A
Interval
cm
0-1
1-2
2-3
3-4
4-5
5-6
6-7
7-8
8-9
9-10
10-11
11-12
12-13
13-14
14-15
15-16
16-17
17-18
18-19
19-20
20-21
21-22
22-23
23-24
24-25
25-26
26-27
27-28
28-30
Total PCBs Mid-Interval
ng/g Dry Date
73.7
60.7
67.6
83.7
91.1
102
119
140
148
199
217
230
208
190
222
221
204
142
88.8
71
31.4
23.5
12.5
10.5
9.52
3.58
2.19
1.37
1.62
1994.16
1992.78
1991.16
1989.45
1987.37
1985.17
1983.07
1980.81
1978.61
1976.55
1974.43
1972.21
1969.91
1 967.54
1965.35
1963.31
1961.05
1958.70
1 956.47
1 954.23
1951.82
1949.47
1 947.27
1945.07
1942.78
1940.30
1937.73
1934.83
1930.81
LM95-061A
Interval
cm
0-0.5
0.5-1
1-1.5
1.5-2
2-2.5
2.5-3
3-3.5
3.5-4
4-5
5-6
6-7
7-8
8-9
9-10
10-11
11-12
12-13
13-14
14-15
15-16
16-17
17-18
18-19
20-22
22-24
24-26
26-28
28-30
Total PCBs
ng/g Dry
91.1
111
110
127
117
120
122
136
100
106
103
117
119
117
144
164
159
128
86.9
65
50.4
29.2
17.6
7.02
4.51
4.22
2.62
1.51
Mid-Interval
ng/g Dry
1995.37
1994.78
1994.28
1993.89
1993.37
1992.63
1991.90
1991.12
1989.61
1987.43
1985.08
1982.53
1979.84
1977.08
1 974.34
1971.69
1968.97
1966.30
1963.55
1960.58
1 957.68
1954.65
1951.57
LM95-086A
Interval
cm
0-0.5
0.5-1
1-1.5
1.5-2
2-2.5
2.5-3
3-3.5
3.5-4
4-5
5-6
6-7
7-8
8-9
9-10
10-11
11-12
12-13
13-14
14-15
15-16
Total PCBs
ng/g Dry
78.3
64.6
66.6
75.7
77.2
77
80.1
76.6
77.7
102
101
97.2
82
34.6
14.5
7.84
3.54
2.61
1.51
0.893
Mid-Interval
Date
1994.90
1993.30
1991.40
1989.30
1987.10
1984.60
1982.00
1979.00
1974.20
1968.10
1961.90
1955.70
1 949.50
1 942.00
1933.80
1925.00
1916.50
1908.60
1900.50
1891.60
326
-------
Figure 4.5.14. The sampling sites of the sediment box core samples (LM94-15A, LM95-61 A, LM95-87A)
taken during the LMMBP for which vertical PCB concentration profiles were analyzed and available.
Table 4.5.6. Available Inventories of PCBs in Lake Michigan Sediments
System
Lake Michigan
Green Bay
Lake Michigan + Green Bay
Inventory (kg)
75,000
14,565
60,000
Source
Golden et al., 1993
Wisconsin Department of Natural Resources, 2003
Eadie and Van Hoof (personal communication, 2006)
327
-------
4.5.4.5.2 Results From the LM2-Toxic PCB
Hindcast
As mentioned previously, in order to ensure the
predictability of a water quality model, a long-term
PCB hindcast has conventionally been used for either
confirming a set of key model parameters defined by
the combination of literature sources and short-term
model calibration or directly as a long-term calibration
tool if a short-term model calibration has not been
conducted. LM2-Toxic went through a very
comprehensive and data intensive short-term
calibration using LMMBP-generated field data. The
PCB hindcast for this report is to confirm the model
parameters defined by the short-term calibration and
to provide confidence in modeling long-term
predictions.
Results from the LM2-Toxic PCB hindcast presented
and discussed in this section include 1} water column
PCB concentrations from the hindcast versus
available historical data, 2) sediment PCB
concentrations from the hindcast versus sediment
core profiles, 3) mass budget checking for the
hindcast simulation period, and 4) a comparison
between model-generated PCB sediment inventory
and available sediment inventory estimations made
by other researchers. In general, the results
demonstrate the LM2-Toxic PCB hindcast is
reasonably successful for reproducing PCB
concentrations in both the water column and the
sediments of Lake Michigan.
The results of the LM2-Toxic PCB hindcast are
presented in Figures4.5.15through 4.5.17forannual
and monthly average concentrations in the water
column and annual average concentrations in the
sediments. In general, the results from the hindcast
are reasonably good for both the water column and
the sediment relative to the available historical water
column field data and the three LMMBP sediment
cores. The results demonstrate that the model is
able to generate the temporal trends of PCB
concentrations which closely match the observed
trends.
Some of the historical data derived before and during
1980 (Chambers and Eadie, 1980; Rice etal., 1982)
are not shown in Figures 4.5.15 and 4.5.16. Due to
the limitation of the analytical method, the objective
of the research studies, and the water depth (< 1
meter below the water surface), the
representativeness of these early measurements for
the lake as a whole is questionable. Taking account
of the possible variation in PCB concentrations with
water depth, the open lake value of 1.2 ng/L ± 0.46
(for surface water only - five meters) for 1980 from
Swackhamer and Armstrong (1987) was considered
to be more appropriate comparison with the lake-
wide annual or monthly average concentration
generated from the model hindcast.
It is important to consider temporal variation when
comparing model results with historical data. For
most of the historical data, sampling occurred during
a two-week or less period during the year (usually
summer). Compared to the annual lake-wide
average water column concentration in Figure4.5.15,
the monthly lake-wide average concentration in
Figure 4.5.16 demonstrates significant variation in
the water column PCB concentrations for any given
year. The monthly variation in the lake PCB
concentrations becomes most intense in the top layer
of the water column (within 10 meter depth) with the
highest concentrations occurring in the summer
months (July to September) and lowest
concentrations in the winter months (December to
February).
The vertical profile of PCB sediment concentrations
assembled based on the LM2-Toxic PCB hindcast
outputs for the depositional area of the lake was
compared to the three sediment core profiles (LM94-
015A, LM95-061A, and LM95-086A) measured and
analyzed for the LMMBP- The concentration profile
from the model is porosity normalized. Considering
the fact that the model-generated vertical PCB
sediment concentration profile is a lake-wide average
vertical profile and based on the outputs of the model
with a very coarse spatial segmentation, the model
did a fairly good job of matching the general trend of
the vertical profiles from the three sediment cores
and had the closest match with core LM95-061.
There are certainly unmatched portions between the
model-generated profile and the profiles from the
sediment cores. But it is very important to understand
that the common purpose of a hindcast is to provide
confirmation of the general conceptualization and
parameterization of a model. The successfulness of
a model hindcast should not be judged only on
whether the model output can match field
328
-------
3.5i-
LMT-Toxic model (lakewide annual average)
historical data
3.0 -I
• 2.0
u
§ 1.0
15 0
o 1940 1950 1960 1970 1980 1990 2000
4-<
Years
Figure 4.5.15. Annual lake-wide average total PCB water column concentrations from the LM2-Toxic
PCB hindcast simulation.
— LMT-Toxic model (lakewide monthly average, top layer)
— LMT-Toxic model (lakewide monthly average)
• historical data
2000
Figure 4.5.16. Monthly lake-wide average total PCB water column concentrations from the LM2-Toxic
PCB hindcast simulation. High concentration in the summer months and low concentration in the
winter months.
329
-------
CM
o Depositional zone (LM2-Toxic)
A LM94-15A (core data)
• LM95-061 (core data)
• LM95-086 (core data)
$ Depositional zone surfacial sediment average
(LMMBP)
15 1940 1950 1960 1970 1980 1990 2000
5 Years
Figure4.5.17. Annual average total PCB concentration profiles in the sediment depositional zone from
the LM2-Toxic PCB hindcast simulation.
observation at one point or a
temporal profile.
short portion of the
The discrepancy between the LM2-Toxic PCB
hindcast generated sediment profile and the profiles
from the sediment cores becomes most noticeable
for the portion after 1990. This is the part of the
profiles that represents the unconsolidated portion -
the surficial sediment mixed layer. The explanations
for the discrepancy could be:
1) The model does not have adequately
reconstructed historical physical conditions such
as wave, water circulation, and temperature time
functions. These time functions can dictate
organic carbon and PCB concentrations in the
lake and the solids cycling rates such as
resuspension, burial, and settling. For example,
a severe weather event could create a major
localized resuspension event in the lake and,
therefore, could elevate PCB concentrations in
the local water column and surficial sediments
along the path impacted by water transport. The
2)
elevated concentrations could also remain for a
long time after the major event. The recorded
maximum wave heights from the southern
National Oceanic and Atmospheric
Administration's (NOAA) buoy 45007 indicate that
a major storm event occurred on September 23,
1989 which induced a maximum wave height of
5.6 meters. The sediment profile from core
LM95-061A shows a step increase in PCB
concentrations just after the event. The step
increase would very likely be caused by re-
depositing the sediments with higher PCB
concentrations resuspended from a nearby area.
The unit for the sediment PCB concentration
output from the LM2-Toxic is ng/L. In order to
compare it with the PCB sediment profiles (ng/g
dw) from the cores, porosity is used to convert
the model-generated PCB sediment
concentration from the unit of ng/L to ng/g dw.
The converted PCB sediment concentration is
extremely sensitive to the porosity used in the
conversion. The value (0.953) for the porosity
330
-------
used in the conversion is the average of the
porosities at the sediment sampling sites located
within the depositional area of the lake. The
range of the porosities at these sediment
sampling sites is between 0.943 and 0.966.
3) Some physical-chemical processes such as the
BNL and its associated transport were not built in
the LM2-Toxic model. A seasonal persistent BNL
with high total suspended solid is not an
uncommon phenomenon in the Great Lakes and
has been observed and documented by
numerous researchers (Chambers and Eadie,
1980, 1981; Eadie et al., 1984; Baker and
Eisenreich, 1985; Baker ef a/., 1991; Eadie, 1997;
Hawley, 2003). The BNL is usually formed in a
large lake such as a Great Lake during summer
stratification. The thickness of the BNL is from a
few meters up to 10 meters above the bottom of
the lake (Chambers and Eadie, 1980, 1981;
Eadie et al., 1984). Total suspended solids and
associated chemicals in both particulate and
dissolved phases increase exponentially from the
top to the bottom of the BNL. The suspended
solids in the hypolimnion with less attached
pollutant will adsorb more PCBs when the solids
settle through this layer to the lake bottom. The
existence of the BNL and the physical-chemical
processes associated with this layer could
elevate the concentration of PCBs in the
sediments. Compared to the hypolimnion,
samples collected in the BNL during the LMMBP
(August 1994, August 1995, and September 1995
cruises) show a 4 to 20 times higher particulate
PCB concentrations and 1.5 times higher
dissolved PCB concentrations. Because the BNL
and the associated processes are not built in the
LM2-Toxic, the PCB hindcast could underpredict
PCB concentrations in certain areas of the lake
sediments.
4) There are uncertainties associated with the
derived historical primary production, PCB
loading, and vapor-phase time functions.
5) The uncertainties related to model parameters
including calibration parameters could also
propagate through the model to the predicted
sediment PCB concentrations.
Mass budget checking is a necessary step to ensure
that a water quality model does not generate or lose
mass through the entire simulation period. Figure
4.5.18 shows the mass budget of IPCB (sum of 54
PCB congeners for Lake Michigan during the
simulation period of the LM2-Toxic PCB hindcast
(1949-1995). The inventories in the mass budget
diagram represent the IPCB masses left in both the
water column and sediment at the end (last time
step) of the model simulation. Each mass
component with an arrow in Figure 4.5.18 indicates
a mass flux in or out of the system during the entire
simulation period. A simple mass budget check was
done based on the numbers in Figure 4.5.18. Table
4.5.7 lists the IPCB inventories, the mass fluxes, and
the calculation of the mass budget checking. The
result of the mass budget checking shows that there
is only about 0.66% of mass not being accounted for
over the 47 years of the hindcast simulation period.
This small amount of uncounted mass over the entire
hindcast period should not have much of an impact
on the accuracy of PCB concentrations computed by
the LM2-Toxic.
The most reliable confirmation of the LM2-Toxic
model comes from the comparison between the total
PCB sediment inventory (total PCB = IPCB x 1.1668
= 52,278 kg x 1.1668 = 60,998 kg) calculated from
the LM2-Toxic PCB hindcast and the latest
estimation of the total PCB sediment inventory (about
60,000 kg) provided by Eadie and Van Hoof
(personal communication, see Part 1, Chapter 7).
The factor 1.1668 used to convert the IPCBs to the
total PCBs was derived from the LMMBP data (Part
1, Chapter 3). The latest sediment inventory
estimate from Eadie and Van Hoof was based on a
large quantity of information collected during the
LMMBP for Lake Michigan sediment, including more
than 60 box core PCB sediment profiles and also
roughly 60 ponar and gravity core samples. A few
previous and most recent studies listed in Table 4.5.6
indicate that the lake-wide sediment inventory is likely
between 60,000 kg and 75,000 kg. The sediment
PCB inventory (60,988 kg) computed from the LM2-
Toxic PCB hindcast is within the range provided in
the literature.
331
-------
loads
168048 kg
net volatilization
114401 kg
water column inventory
692kg
dispersion
30kg
Chicago diversion
168kg
net advection
1643 kg (2516-873)
sediment inventory
52278 kg
Figure 4.5.18. IPCB mass budget of Lake Michigan during the period of the LM2-Toxic PCB hindcast
(1949-1995).
Table 4.5.7. Calculations in PCB Mass Budget Checking for the LM2-Toxic PCB Hindcast
Mass of In-Flux (kg)
Mass of Out-Flux (kg)
Inventory (Kg)
Loads
Dispersion
16,048 Net Volatilization
30 Net Advection
Chicago Diversion
114,401 Water Column
1,643 Sediment
168
692
52,278
Difference in PCB mass over the hindcast simulation period (47 years) = (Inventory + Mass of out-flux)
(Mass of in-flux) = (692 + 52,278 + 114,401 + 1,643 + 168) - (168,048 + 30) = 169,182 - 168,078 =
1,104kg
Percentage of the difference in PCB mass over the PCB total input into Lake Michigan = difference in
PCB mass over the hindcast simulation/mass of in-flux =1,104 kg/168,078 kg = 0.657%
332
-------
Given the adequate level of reconstructed historical
forcing functions including loading and vapor phase
concentrations, the overall results from the LM2-
Toxic PCB hindcast demonstrate that the model is
able to generate PCB concentrations in both the
water column and sediment that compare reasonably
well with the available historical data and estimated
sediment total PCB inventories.
References
Anderson, D.J., T.B. Bloem, R.K. Blankenbaker, and
T.A. Stanko. 1999. Concentrations of
Polychlorinated Biphenyls in the Water Column of
the Laurentian Great Lakes: Spring 1993. J.
Great Lakes Res., 25(1 ):160-170.
Baker, J.E. and S.J. Eisenreich. 1985. Chlorinated
Hydrocarbon Cycling in the Benthic Nepheloid
Layer of Lake Superior. Environ. Sci. Technol.,
19(9):854-861.
Baker, J.E., S.J. Eisenreich, and B.J. Eadie. 1991.
Sediment Trap Fluxes and Benthic Recycling of
Organic Carbon, Polycyclic Aromatic
Hydrocarbon, and Polychlorobiphenyl Congeners
in Lake Superior. Environ. Sci. Technol.,
25(3):500-509.
Bamford, H.A., D.L. Poster, R.E. Huie, and J.E.
Baker. 2002. Using Extrathermodynamic
Relationships to Model the Temperature
Dependence of Henry's Law Constants of 209
PCB Congeners. Environ. Sci. Technol.,
36(20):4395-4402.
Bicksler, J. 1996. PCBs in the Spring-Time Water
Column of the Great Lakes. M.S. Thesis,
University of Minnesota, Minneapolis, Minnesota.
160pp.
Bierman, V.J., J.V. DePinto, T.C. Young, P.W.
Rodgers, S.C. Martin, and R. Raghunathan.
1992. Development and Validation of an
Integrated Exposure Model for Toxic Chemicals in
Green Bay, Lake Michigan. U.S. Environmental
Protection Agency, Office of Research and
Development, ERL-Duluth, Large Lakes Research
Station, Grosse lie, Michigan. 381 pp.
Buehler, S.S., I. Basu, and R.A. Hites. 2002. Gas-
Phase Polychlorinated Biphenyl and
Hexachlorocyclohexane Concentrations Near the
Great Lakes: A Historical Perspective. Environ.
Sci. Technol., 36(23):5051-5056.
Buehler, S.S., I. Basu, and R.A. Hites. 2004.
Causes of Variability in Pesticide and PCB
Concentrations in Air Near the Great Lakes.
Environ. Sci. Technol., 38(2):414-422.
Chambers, R.L. and B.J. Eadie. 1980. Nearshore
Chemistry in the Vicinity of the Grand River,
Michigan. National Oceanic and Atmospheric
Administration, Great Lakes Environmental
Research Laboratory, Ann Arbor, Michigan.
NOAA Technical Memorandum ERL GLERL-28,
28pp.
Chambers, R.L. and B.L. Eadie. 1981. Nepheloid
and Suspended Particulate Matter in Southeastern
Lake Michigan. Sedimentology, 28(3):439-447.
Delaware River Basin Commission. 2003.
Calibration of the PCB Water Quality Model for the
Delaware Estuary for Penta-PCBs and Carbon.
Delaware River Basin Commission, West Trenton,
New Jersey. 147 pp. (Website:
http:www.epa.gov/reg3wapd/tmdl/pa_tmdl/
delaware%20river/ModelCalibrationReport.pdf).
DePinto, J.V., R. Raghunathan, P. Sierzenga, X.
Zhang, V.J. Bierman, Jr., P.W. Rodgers, and T.C.
Young. 1993. Recalibration of GBTOX: An
Integrated Exposure Model for Toxic Chemicals in
Green Bay, Lake Michigan. Final Report. U.S.
Environmental Protection Agency, Office of
Research and Development, ERL-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
132pp.
DePinto, J.V., W.M. Larson, J. Kaur, and J. Atkinson.
2003. LOTOX2 Model Documentation - In
Support of Development of Load Reduction
Strategies and a TMDL for PCBs in Lake Ontario.
Submitted to New England Interstate Water
Pollution Control Commission, Boott Mills South,
Lowell, Massachusetts. 122 pp.
333
-------
Eadie, B.J., R.L. Chambers, W.S. Gardner, and G.L.
Bell. 1984. Sediment Trap Studies in Lake
Michigan: Resuspension and Chemical Fluxes in
the Southern Basin. J. Great Lakes Res.,
10(3):307-321.
Eadie, B.J., H.A. Vanderploeg, J.A. Robbins, and
G.L. Bell. 1990. Significance of Sediment
Resuspension and Particle Settling. In: M.M.
Tilzer and C. Serruya (Eds.), Large Lakes:
Ecological Structure and Function, pp. 196-209.
Springer-Verlag, New York, New York.
Eadie, B.J. 1997. Probing Particle Processes in
Lake Michigan Using Sediment Traps. Water, Air,
and Soil Pollut., 99:133-139.
Endicott, D.D. 2005. 2002 Lake Michigan Mass
Balance Project: Modeling Total PCBs Using the
MICHTOX Model. In: R. Rossmann (Ed.),
MICHTOX: A Mass Balance and Bioaccumulation
Model for Toxic Chemicals in Lake Michigan, Part
2. U.S. Environmental Protection Agency, Office
of Research and Development, National Health
and Environmental Effects Research Laboratory,
MED-Duluth, Large Lakes Research Station,
Grosse lie, Michigan. EPA/600/R-05/158,140 pp.
Endicott, D.D., W.L. Richardson, and D.J. Kandt.
2005. 1992 MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan. In: R. Rossmann (Ed.),
MICHTOX: A Mass Balance and Bioaccumulation
Model for Toxic Chemicals in Lake Michigan, Part
1. U.S. Environmental Protection Agency, Office
of Research and Development, National Health
and Environmental Effects Research Laboratory,
MED-Duluth, Large Lakes Research Station,
Grosse lie, Michigan. EPA/600/R-05/158,140 pp.
Filkins, J.C., J.M. Townsend, and S.G. Rood. 1983.
Organochlorines in Offshore Waters of the Great
Lakes, 1981. U.S. Environmental Protection
Agency, Office of Research and Development,
ERL-Duluth, Large Lakes Research Station,
Grosse lie, Michigan. 12pp.
Gardiner, R.D., M.T. Auer, and R.P. Canale. 1984.
Sediment Oxygen Demand in Green Bay (Lake
Michigan). In: M. Pirbazari and J.S. Devinny
(Eds.), Proceedings of the 1984 Specialty
Conference on Environmental Engineering, pp.
514-519. American Society of Civil Engineers,
New York, New York.
Gobas, F.A.P.C., M.N.Z. Graggen, and X. Zhang.
1995. Time Response of the Lake Ontario
Ecosystem to Virtual Elimination of PCBs. Environ.
Sci. Technol., 29(8):2038-2046.
Golden, K.A., C.S. Wong, J.D. Jeremiason, S.J.
Eisenreich, G. Sanders, J. Hallgren, D.L.
Swackhamer, D.R. Engstrom, and D.T. Long.
1993. Accumulation and Preliminary Inventory of
Organochlorines in Great Lakes Sediments.
Water Sci. Technol., 29(8-9):19-31.
Hall, D. and D. Robertson. 1998. Estimation of
Contaminant Loading from Monitored and
Unmonitored Tributaries to Lake Michigan for the
USEPA Lake Michigan Mass Balance Study.
Quality Systems and Implementation Plan.
Submitted October 23,1998. U.S. Environmental
Protection Agency, Great Lakes National Program
Office, Chicago, Illinois. 19 pp.
Hawker, D.W. and D.W. Connell. 1988. Octanol-
Water Partition Coefficients of Polychlorinated
Biphenyl Congeners. Environ. Sci. Technol.,
22(4):382-387.
Hawley, N. 2003. Observations of the Intermediate
and Benthic Nepheloid Layers in Southern Lake
Michigan During the Summer of 1995. National
Oceanic and Atmospheric Administration, Great
Lakes Environmental Research Laboratory, Ann
Arbor, Michigan. NOAA Technical Memorandum
ERLGLERL-124, 30pp.
Hillery, B.L., I. Basu, C.W. Sweet, and R.A. Hites.
1997. Temporal and Spatial Trends in a Long-
Term Study of Gas-Phase PCB Concentrations
Near the Great Lakes. Environ. Sci. Technol.,
334
-------
Jeremiason, J.D., K.C. Hornbuckle, and S.J.
Eisenreich. 1994. PCBs in Lake Superior, 1978-
1992: Decreases in Water Concentrations Reflect
Loss by Volatilization. Environ. Sci. Technol.,
28(5):903-914.
Lefkovitz, L.F. 1987. The Particle Mediated
Fractionation of PCBs in Lake Michigan. M.S.
Thesis, University of Wisconsin, Madison,
Wisconsin. 238 pp.
Mackay, D. 1988. A Model of PCB Behaviour in
Lake Ontario. In: Report on Modeling the
Loading-Concentration Relationship for Critical
Pollutants in the Great Lakes, Appendix V. Task
Force on Chemical Loadings, Water Quality
Board, International Joint Commission, Windsor,
Ontario, Canada. 275 pp.
McCarty, H.B., J. Schofield, K. Miller, R.N. Brent, P.
Van Hoff, and B. Eadie. 2004. Results of the
Lake Michigan Mass Balance Study:
Polychlorinated Biphenyls and frans-Nonachlor
Data Report. U.S. Environmental Protection
Agency, Great Lakes National Program Office,
Chicago, Illinois. EPA/905/R-01/011, 289 pp.
Miller, S.M., M.L. Green, J.V. DePinto, and K.C.
Hornbuckle. 2001. Results from the Lake
Michigan Mass Balance Study: Concentrations
and Fluxes of Atmospheric Polychlorinated
Biphenyls and frans-Nonachlor. Environ. Sci.
Technol., 35(2):278-285.
Miller, S.M. 2003. The Effects of Large-Scale
Episodic Sediment Resuspension on Persistent
Organic Pollutants in Southern Lake Michigan.
Ph.D. Thesis, The University of Iowa, Iowa City,
Iowa. 194pp.
Pearson, R.F., K.C. Hornbuckle, S.J. Eisenreich, and
D.L Swackhamer. 1996. PCBs in Lake Michigan
Water Revisited. Environ. Sci. Technol.,
30(5):1429-1436.
Rice, C.P., B.J. Eadie, and K.M. Erstfield. 1982.
Enrichment of PCBs in Lake Michigan Surface
Films. J. Great Lakes Res., 8(2):265-270.
Robbins, J.A., N.R. Morehead, R.W. Rood, D.N.
Edgington, and S. Meyer. 1999. Accumulation
and Near-Surface Mixing of Sediments in Lake
Michigan as Determined for the Lake Michigan
Mass Balance Program, Volumes 1 and 2. Project
Report. U.S. Environmental Protection Agency,
Office of Research and Development, ERL-Duluth,
Large Lakes Research Station, Grosse lie,
Michigan. 503 pp.
Rodgers, P.W., T.A.D. Slawecki, J.V. DePinto, and
W. Booty. 1988. LTI Toxics Model Application:
PCBs in Lake Ontario - An Exploratory
Application. In: Report on Modeling the Loading-
Concentration Relationship for Critical Pollutants
in the Great Lakes, Appendix III. Task Force on
Chemical Loadings, Water Quality Board,
International Joint Commission, Windsor, Ontario,
Canada. 275 pp.
Rygwelski, K.R. (Ed.). 2005. Lake Michigan Mass
Balance Project (LMMBP) PCB Peer Review
Report. U.S. Environmental Protection Agency,
Office of Research and Development, National
Health and Environmental Effects Research
Laboratory, MED-Duluth, Large Lakes Research
Station, Grosse lie, Michigan. 31 pp.
Schneider, A.R., H.M. Stapleton, J. Cornwell, and
J.E. Baker. 2001. Recent Declines in PAH, PCB,
and Toxaphene Levels in the Northern Great
Lakes as Determined from High Resolution
Sediment Cores. Environ. Sci. Technol.,
35(19):3809-3815.
Schwab, D. and D. Beletsky. 1998. Lake Michigan
Mass Balance Study: Hydrodynamic Modeling
Project. National Oceanic and Atmospheric
Administration, Great Lakes Environmental
Research Laboratory, Ann Arbor, Michigan.
NOAA Technical Memorandum ERL GLERL-108,
53pp.
Swackhamer, D.L. and D.E. Armstrong. 1987
Distribution and Characterization of PCBs in Lake
Michigan Water. J. Great Lakes Res., 13(1):24-
36.
Thomann, R.V. and D.M. Di Toro. 1983. Physico-
Chemical Model of Toxic Substances in the Great
Lakes. J. Great Lakes Res., 9(4):474-496.
335
-------
U.S. Environmental Protection Agency. 1997. Lake
Michigan Mass Balance Project Modeling Work
Plan. Final Report. U.S. Environmental
Protection Agency, Office of Research and
Development, ERL-Duluth, Large Lakes Research
Station, Grosse He, Michigan. 30 pp.
U.S. Environmental Protection Agency. 2000.
Hudson River PCBs Reassessment RI/FS Phase
3 Report: Feasibility Study, Books 1-6. U.S.
Environmental Protection Agency, Region 2, New
York, New York. 2,095 pp.
Vollenweider, R.A., M. Munawar, and P. Stadelmann.
1974. A Comparative Review of Phytoplankton
and Primary Production in the Laurentian Great
Lakes. J. Fish. Res. Board Canada, 31(5):739-
762.
Wisconsin Department of Natural Resources. 2003.
Estimates of PCB Mass, Sediment Volume, and
Surface Sediment Concentrations in Operable Unit
5, Green Bay Using an Alternative Approach.
Wisconsin Department of Natural Resources,
Madison, Wisconsin. White Paper Number 19,23
pp.
336
-------
PART 4
LM2-TOXIC
Appendix 4.5.1. Results From Thermal Balance Model
337
-------
o
-------
Segment 7
120 240 359 479 599
Time (Day 0 = 4/1/94)
40
32
o
-------
30
o 24
18
12
6
Segment 1 1
120 240 359 479 599
Time (Day 0 = 4/1/94)
Segment 12
120 240 359 479 599
Time (Day 0 = 4/1/94)
30
o 24
18
6
0
Segment 13
120 240 359 479 599
Time (Day 0 = 4/1/94)
30
Segment 14
120 240 359 479 599
Time (Day 0 = 4/1/94)
30
o 24
a
.3 18
o
aJ 10
CL 12
E
01 _
H- 6
0
Segment 15
120 240 359 479 599
Time (Day 0 = 4/1/94)
30
o 24
CP
2 18
a) - ,-,
CL 1 2
Segment 16
CD
I
. ~
a. I /
0
0
Segment 18
120 240 359 479 599
Time (Day 0 = 4/1/94)
0 120 240 359 479
Time (Day 0 = 4/1/94)
599
-------
30
o 24
0)
3 18
12
6
0
Segment 20
120 240 359 479 599
Time (Day 0 = 4/1/97)
30
o 24
-------
30
24
o
-------
o
03
13
O
i_
0)
CD
30
24
18
12
6
0
Segment 35
0 120 240 359 479 599
Time (Day 0 = 4/1/97)
Temperature
343
-------
o
CD
24 -
12
0
Segment 36
0 120 240 359 479 599
Time (Day 0 = 4/1/94)
o
QJ
O
\_
CD
30
24
12
0
Segment 37
0 120 240 359 479 599
Time (Day 0 = 4/1/94)
30
o
CD
o
CD
24 -
18
12
6
0
Segment 38
0 120 240 359 479 599
Time (Day 0 = 4/1/94)
o
CD
i_
=>
-»-j
o
0)
30
24
12
0
Segment 39
0 120 240 359 479 599
Time (Day 0 = 4/1/94)
o
CD
30
24
18
-
0)
Q_
E
0)
12
0
Segment 40
0 120 240 359 479 599
Time (Day 0 = 4/1/94)
30
24
o
-------
Vertical Dispersion Coefficient Plot
Interface b/w segment 1 and 11
Vertical Dispersion Coefficient Hot
Interface b/w segment 2 and 12
Vertical Dispersion Coefficient Rot
Interface b/w segment 3 and 13
J2
D QOOO
100 200 300 «0 500 600 TOD
Day (0 = 04/01/94)
100 200 300 400 500 600 TOO
Day (0=04/01/94)
100 200 300 400 500 600 TOO
Day(0 = 04/01/94)
Vertical Dispersron Coefficient Plot
Interface b/w segment 4 and 14
Vertical Dispersion Coefficient Plot
Interface b/w segment 5 and 15
Vertical Dispersion Coefficient Plot
Interface b/w segment G and 16
»
0 0003
So
V ^ 0002
0 •—
«
0 0001
s
" Q003
o «•
^ >£ 0002
e E^
w
fe 0.001
.2
\ 1 V ^X /
1 Ul^
-------
Vertical Dispersion Coefficient Plot
Interface b/wsegmentH and 20
Vertical Dispersion Coefficient Plot
Interface b/wsegment 12 and 21
Vertical Dispersion Coefficient Plot
Interface b/w segment 13 and 22
\ j v J
2
c
.2
" 0.003
£
S^r
O -v 0.002
c E
o •—
E
1 °™
° 0030
\ J v_ /
n
« 0003
C
o "»"
° -V 0.002
c E
o
« 0001
i
0 0000
*- y >- / —
0 100203300400500600700
Day (0 = 04/01/94)
100 200 300 400 500 600 700
Day (0 = 04/01/94)
100 200 300 «0 £00 600 700
Day (0 = 04/01/94)
Vertical Dispersion Coefficient Plot
Interface b/w segment 14 and 23
Vertical Dispersion Coefficient Plot
Interface b/w segment 15 and 24
Vertical Dispersion Coefficient Plot
Interface b/w segment 16 and 25
0 100200X0400500600700
Day(0 - 04/01/94)
100 200 300 4DO 500 600 700
Day (0 = 04/01/94)
t> -C 0002
100 200 300 400 500 600 700
Day(0 = 04/01794)
Vertical Dispersion Coefficient Plot
Interface b/w segment 17 and 26
Vertical Dispersion Coefficient Plot
Interface b/w segment IBand 27
Vertical Dispersion Coefficient Plot
Interface b/w segment 19 and 28
C
** 0003
i
3^0002
i oooi
a.
5 OOOD
c
^ 0003
O "S"
U r£ 0002
c E
£ 0001
1
5 000)
e 0003
o| 0.002
1 «"
^j0 r-^,^^ r-v.
100 200 300 400 500 600 700
Day{0 = 04/01/94)
100 200 300 400
Day (0=04/01/94)
500600700
100 200 300 400 500 600 TOO
Day (0=04/01/94)
Vertical Dispersion Coefficient
346
-------
Vertical Dispersion Coefficient Plot
Interface b/w segment 20 and 29
Vertical Dispersion Coefficient Plot
Interface b/w segment 21 and 30
(« 0.004
I
" 0003
8 l?
°
-------
Vertical Dispersion Coefficient Plot
Interface b/w segment 29 and 36
Vertical Dispersion Coefficient Plot
Interface b/w segment 30 and 37
c
0>
'o
£
o
o 1?
= ~E
o ^~
0
a
.2
0.004
0.003
0.002
0001
n onn
^__y^_ Ar7 ^-^ f\f
f
.2!
O
t
o
o
c
0
e
0)
Q.
5
U UU4 •
0.003
I?
•V 0.002
0.001
0.000 J
> — - — , ~y^ ~X -.. __,
J»
0.003
0.002
0.001
100 200 300 400 500 600
Day (0 = 04/01/94)
100 200 300 400 500 600 700
Day (0 = 04/01/94)
Vertical Dispersion Coefficient Plot
Interface b/w segment 33 and 40
Vertical Dispersion Coefficient Plot
Interface b/w segment 34 and 41
2
" 0.003
O yT
*-* i — -''yv . , — , — > r . ^r~i o
0.003
0.002
0.001
0.000 -
\ ^ ^_/ \ Ht~
100 200 300 400 500 600 700
Day (0 = 04/01/94)
200 300 400 500
Day (0 = 04/01/94)
Vertical Dispersion Coefficient
348
-------
Appendix 4.5.2
Calibrated Results for Organic Carbons
349
-------
4-r
_ 3-
2"
75i
£ 2-
o
o
Q 1-
segment 1
1 T-1 «^- * * *
segment 2
—U--^t--^.
•
segment 3
— r-*-*-*-~^«^— 1<_.
segment 4
Jan94 Jul94 De'c94 JuJ95 Dec95 Jan94 JuJ94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 JuJ95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec£
_ 3
_j
^i
£ 2.
8
° 1.
Q
segment 5
» j-t— *-*
• * "~
segment 6
»
— - t t i —f-.~
» • ' • ~-
segment 11
. j J-"- t * *
segment 12
- • • « -*^ « ^» »
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Ju(95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
A
-, 3-
_i
— 2-
O
O
° 1
0
Ja
4
^ 3
Q"
T5>
£ 2
O
O
Q 1-
0-
Ja
4-
— 3^
_j
1
& 2
O
8 ,
0-
segment 13
segment 14
— *-*-*-*-~__t_^JU*^
segment 15
— i i • "' „ ».*
T- 4— — ^— -»•*-.
segment 16
h i • *~ -~_A_-i-L.
n94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
segment 20
- . ,-f * » *
segment 21
1 ' | T ** — A *-
segment 22
— -*- * *
« ^^ r*~
segment 23
* * • — r*-
194 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul^S Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
segment 24
' » " « -* — v *>-
segment 25
— »
• *-.— •'"•.*-
~s Model Output
» Cruise Mean
error bars = standard error
Dissolved Organic Carbon (mg/L)
Lake Michigan layers 1 , 2, 3
upper 30 meter
water segments
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
350
-------
DOC (mg/L)
3 ->• W W *
i i i i
segment 29
segment 30
segment 31
segment 32
* - t t
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
^ 3-
.§ 2-
O
o
0-
segment 33
~~ — * * * — i Ji
•
segment 34
— 1*^_^__ *
"
segment 36
* *
segment 37
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
-s Model Output
,) » Cruise Mean
_ 3-
I 2-
o
8,.
0-
segment 38
error bars = standard error
OBa^cl ll Dissolved Organic Carbon (mg/L) ^9^enl 40
Lake Michigan layers 4 & 5
30 meters to bottom
water segments
— * 1 . ' •
, «
.
segment 41
i^ * +
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Ju)95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
351
-------
4-r
^ 3-
_J
-§- 2-
o
0
g
segment 7
fVH~l*.» +
T L^w~ "* n^. Ay^^* " ^^*»
segment 8
^[\
L - — -^ ^Tf^f Lr-111-
i
^* «•
segment 9
i »A 1
^S *A/M
*
segment 1 0
Jan94 Jul94 Dec94 Jui95 Dec95 Jan94 JuJ94 Dec94 JuJ95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
_ — . — .
3-
D)
& 2-
o
o
1 -
o
segment 17
^
^4— *-^j^l*^
segment 1 8
»
«w4^v-i^ L-l*^
I segment 19
\ * t
^^ t
^V/^-^-^iJ^
segment 26
'--^U-^.^^
*
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9!
4
3-
D)
O
o
Q 1
0-
segment 27
**"**X^1 ff^. _*-u-.JL
»^1V^™^C*' """
segment 28
i
V
V"""^
» T^^^V^T^XJ^
» **
segment 35
ta%JU
•«v Model Output
» Cruise Mean
error bars = standard error
Dissolved Organic Carbon
(mg/L)
Green Bay
water segments
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
352
-------
4-r
^3-
j'
15)
§2-
o
*,.
Q
segment 42
3-
2-
1 -
n.
segment 43
3-
2-
1 -
n.
segment 44
3-
2-
1 •
n.
segment 45
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
A A A A
_3-
T5>
§2-
O
o
°1.
Q
segment 46
3-
2-
1-
n
segment 47
3-
2-
1.
n.
segment 48
3-
2-
1-
n.
segment 49
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Ju!95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
A A A A
_3-
_j
-§. 2-
0
0
°1-
0-
Ja
4-
_3-
2"
^i
£2-
O
o
Q1-
0-
Jat
4-
~3-
!>>
Q
1.
o.
segment 50
3-
2
1 -
n.
segment 51
3-
2
1 •
n.
segment 52
3-
2
1
n.
segment 53
i94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
A A A
segment 54
3-
2-
1-
n.
segment 55
3-
2-
1
n
segment 56
3-
2-
1 •
n.
segment 57
194 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
A
segment 58
3-
2-
1-
0-
—• Model Output
Dissolved
Organic Carbon
(mg/L)
Lake Michigan
sediment segments
segment 59
A
3
2
1
0
segment 60
3
2
1
n
segment 61
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
353
-------
4-r
_ 3-
^
~ai
£ 2-
o
o
Q 1.
segment 62
i -
3-
2-
1 -
n
segment 63
3-
2-
1 -
n.
segment 64
3-
2-
1-
n.
segment 65
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
A * * A
3-
_j
15)
£ 2-
o
o
Q 1.
o
segment 66
3-
2-
1 -
n
segment 67
3-
2-
1-
n.
segment 68
3-
2-
1-
n.
segment 69
»—
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
4 A A A
_ 3-
_J
^)
S 2
O
o
Q 1
0-
Ja
4
_ 3-
3"
^
£ 2-
O
o
Q 11
0-
segment 70
3-
2-
1 -
ru
segment 71
3-
2
1 .
n
segment 72
3-
2-
1-
n
segment 73
ri94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9"
. .. . . A A „
segment 74
3-
2-
1-
n.
segment 75
3-
2
1-
n
segment 76
3-
2-
1-
segment 77
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jui94 Dec94 Jul95 Dec9
~- Model Output
Dissolved
Organic Carbon
(mg/L)
Lake Michigan
sediment segments
354
-------
3-
CT
I 2-
o
O
0 1.
Q
segment 78
"~— ^
^
3-
2
1 •
n.
segment 79
y|u~>^*^~^ -.
^~'-^'*«^Vx<-^>«lVv"^"">'
3-
2-
1 -
n
segment 80
^-v^
^^^^-^"^X/**"" ^/~'*~v
3-
2-
1-
n
segment 81
^^^^%v*-<-^^ll*^X^—*A^*sJ**^
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec£
A A A t
^ 3-
j"
£ 2-
o
o
u 1-
o
segment 82
^***~~r~**s~~*^S-'
3-
2-
1-
n.
segment 83
>ff^ftUUi. .t i^Mti^ --»-
"^ "*-*"-*"
3-
2-
1H
n
segment 84
i^ftUfct,^ 1*,. ^^t
J"l^"-*u-'p"*'
3
2-
1.
n
segment 85
^^^^V-^^, <*t*^^^^_^_._JL_ __fUn_
-• I_TT»*
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 DecS
4 A A A
Cj
ll-
o
n
0 1-
Q-
segment 86
^ — ^ "^-i* ^ - -
^^^~
3-
2
1 .
n.
segment 87
^^^ Hi.^^^ -nf*^^^ m lh* ""^
3-
2-
1-
n
segment 88
I
^^X^^/X^N
3
2
1-
n
segment 89
1
^XJ^v J^
^W ™^^n^^ \f>
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 DecS
4 A . A t
„ 3-
1 2-
o
o
0 1.
0-
segment 90
v
^X/^-v-^J^
3-
2
1 -
n.
segment 91
I
v.^
^W ^^u-^^ \/*
3-
2-
1-
n
, » ,
vVV\ r^jlA
^\r ^
segment 92
3-
2
1 -
n
\ A i
VV\A w\ /u
v^y VI
segment 93
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 JuJ94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
3-
_i
01
E -i
O
n
1.
n-
W\A Af\
VV /A v\
\^J Vlv v
N/
segment 94
•«. Model Output
Dissolved
Organic Carbon
(rng/L)
Green Bay
sediment segments
Jan94 Jul94 Dec94 Jul95 Dec95
355
-------
0.5-1
0.4-
=6, 0.3-
e
0 0.2
CD
0.1-
segment 1
fV"*^ /^"^v
^/ S*\±s .*V
-
segment 2
«^-». A""*,*'.
-/ .^\^/ /«x
segment 3
«/*«~y y'^>~*\
^/ +~*\^jl »^V
segment 4
W**~-v /"*"«\
— ' » *\^X *\
Jan94 JuJ94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 JuJ95 Dec95 Jan94 Jui94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 JuJ95 Dec£
0.5-
0.4
Bi 0.3-
E
0 0.2-
CQ
0.1-
segment 5
±A*^~^ /^vs_
y VK^/* v^\
"
.
segment 6
«.rt_jT-^ /^^"••M.
^/ +?\*S « v\
segment 11
I
I
4 ^-r^,^ y**^^^
-/ r*v^x /^
segment 12
T
1
f -»-l'u^ v'*'^^
^/ •'v^x /?v>
Jan94 Ju!94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 JuJ94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jui95 Dec9
O.b-
0.4-
^ 0.3]
0 0.2-
CD
0.1-
segment 13
i
*/•">-»_ v^">^
-X^^K^XVX,
segment 14
i
., , /*v^_
^/"^X^XVX,
segment 15
^
* j— u»<_
-/^^'V^rT^V
segment 16
4
* y*^»>.
^/*~*\^r^\
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
0.4-
Bi 0.3-
0 0.2
CD
0.1-1
0-
segment 20
«Vr
-^ ^^ — i^— "J-^
segment 21
«il
^/ 1^*S — *s~*~^
•
segment 22
« » *
^/ "f^^^^t^.
segment 23
f« I
^/"^^^s^.X*****^
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
0.4 ]
? 03l
E
0 0.2-
CQ
0.1-
o-
segment 24
4
segment 25
* « «
^y^-^A^/^*^
-% Model Output
« Cruise Mean
error bars = standard error
Biotic Carbon (mg/L)
Lake Michigan layers 1, 2, 3
upper 30 meter
water segments
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
356
-------
D)
O
0)
u.u-
0.4-
0.3-
0.2 J
0.1-
n-
segment 29
***
^.-L^r *, r-,t ,^
^^ *" *
segment 30
segment 31
segment 32
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
0.4.
'& 0.3.
E
0 0.2.
CD
0.1-
0-
segment 33
*««
segment 34
**'
segment 36
^^f +
segment 37
r— * *
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
^i
o
CO
0.4-
0.3-
0.2-
0.1-
n-
segment 38
• *— «-T «• * i -
segment 39
• _*.
»» Model Output
» Cruise Mean
error bars = standard error
Biotic Carbon (mg/L)
Lake Michigan layers 4 & 5
30 meters to bottom
water segments
^ ^'
•
«
. 4
segment 40
*-i
segment 41
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
357
-------
E,
g
CO
0.4
0.3-
0.2-
0.1-
segment 7
segment 8
segment 9
2.0
1.6-
1.2-
0.8-
0.4-
0
segment 10
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
0.5-,
O)
O
GO
0.4.
0.3.
0.2-
0.1-
segment 17
segment 18
segment 19
segment 26
0
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
Q.5-,
01
-------
0.5
0.4-
0.3-
0.2-
0.1
segment 1
segment 2
segment 3
segment 4
0
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 JuI94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
0.5,
0.4.
0.3
0.2.
0.1.
segment 5
segment 6
segment 11
segment 12
0
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
0.5-,
T5,
0
0.
0.4
0.3
0.2.
0.1
0
segment 13
segment 14
segment 16
Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
0.5'
O
O
a.
0.4-
0.3
0.2
0.1-
segment 20
segment 21
segment 22
segment 23
0
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
0.4
01 0.3-
o 0.2
segment 25
•s. Model Output
« Cruise Mean
error bars = standard error
Participate Detrital Carbon (mg/L)
Lake Michigan layers 1 , 2, 3
upper 30 meter
water segments
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
359
-------
0.5
0.4-
|> 0.3-
n 0.2
0.1
segment 29
segment 30
segment 31
segment 32
0
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
0.4-
^ 0.3.
0 0.2.
0.
O.L
0
segment 33
JL
V. ^^SI^V^^L ^^^^
segment 34
* .
^ *(«W'V^SL ~ff*^^
x«x \_ ^^
* «
segment 36
I **
- _ ~ T \"***~ ^T I ~
* * *
.
segment 37
- — I iT'^N^. ^t^^'
* * #
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec
0.5-,
n A
;J
0) 0.3-
8 0.2-
0.
01-
0-
segment 38
~«>^iJ, . ~ ^ r__^»^
* I * *"~ *
1— i i
segment 39
•^. Model Output
*• Cruise Mean
error bars = standard error
Participate Detrital Carbon
(mg/L)
Lake Michigan layers 4 & 5
30 meters to bottom
water segments
•~-, * — :f^**
^*\\ '*"' «
segment 40
i »J»
<***4*~*~**f^****^jr**~*a™"^
* « * **
segment 41
T
r\v-^<~-.»,..-xv>^_
-------
O
Q
Q.
1.0
0.8-
0.6-
0.4-
0.2
0
segment 7
1.0
0.8-
0.6-
0.4-
0.2-
0
segment 8
1.0
0.8-
0.6
0.4-
0.2-
0
segment 9
0
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
1.0
O
Q
0.
0.8-
0.6.
0.4
0.2-
0
segment 17
1.0
0.8.
0.6.
0.4
0.2
segment 18
0
1.0
0.8-
0.6.
0.4
0.2.
0
segment 19
1.0
0.8-
0.6-
0.4
0.2-
segment 26
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
0.8
3
"& 06
0 0.4.
Q.
0.2-
segment 27
^ *v- «
0.8-
0.6.
0.4-
0.2-
segment 28
0.8
0.6.
0.4-
0.2-
segment 35
-. Model Output
» Cruise Mean
error bars = standard error
Particulate
Detrital Carbon
(mg/L)
Green Bay
water segments
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
361
-------
10000-1
7500-
0.
£ 5000-
O
°- 2500-
n.
segment 42
segment 43
segment 44
segment 45
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
OOUU-,
7500-
5000-
2500-
n.
segment 46
segment 47
segment 48
segment 49
Jan94 Jul94 Dec94 Jul95 Dec95Jan94 Jul94 Dec94 Jul95 Dec95Jan94 Jul94 Dec94 Jul95 Dec95Jan94 Jul94 Dec94 Jul95 Dec95
1UUVJU -
_ 7500-
15>
,§ 5000-
O
O
°- 2500-
0.
segment 50
.
segment 51
segment 52
segment 53
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
10000-
7500-
£ 5000
o
Q
°- 2500 -I
segment 54
•~> Model Output
Participate
Detrital Carbon
(mg/L)
Lake Michigan
sediment segments
segment 55
segment 56
segment 57
Jan94 Jul94 Dec94 Jul95 Dec95Jan94 Jul94 Dec94 Jul95 Dec95Jan94 Jul94 Dec94 Jul95 Dec95Jan94 Jul94 Dec94 Jul95 Dec95
„ 7500-
_j
^>
§5000-
Q
Q.
2500-
0-
segment 58
segment 59
-
segment 60
segment 61
362
-------
IUIAAJ •
,-. 7500-
I
~ 5000-
o
Q
Q.
2500-
Q
segment 62
segment 63
segment 64
segment 65
Jan94 Jul94 Dec94 Jul95 Dec95Jan94 Jul94 Dec94 Jul95 Dec95Jan94 Jul94 Dec94 Jul95 Dec95Jan94 Jul94 Dec94 Jul95 Dec£
,-, 7500-
_i
^i
~ 5000-
O
Q
0.
2500-
Q
segment 66
.
segment 67
segment 68
segment 69
Jan94 Jul94 Dec94 Jul95 Dec95Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
innnn
_ 7500-
_j
£-5000-
Q
Q.
2500-
0 -
segment 70
segment 71
.
segment 72
-
segment 73
Jan94 Jul94 Dec94 Jul95 Dec95Jan94 Jul94 Dec94 Jul95 Dec95Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
10000
„ 7500
^
JSODO.
Q
0_
2500-
0-
segment 74
segment 75
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 J
segment 76
.
segment 77
u!95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 DecS
—• Model Output
Particulate
Detrital Carbon
(mg/L)
Lake Michigan
sediment segments
363
-------
10000.
_ 7500-
O)
E. 5000
0
Q
°- 2500-
n.
segment 78
•
segment 79
segment 80
segment 81
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
en
o
Q
uuuu-
7500-
5000-
2500-
n.
segment 82
segment 83
segment 84
segment 85
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
7500-
5000 .
2500.
o.
segment 86
segment 87
segment 88
segment 89
Jan94 Jul94 Dec94 Jul95 Dec95Jan94 Jul94 Dec94 Jul95 Dec95Jan94 Jul94 Dec94 Jul95 Dec95Jan94 Jul94 Dec94 Jul95 Dec95
„ 7500-
O)
.§ 5000-
O
Q
°~ 2500-
0-
segment 90
segment 91
segment 92
segment 93
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
_ 7500-
_i
^>
o5000'
Q
Q.
2500-
0-
segment 94
~» Model Output
Participate
Detrital Carbon
(mg/L)
Green Bay
sediment segments
Jan94 Jul94 Dec94 Jul95 Dec95
364
-------
Appendix 4.5.3
Calibration Results for PCB28+31 and IPCBs
365
-------
PCB28+31 (ng/L) in segment 1
PCB28+31 (ng/L) in segment 2
PCB28+31 (ng/L) in segments
0.05
0.04-
0.03
0.02
0.01
0
Jan
F
0.05
0.04
0.03
0.02
0.01
0
Jan
0.05
0.04
0.03
0.02
0.01
0
— Model Output
» Cruise Mean
T I
v**kri4^\K^^'i4-vv
94 Jul.94 Dec.94 Jul.95 Dec
•CB28+31 (ng/L) in segment 4
— Model Output
* Cruise Mean
*
94 Jul.94 Dec.94 Jul.95 Del
0.0*
0.04
0.03
0.02 -
0.01 -
0
.95 Jan
F
0.05
0.04
0.03
0.02
0.01
0
=-95 Jan
— Model Output
* Cruise Mean
>^f~f**~*~\***iti^~~~/f^$*r*
.94 Jul.94 Dec.94 Jul.95 De
"CB28+31 (ng/L) in segment 5
— Model Output
» Cruise Mean
§
v^—^ 4J-^_«_J-»V-
94 Jul.94 Dec.94 Jul.95 Dec
PCB28+31 (ng/L) in segment 1 1 PCB28+31 (ng/L) in segment 1 2
— Model Output
• Cruise Mean 0.04 -
0.03 -
0.02
-• 1 1 1 0
— Model Output
* Cruise Mean
^~Ui4_j_X~
0.04
0.03
0.02
0.01
0
:.9S Ja
F
0.05
0.04
0.03
0.02
0.01
0
95 Jan
F
0.05
0.04
0.03
0.02
0.01
n -
— Model Output
* Cruise Mean
,
w-v-i-*-M~*_^-r*«-.
•^wtw^v-
n.94 Jul.94 Dec.94 Jul.95 Dec
•CB28+31 (ng/L) in segment 6
— Model Output
* Cruise Mean
*
— «
94 Jul.94 Dec.94 Jul.95 Dec.9
>CB28+31 (ng/L) in segment 13
— Model Output
« Cruise Mean
O_u__
^T~
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 in dissolved phase (ng/L)
Lake Michigan layers 1, 2, and 3
Upper 30 m water segments
Error bars = standard error
366
-------
PCB28+31 (ng/L) in segment 14
PCB28+31 (ng/L) in segment 15
PCB28+31 (ng/L) In segment 16
0.05
0.04
0.03
0.02
0.01
— Model Output
* Cruise Mean
0.04 -
0.03
0.02
0.01
n •
— Model Output
» Cruise Mean
^_^-t_t~w^_^-»-^V
0.04 •
0.03
0.02
0.01
n
— Model Output
* Cruise Mean
^_^t-2^__^»^^*-^_
•
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan'94 Jul-94 Dec.94 Jul.95 Dec.95
PCB28+31 (ng/L) in segment 20 PCB28+31 (ng/L) in segment 21
PCB28+31 (ng/L) in segment 22
0.05
0.04
0.03
0.02
0.01
— Model Output
» Cruise Mean
o.os
0.04
0.03
0.02
0.01 •
— Model Output
» Cruise Mean
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan-94 Jul-94 Dec'94 Jul'95 Dec-95
PCB28+31 (ng/L) in segment 23
PCB28+31 (ng/L) in segment 24
0.05
0.04
0.03
0.02
0.01
— Model Output
* Cruise Mean
0.04
0.03
0.02
0.01
— Mod el Output
« Cruise Mean
v ***^f~^At^*. j - i *-*V^r
* ^ »
0.05
0.04
0.03
0.02
0.01
— Model Output
» Cruise Mean
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ng/L) in segment 25
0.05
0.04
0.03
0.02
0.01
- Model Output
• Cruise Mean
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
Jan.94 Jul.94 Dec.94 Jul. 95 Dec.95
PCB28+31 in dissolved phase (ng/L)
Lake Michigan layers 1, 2, and 3
Upper 30 m water segments
Error bars = standard error
367
-------
PCB28+31 (ng/L) in segment 29 PCB28+31 (ng/L) In segment 30 PCB28+31 (ng/L) in segment 31 PCB28+31 (ng/L) in segment 32
0.05
0.04
0.03
0.02
0.01
0 -
Jan
F
0.05
0.04
0.03 -
0.02
0.01
0
Jan
f
0.05
0.04
0.03 -
0.02 -
0.01 -
0
— Model Output
» Cruise Mean
^ sSt^J-—^
•* * *{
.94 Jul.94 Dec.94 Jul.95 Dec
CB28+31 (ng/L) in segment 33
— Model Output
» Cruise Mean
* .
.94 Jul.94 Dec.94 Jul.95 De<
CB28+31 (ng/L) In segment 38
— Model Output
* Cruise Mean
TT'Xr
0.05
0.04
0.03
0.02
0.01
0
.95 Jan
F
0.05
0.04
0.03
0.02
0.01 -
0
.95 Jar
F
0.05
0.04
0.03 -
0.02 -
0.01 -
0 -
— Model Output
» Cruise Mean
f I
.94 Jul.94 Dec.94 Jul.95 Dec
•CB28+31 (ng/L) In segment 34
— Model Output
» Cruise Mean
-*^—^~
.94 Jul.94 Dec.94 Jul.95 Dec
'CB28+31 (ng/L) in segment 39
— Model Output
» Cruise Mean
^TT^^^T
0.04 -
0.03
0.02 -
0.01 -
0
.95 Jan
F
0.05 •
0.04
0.03
0.02
0.01
0
.95 Jan
p
0.05-
0.04
0.03
0.02
0.01
0 •
— Model Output
» Cruise Mean
* «
I.94 Jul.94 Dec.94 Jul.95 Dei
'CB28+31 (ng/L) in segment 36
— Model Output
* Cruise Mean
"^^^^r
.94 Jul.94 Dec.94 Jul.95 Dec
CB28+31 (ng/L) in segment 40
— Model Output
« Cruise Mean
» •
0.04
0.03
0.02
0.01
0
:.9S Jar
F
0.05 -
0.04 -
0.03 -
0.02
0.01
0
.95 Jar
F
O.OS
0.04
0.03
0.02
0.01
o -
— Model Output
» Cruise Mean
*
.94 Jul.94 Dec.94 Jul.95 Dec
>CB28+31 (ng/L) in segment 37
— Model Output
* Cruise Mean
**** " '.
.94 Jul.94 Dec.94 Jul.95 Dec
CB28+31 (ng/L) in segment 41
— Model Output
» Cruise Mean
' » * * * «
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 in dissolved phase (ng/L)
Lake Michigan layers 4 and 5
30 m to bottom water segments
Error bars = standard error
368
-------
PCB28+31 (ng/L) in segment?
PCB28+31 (ng/L) in segment 10
0.1
0.08
O.OE
0.04
0.02
— Model Output
* Cruise Mean
0.1
0.08
0.06
0.04
0.02
— Model Output
» Cruise Mean
U.3
0.4
0.3
0.2
0.1
0 -
— Model Output
» Cruise Mean
*
•
1
0.8
0.6
0.4
— Model Output
» Cruise Mean
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.96 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ng/L) in segment 17 PCB28+31 (ng/L) In segment 18 PCB28+31 (ng/L) in segment 19 PCB28+31 (ng/L) in segment 26
— Model Output
« Cruise Mean
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.96 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ng/L) in segment 27 PCB28+31 (ng/L) in segment 28
o.s
PCB28+31 (ng/L) In segment 35
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 in dissolved phase (ng/L)
Green Bay water segments
Error bars = standard error
369
-------
0.4i
5o.3-
O)
c
"0.2-
CO
CM
CD
00.1-
0.
segment 42
""-•x.
^^^^^
"""""-N^^.
U.4-1
0.3-
0.2-
0.1
segment 43
^^^
^^^^^^
"" ^-^^
0.3^
0.2-
0.1-
n.
segment 44
^^^^fc.-
' ' -~^__
u.t-
0.3-
0.2
0.1-
n-
segment 45
^-i^,
— «^_^_^^
~^*~
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 JuJ94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul'95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
„. r\ A n A n A
0.4-
?0.3-
D)
c
"0.2-
co
CM
m
00.1-
Q.
segment 46
0.3-
0.2
0.1-
segment 47
0.3-
0.2
0.1-
n.
segment 48
0.3-
0.2-
0.1-
segment 49
— t
" '
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
U.4'
^°-3'
"0.2
CO
CM
CO
00.1-
n-
segment 50
0.3-
0.2i
0.1-
n-
segment 51
0.3-
0.2-
0.1-
n.
segment 52
•^-^^
^^~^^^
^ — >^_^
^
~~-~.
0.3-
0.2-
0.1-
n-
segment 53
>
""""
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
I0-3'
"0.2-
CO
CM
cn
00.1-
0-
segment 54
^*"*»*»fci^
^^^
0.3-
0.2-
0.1-
segment 55
^ ^_
0.3-
0.2-
0.1-
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jar
segment 56
i94 Jul94 DecS4 Jul95 Dec
U.4-
0.3-
0.2-
0.1-
95 Jar
segment 57
i94 Jul94 Dec94 Jul95 Dec.
0.4
D)
0.3-
"0.2-]
CO
fM
CO
segment 58
0.4
0.3-
0.2-
0.1-
segment 59
— . Model Output
dissolved
PCB 28 + 31 (ng/L)
Lake Michigan
sediment segments
0.4
0.3-
0.2-
0.1-
°"J ' i • 1 0-J 1 i . 1 Q| : | Q| _ _ ,
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
segment 60
0.4
0.3-
0.2-
0.1-
segment 61
370
-------
ux-
i°-3'
So.2-
00
CM
CD
00.1-
CL
0-
Jar
0.4-
=d 0.3-
c
?0.2-
00
CM
£fl
00.1-
0-
Jar
0.4-
^°-3-
"0.2-
co
CM
m
00.1-
0-
Ja
0.4-
en
c
"0.2-
co
CM
m
00.1-
0-
Jar
segment 62
i94 Jul94 Dec94 Jul95 Dec
segment 66
i94 Jul94 Dec94 Jul95 Dec
segment 70
i94 Jul94 Dec94 Jul95 Dec
segment 74
i94 Jul94 Dec94 Jul95 Dec
0.3-
0.2-
0.1 J
0-
95 Ja
0.4-1
0.3-
0.2-
0.1-
O-l
95 Jar
0.4-I
0.3
0.2-
0.1-
o-l
95 Jar
0.4-
0.3-
0.2-
0.1-
O-l
95 Jar
segment 63
i94 Jul94 Dec94 Jul95 Dec
segment 67
i94 Jul94 Dec94 Jul95 Dec
segment 71
194 Jul94 Dec94 Jul95 Dec
segment 75
194 Jul94 Dec94 Ju
I95 Dec
0.3-
0.2-
0.1-
0-
95 Jar
0.4-
0.3-
0.2-
0.1-
oJ
95 Jar
0.4n
0.3-
0.2-
0.1-
o-l
95 Jar
0.4-
0.3-
0.2-
0.1-
o-l
95 Jar
segment 64
i94 Jul94 Dec94 Jul95 Dec
segment 68
i94 Jul94 Dec94 Jul95 Dec
segment 72
i94 Jul94 Dec94 Jul95 Dec
segment 76
194 Jul94 Dec94 Jul95 Dec
•x Model Output
dissolved
0.3-
0.2-
0.1-
O-l
95 Jar
0.4-
0.3-
0.2-
0.1-
O-l
95 Jar
0.4-
0.3-
0.2-
0.1-
o-l
95 Jar
0.4-1
0.3-
0.2-
0.1-
o-l
95 Jar
segment 65
i94 Jul94 Dec94 Jul95 DecJ
segment 69
i94 Jul94 Dec94 Jul95 Dec?
segment 73
)94 Jul94 Dec94 Jul95 Dec£
segment 77
i94 Jul94 Dec94 Jul95 Dec<
PCB 28 + 31 (ng/L)
Lake Michigan
sediment segments
371
-------
0.4-]
I0'3'
«0.2-
CO
CM
03
Q-
segment 7Q
U.4-
0.3-
0.2
0.1-
segment 79
U.H-
0.3-
0.2-
0.1-
n.
segment 80
U.T
0.3-
0.2-
0.1-
n.
segment 81
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 JuISS Dec£
0.4-
i>3-
«0.2-
co
CM
m
oo.1
segment 82
0.3-
0.2
0.1
segment 83
0.3-
0.2-
0.1-
n.
segment 84
0.3-
0.2-
0.1
n.
segment 85
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
r\ A r\ A r\ A r\ A
?0.3-
O>
C
"0.2-
co
CM
m
00.1-
0-
segment 86
0.3-
0.2
U.1-
r\.
segment 87
0.3-
0.2-
0.1-
segment 88
0.3-
0.2-
0.1-
segment 89
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul'95 Dec95 Jan94 Jul'94 Dec94 Jul95 Dec9
|°-3-
« 0.2-
co
CM
CO
00.1-
0-
Ja
0.4-
I0-3
« 0.2
CO
CM
CD
00.1
0-
segment 90
0.3-
0.2-
0.1-
n.
segment 91
^^^_
0.3-
0.2-
0.1-
n
segment 92
0.3-
0.2-
0.1-
segment 93
i94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 JuJ94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9
segment 94
-v Model Output
dissolved
PCB 28 + 31 (ng/L)
Green Bay
sediment segments
Jan94 Jul94 Dec94 Jul95 Dec95
372
-------
PCB28+31 (ng/L) in segment 1
PCB28+31 (ng/L) in segment 2
PCB28+31 (ng/L) in segment 3
0.04
0.03
0.02
0.01
0
Jan
0.05
0.04
0.03
0.02 •
0.01
0
Jan
0.05
0.04
0.03
0.02
0.01
0
— Model Output
» Cruise Mean
^M^>^>-
94 Jul.94 Dec.94 Jul.95 Dec
'CB28+31 (ng/L) in segment 4
— Model Output
» Cruise Mean
^-L*~^L*~^-^
0.04
0.03
0.02
0.01
0
.95 Jan
0.05
0.04
— Model Output
» Cruise Mean
Jj-,^^^
.94 Jul.94 Dec.94 Jul.95 Dec
PCB28+31 (ng/L) in segment 5
— Model Output
» Cruise Mean
0.03 -
0.02 -
0.01
0.04 -
O.OJ -
0.02
0.01
0
95 Jar
0.05 -I
0.04
0.03
0.02
0.01 -
" 0
— Model Output
* Cruise Mean
^t-^1^—
.94 Jjl.94 Dec.94 Jul.95 Dec
PCB28+31 (ng/L) in segment 6
— Model Output
» Cruise Mean
-J^— ^_— -r-r-
0 -I r-*-^* r '- r-*-» 1
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ng/L) in segment 11
— Model Output
» Cruise Mean
"-J*H-^-r^
F
0.05
0.04
0.03
0.02
0.01
0 H
•CB28+31 (ng/L) in segment 12
— Model Output
* Cruise Mean
* " "
^- — ^%V' '"IT"""-
F
0.05
0.04
0.03
0.02
0.01 -
0
'CB28+31 (ng/L) in segment 13
— Model Output
» Cruise Mean
<_Ur>-~-r>-
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.96
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+3i in particulate phase (ng/L)
Lake Michigan layers 1, 2, and 3
Upper 30 m water segments
Error bars = standard error
373
-------
PCB28+31 (ng/L) in segment 15
PCB28+31 (ng/L) in segment 16
.OS --
.04 -
.03
.02
.01
— Model Output
» Cruise Mean
— Model Output
0.04 - » Cruise Mean 0.04 •
0.03 - 0.03
0.02 - 0.02
0.01
J°an.94 MM Dec.94 J.I.H Dec.95 Ja"-M Jul-94 DeC-94 Jul'9S DeC'9i
f
0.05
0.04 -
0.03
0.02
0.01
0
Jar
O.OS
0.04
0.03
0.02
0.01
0
'CB28+31 (ng/L) in segment 20
— Model Output
» Cruise Mean
•^^^^^
1.94 Jul.94 Dec.94 Jul.95 De
PCB28+31 (ng/L) in segment 23
— Model Output
» Cruise Mean
"— i>T>-^» — rr-
F
0.05
0.04
0.03 -
0.02-
0.01
0
c.95 Jan
1
0.05
0.04
0.03
0.02
0.01
3 0
'CB28+31 (ng/L) in segment 21
— Model Output
« Cruise Mean
.94 Jul.94 Dec.94 Jul.95 Dec
'CB28+31 (ng/L) in segment 24
— Model Output
* Cruise Mean
^__t-« •**_»—• •»-*•**
1— * * 1 1—
0.01
0-
Jan
0.0
0.0
— Model Output
* Cruise Mean
• **7^ •-^T^—
94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ng/L) in segment 22
— Model Output
4 - » Cruise Mean
0.03-
0.02
0.0
.95 J
O.OE
0.04
0.03
0.02
0.01
0
1 -
an.94 Jul.94 Dec.94 Jul.95 Dec
PCB28+31 (ng/L) in segment 25
— Model Output
* Cruise Mean
^__»j. — .. _^ — K-_
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 in particulate phase (ng/L)
Lake Michigan layers 1, 2, and 3
Upper 30 m water segments
Error bars = standard error
374
-------
0.04
0.03
0.02
0.01
— Model Output
» Cruise Mean
\
•kii*—*^! — ,-JJ
0.04
0.03
0.02 •
0.01
— Model Output
* Cruise Mean
I
I
I ^ f ,-f
0.04
0.03 -
0.02 -
0.01 -
— Model Output
» Cruise Mean
* . .[-f-^ ~_
0.04
0.03
0.02-
0.01 -
— Model Output
» Cruise Mean
^__t-t— 4-— _•— — ri—-
O-l 1 1 < • — 1 0-1 ^ 1 1 1 0-1 r-=— 1 , : 1 0-1 r-2 , , : 1
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ng/L) in segment 33 PCB28+31 (ng/L) in segment 34 PCB28+31 (ng/L) in segment 36 PCB28+31 (ng/L) in segment 37
0.04
0.03
0.02
0.01 -
— Model Output
* Cruise Mean
^ -- -*-r*-
0.04
0.03
0.02
0.01
— Model Output
» Cruise Mean
±_iL_ , _ ^___^_
0.04
0.03
0.02 i
0.01 -
— Model Output
» Cruise Mean
t » i
0.04
0.03
0.02
0.01
— Model Output
» Cruise Mean
. -*.n _-!-*-
o ._ ^* , — -.; — , — _: — . o r-* , ^— i — - — 0-- r^—* — , : 1 0 r^-^ — , : ,
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ng/L) in segment 38 PCB28+31 (ng/L) in segment 39 PCB28+31 (ng/L) in segment 40 PCB28+31 (ng/L) in segment 41
0.04-
0.03 -
0.02 -
0.01 -
0 -
— Model Output
» Cruise Mean
0.04
0.03
0.02
0.01 -
0 -I
— Model Output
* Cruise Mean
*-' i 7 > -r-i —
0.04
0.03 •
0.02
0.01
0 -I
— Model Output
» Cruise Mean
0.04 •
0.03
0.02 -
0.01 •
0 -t
— Model Output
» Cruise Mean
1 *-»T ' * T1 —
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 in particulate phase (ng/L)
Lake Michigan layers 4 and 5
30 m to bottom water segments
Error bars = standard error
375
-------
PCB28+31 (ng/L) in segment 7
PCB28+31 (ng/L) in segments
PCB28+31 (ng/L) in segments
PCB28+31 (ng/L) in segment 10
0.1
0.08
0.06
0.04
0.02
-Model Output
> Cruise Mean
0.1
0.08
0.06
0.04
0.02
— Model Output
* Cruise Mean
0.5
0.4
0.3
0.2
— Model Output
* Cruise Mean
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ng/L) in segment 17
PCB28+31 (ng/L) in segment 18
PCB28+31 (ng/L) in segment 19
PCB28+31 (ng/L) in segment 26
— Model Output
» Cruise Mean
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 jan.M Ju| 94 Dec94 Ju| 95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ng/L) in segment 27
PCB28+31 (ng/L) In segment 28
PCB28+31 (ng/L) in segment 35
— Model Output
* Cruise Mean
0.1
0.08
0.06
0.04 •
0.02
— Model Output
» Cruise Mean
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 in particulate phase (ng/L)
Green Bay water segments
Error bars = standard error
376
-------
PCB28+31 (ug/L) in segment 42
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 43
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (fig/L) in segment 44
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 45
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 46
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 47
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 48
PCB28+31 (ug/L) in segment 49
PCB28+31 (ug/L) in segment 50
— Model Output
4-1
3
2
1
— Model Output
" —I
3
2 -
1 -
0 -
— Model Output
—
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.96 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
in particulate phase (pg/L)
Lake Michigan sediment segments
377
-------
PCB28+31 (ug/L) in segment 51
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 52
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment S3
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 54
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 55
— Model Output
PCB28+31 (ug/L) in segment 56
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 57
3
2
1
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) In segment 60
3
2
1
n •
— Model Output
PCB28+31 (ug/L) in segment 58
PCB28+31 (ug/L) in segment 59
— Model Output
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan-94 Jul-94 Dec'94 Jul'95 Dec'95
PCB28+31 (u.g/L) In segment 61
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
Jan.94 Jul.94 Dec.94 Jul.96 Dec.95
PCB28+31 in particulate phase (pg/L)
Lake Michigan sediment segments
378
-------
PCB28+31 (ug/L) in segment 62
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 63
-Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 64
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 65
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 66
PCB28+31 (ug/L) in segment 67
— Model Output
3
2
1 •
0 -I
— Model Output
I r- • 1 1 1
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 68
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 69
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 70
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 in particulate phase (M9/L)
Lake Michigan sediment segments
379
-------
PCB28+31 ((ig/L) in segment 71 PCB28+31 (jig/L) in segment 72 PCB28+31 (ug/L) in segment 73
4-1
3
2
1
0
— Model Output
1
3
2
1
0 -
— Model Output
3
2
1
— Model Output
Jan.94 Jul.94 Dec.94 Jul.9S Dec.95 Jan.94 Jul.94 Dec.94 Jul.96 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 74
PCB28+31 (ng/L) in segment 75
PCB28+31 (fig'L) in segment 76
— Model Output
4
3
2
1
n .
— Model Output
3
2
1
0 •
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 77
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 in participate phase (M9/L)
Lake Michigan sediment segments
380
-------
PCB28+31 (ug/L) in segment 78
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 79
- Model Output
PCB28+31 (ug/L) in segment 80
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 81
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 82
— Model Output
PCB28+31 (ug/L) in segment 83
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 84
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 85
2 •
— Model Output
PCB28+31 (ug/L) in segment 86
1 -
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+3i in particulate phase (pg/L)
Green Bay sediment segments
381
-------
PCB28+31 (ug/L) In segment 87
2
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.96
PCB28+31 ((ig/L) in segment 88
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 89
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 90
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 91
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 92
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 93
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (pg/L) in segment 94
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 in participate phase (|jg/L)
Green Bay sediment segments
382
-------
U.LD'
5" 0.04
r 0.03-
to
8 0.02-
03
CL 0.01-
Q.
segment
Jan94 Jul94 Dec94 Jul95
005
j 0.04.
X 0.03.
CO
a 0.02.
CO
CL 0.01-
o
1
h
^J^
Dec95 Jan94 Jul94
segment 5
4
Jan94
0 05
j 0.04-
^ 0.03-
co
8 0.02-
m
CL 0.01-
0-
Ja
5 0.04-
~ 0.03-
co
8 0.02-
CD
CL 0.01-
Jul94 Dec94 Jul95
segment 2
r
•^^^^^f^l
segment 3
segment 4
T
Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
segment 6
Dec95Jan94 Jul94
segment 13
j
**
194
•^
1
Jan94
J 0.04-
ra
^ o.o3-
CO
+
8 0.02-
CQ
O
CL 0.01-
n-
Jul94 Dec94 Jul95
segment
2
'*
Jul94 Dec94 Jul95
segment 11
V>^T""*r"ASfc
-------
0.05-
j 0.04-
^ 0.03-
(0
8 0.02-
CO
CL 0.01-
0-
Jar
0.05 -I
j 0.04-
^ 0.03-
8 0.02-
CD
CL O.OL
0-
Jar
0.05 -,
5" 0.04-
Dl
~ 0.03-
3 0.02.
CO
a! 0.01-
0-
segment 29
, I
SfM-rf
•
.
segment 30
t
UVV"~IT*
segment 31
^rr^^r^
segment 32
v_l<-K^t>
i94 Jul94 Dec94 Jul95 Dec95 Jan94 JuJ94 Dec94 Jul95 Dec95 Jan94 JuJ94 Dec94 Jul95 Dec95 Jan94 JuJ94 Dec94 Jui95 Dec!
segment 33
^?-^r~
segment 34
*• »
segment 36
A~_^--''~X\
*»« * , *^
segment 37
~^-^~~*\
*H . *{
i94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 DecS
segment 38
^r>^s-<^
segment 41
** « *»
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
384
-------
0.10
0.08
segment 7
^ 0.06 ]
? I
3 0.04-1
£ 0-02 W^HKfc*'-^
segment 9
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
0.10-,
0.08 J
segment 26
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
0.10,
5 0.08-
"a
" 0.06.
n
S 0.04-I
-------
PCB28+31 (ug/L) In segment 42
— Model Output
Jsn.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (pg/L) In segment 43
PCB28+31 (ng/L) in segment 44
PCB28+31 (|ig/L)ln segment 45
— Model Output
~— —
3
2
1
0
— Model Output
"^- —
3
2
1 •
0
— Model Output
~~ —
PCB28+31 (u.g/L) in segment 46
PCB28+31 (ug/LJ in segment 47
PCB28+31 (u.g/L) In segment 48
PCB28+31 (u,g/L) in segment 49
— Model Output
4 •
3 -
2
0
— Model Output
3
2
0
— Model Output
3
2
0
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jui.fl4 Oec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (jig/L) in segment 50
PCB28+31 (iig/L) in segment 51
PCB28+31 (ng/L) in segment 52
PCB28+31 (|ig/L) in segment 53
— Model Output
.94 Jul.94 Doc. 9 4 Jut. 95 De
3
2 -
1 -
0 -
.95 Jar
— Model Output
.94 Jul.94 Dec.94 Jul.95 Dec
3
2
1
0
.95 Jar
— Model Output
~^~-
.94 Jul.94 Dec ,94 Jul.95 Dec
3
2
1
0
.95 Jan
— Model Output
.94 Jul.94 Dec.94 Jul.BS Dec
PCB28+31 (|tg/L) In segment 54
PCB28+31 (ug/L) in segment 55
PCB28+31 (|ig/L) in segment 56
PCB28+31 (|ig/L) in segment 57
— Model Output
-~
3
2
1 -
— Model Output
— Model Output
3 -
2
1 -
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ttS/L) in segment 58
PCB28+31 |ug/l_) in segment 59
PCB28+31 (u.g/L) in segment 60
— Model Output
4 •
3
2
1
— Model Output
' ' • \ 0 -1 — r- , T , 1
.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.BS De<
4 -I
3
2 -
1
0 -I
.95 Jan
— Model Output
.84 Jul.94 Dec.94 Jul,95 Dec
PCB28+31 (|ig/L) In segment 61
-Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Osc.95
PCB28+3i (dissolved + particulate) (ug/L)
Lake Michigan sediment segments
386
-------
PCB28+31 (fig/L) in segment 62 PCB28+31 (ug/L) in segment 63
PCB28+31 (ug/L) in segment 64
— Model Output
4
3
2
1
0 -I
— Model Output
, , , *
3
2
1
0 -I
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.96 Dec.95 Jan.94 Jjl.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 65
PCB28+31 (jig/L) in segment 66
PCB28+31 (ug/L) In segment 67
— Model Output
3
2
1
— Model Output
4 i
3 J
2 -
1
• i • i 0 j
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 ((ig/L) in segment 68
PCB28+31 (ug/L) in segment 69
PCB28+31 (uglL) in segment 70
— Model Output
4 -]
3
2
1
0 J
— Model Output
3
2
1
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (dissolved + particulate) (|jg/L)
Lake Michigan sediment segments
387
-------
PCB28+31
-------
PCB28+31 (|ig/L) in segment 78
PCB28+31 (ug/L) in segment 79
PCB28+31 (ug/L) in segment 80
— Model Output
3
2
1
n •
— Model Output
3
2
1
n
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 81
PCB28+31 (ug/L) in segment 82
PCB28+31 (|ig/L) in segment 83
— Model Output
— Model Output
3
2 -
1 -
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) in segment 84
PCB28+31 (|ig/L) in segment 85
PCB28+31 (ug/L) in segment 86
3
— Model Output
3
2
1
— Model Output
3
2
1
0 •
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
8+3i (dissolved + particulate) (|jg/L)
Green Bay sediment segments
389
-------
PCB28+31 (ng/U) in segment 87
2
— Model Output
PCB28+31 (ng/U) in segment 88
— Model Output
PCB28+31 (ug/L) in segment 89
— Model Output
Jan.94 Jul.94 Dec.94 Jul.9B Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.96 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ng/L) in segment 90
— Model Output
PCB28+31 (jig/L) in segment 91
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (ug/L) In segment 92
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB28+31 (|ig/L) in segment 93
PCB28+31 (ng'L) in segment 94
— Model Output
3
2
1
n
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
PCB2B+31 (dissolved + participate) (|jg/L)
Green Bay sediment segments
390
-------
ZPCB (ng/L) in segment 1
ZPCB (ng/L) in segment 2
ZPCB (ng/L) in segment 3
0.2
— Model Output
* Cruise Mean
— Model Output
» Cruise Mean
0.4
— Model Output
« Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-96 Dec-95
EPCB (ng/L) in segment 4
ZPCB (ng/L) in segment 5
ZPCB (ng/L) in segment 6
— Model Output
» Cruise Mean
— Model Output
» Cruise Mean
0.3
— Model Output
» Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 11
ZPCB (ng/L) in segment 12
ZPCB (ng/L) in segment 13
0.3
0.2
0 1 -
D -J
— Model Output
» Cruise Mean
*
^ *
*\— "'^•s^ *
^"~ ' * — »^^^~^«j*^™ ~w ""^
* *
0.3
0.2
0 1
n •
— Model Output
» Cruise Mean
. » *
y _ * *
^^ 11^ ' ^-> — ~~ , -JT/1
0.3
0.2
01 -
a -
— Model Output
» Cruise Mean
»
L ^ *
^ ^ — A ^ - - t { ^
*
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
IPCBs in dissolved phase (ng/L)
Lake Michigan layers 1, 2, and 3
Upper 30 m water segments
Error bars = standard error
391
-------
ZPCB (ng/L) in segment 14
0.2
— Model Output
• Cruise Mean
ZPCB (ng/L) in segment 15
0.4
0.2
0.1
— Model Output
» Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 16
0.1
— Model Output
» Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 20
0.4
— Model Output
« Cruise Mean
ZPCB (ng/L) in segment 21
0.1
— Model Output
» Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 22
0.4
— Model Output
» Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 23
0.4
0.2
— Model Output
» Cruise Mean
ZPCB (ng/L) in segment 24
— Model Output
» Cruise Mean
ZPCB (ng/L) in segment 25
— Model Output
» Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
IPCBs in dissolved phase (ng/L)
Lake Michigan layers 1, 2, and 3
Upper 30 m water segments
Error bars = standard error
392
-------
ZPCB (ng/L) in segment 29
ZPCB (ng/L) in segment 30
ZPCB (ng/L) in segment 31
ZPCB (ng/L) in segment 32
0.4
0.3
0.2
0.1
— Model Output
* Cruise Mean
,
1 1 *
0.4 -i
0.3
0.2
0.1 •
— Model Output
« Cruise Mean
-r^^ .
0.3
0.2
0.1 V
— Model Output
» Cruise Mean
^
» ^
^ *" ""*>•>•* — ^~ %^v
0.3
0.2
0.1
— Model Output
« Cruise Mean
4 *
•- — ' "^V^- — -"v^
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95Jan.94 Ju|_94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 33 _ ZPCB <"9'L) in segment 34 ^^ ^^ ^ ^ «,nmonf 1K ZPCB ("3"-) in segment 37
0.4
(M
07-
ni -
— Model Output
• Cruise Mean
»
_^-**» *
— /* V—v^. — — 'V-'
»* « »
Jan-94 Jul-94 Dec-94 Jul-95 Dec
0.3
0.2
0.1 •
0 •
— Model Output
» Cruise Mean
»
-j^J**, *
"" —\^ \r~~.r** — ^-^^V<«.
0.4
0.3
0.2
0.1
95Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 °
Ja
— Model Output
» Cruise Mean
»
>^^V^-. -^^~~v_
^ »
1-94 Jul-94 Dec-94 Jul-95 De
0.3 -
0.2 -
0.1 -
-95Ja«
ZPCB (ng/L) in segment 38 ZPCB (ng/L) in segment 39 ZPCB (ng/L) in segment 40 04
0.4
0.3
0.2
0.1
0
— Model Output
» Cruise Mean
^
^r^~^^
0.2
0.1
0
— Model Output
» Cruise Mean
»
^ » * »
0.4
0.3
0.2
0.1
n
— Model Output
* Cruise Mean
«
. . » *
0.3
0.2
0.1
0
— Model Output
» Cruise Mean
s^*^~. ^—^
/• ^V^ _-tr^ *^^
-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 41
— Model Output
» Cruise Mean
*
,_. r » v^.x. ,,,.'*"-^^^v
^ *
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95Jan-94 Jul'94 Dec-94 Jul-95 Dec-95 Jan_94 Ju|-94 Dec-94 Jul-95 Dec-95
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
IPCBs in dissolved phase (ng/L)
Lake Michigan layers 4 and 5
30 m to bottom water segments
Error bars = standard error
393
-------
EPCB (ng/L) in segment?
EPCB (ng/L) in segment 8
EPCB (ng/L) in segments
EPCB (ng/L) in segment 10
0.4
— Model Output
» Cruise Mean
0.4
0.2
— Model Output
» Cruise Mean
3.2
2.4
0.8
- Model Output
> Cruise Mean
3.2
2.4
1.6
— Model Output
• Cruise Mean
Jan-94 JUI94 Dec-94 Jul-95 Dec-95 Ja"-94 Jul'94 Dec-94 Ju|-95 Dec-96 Jan-94 Jul-94 Dec-94 Jul-95 Dec-96 Jan-94 Jul-94 Dec-94 Jul-95 Dee-95
EPCB (ng/L) in segment 17 EPCB '" segment 18
EPCB (ng/L) In segment 19 £PCB ("a'1-)in segment 26
0.8
— Model Output
« Cruise Mean
— Model Output
» Cruise Mean
0.2
0.2
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
EPCB (ng/L) in segment 27
4 Jul-94 Dec-94 Jul-95 Dec-95
EPCB (ng/L) in segment 28
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
EPCB (ng/L) in segment 35
0.8 -
— Model Output
« Cruise Mean
— Model Output
* Cruise Mean
0.8
0.6
— Model Output
» Cruise Mean
WWi»+**JW*^
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
IPCBs in dissolved phase (ng/L)
Green Bay water segments
Error bars = standard error
— Model Output
» Cruise Mean
Jul'94
Ju''9i Dec.95
394
-------
EPCB (ng/L) In segment 42
— Model Output
EPCB (ng/L) In segment 43
1.8
1.2 -
0.8
- Model Output
EPCB (ng/L) in segment 44
— Model Output
IPCB (ng/L) in segment 45
— Model Output
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jut-94 Dcc-94 Jul-95 Dec-95 Jan-B4 Jul-94 Dec-94 Jul-05 Dec-95 Jan-94 Jul-94 Dcc-94 Jul-95 Dec-B5
EPCB (ng/L) in segment 46
— Model Output
Jan-94 Jul-64 Dec-94 Jul-95 D«c-95
IPCB (ng/L) in segment 47
IPCB (ng/L) in segment 48
ZPCB (ng/L) in segment 49
.4
.8
.2
.6
0 -
— Model Output
3 -I
2.4
1.8
1.2
0.8
0 -i
— Model Output
3 1
2.4 •
1.8
1.2
0.6 •
0
— Model Output
— —
EPCB (ng/L) in segment 50
EPCB (ng/L) (n segment 51
2.4
1.8
1.2
— Model Output
2.4 •
1.8 •
1.2 -
— Model Output
EPCB (ng/L) in segment 52
— Model Output
•95 Jan-94 Jul-94 De
EPCB (ng/L) in segment 53
— Model Output
EPCB (ng/L) in segment 54
EPCB (ng/L) in segment 55
EPCB (ng/L) in segment 58
EPCB (ng/L) in segment 57
2.4
1.8
1.2
D.6
— Model Output
^_^^
2.4
1.8
1.2
0.6
— Model Output
^^— — — __
2.4
1.8
1.2
0.6 •
— Model Output
2.4
i.a
1.2
0.8
— Model Oulput
EPCB (ng/L) in segment 58
EPCB (ng/L) in segment 59
EPCB (ng/L) in segment 60
EPCB (ng/L) In segment 61
— Model Output
2.4
1.8
1.2
0.6
— Model Output
2.4
1.8
1.2
0.6
— Model Output
2.4
1.8 -
1.2
0.6
— Model Output
SPCBs in dissolved phase (ng/L)
Lake Michigan sediment segments
395
-------
ZPCB (ng/L) in segment 64
IPCB (ng/L) in segment 65
3
2.4
1.8
1.2
0.6
0
Ja
3
2.4
1.8
1.2
0.6
O-l
Jan
3
2.4
1.8
1.2
0.6
— Model Output
1-94 Jul-94 Dec-94 Jul-95 De
ZPCB (ng/L) in segment 66
— Model Output
-94 Jul-94 Dec-94 Jul-95 Dec
ZPCB (ng/L) in segment 70
— Model Output
0 I 1 , , 1
Jan-94 Jul-94 Dec-94 Jul-96 Dec
3 •
2.4 •
1.8
1.2
0.6
0 -I
ZPCB (ng/L) in segment 74
— Model Output
1 1 1- 1
3
2.4
1.8
1.2
0.6
0
c-99an
3
2.4
1.8
1.2
0.6
0
-95 Ja
3
2.4-
1.8
1.2
0.6
0 -I
.95 Jar
3 -
2.4-
1.8
1.2-
0.6
0-
— Model Output
3
2.4-
1.8
1.2 -
0.6
— Model Output
T_ , , 1 o 4- 1 1 1
-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec
ZPCB (ng/L) in segment 67
— Model Output
3 •
2.4
1.8
1.2
0.6
ZPCB (ng/L) in segment 68
— Model Output
•1 1 , 1 1 OH 1 1 1 1
n-94 Jul-94 Dec-94 Jul-95 Dec-96 Jan-94 Jul-94 Dec-94 Jul-95 Dec
EPCB (ng/L) in segment 71
— Model Output
3 -
2.4-
1.8
1.2 -
0.6 -
i i 0
-94 Jul-94 Dec-94 Jul-95 Dec-95 Jar
ZPCB (ng/L) in segment 75
— Model Output
3
2.4
1.8
1.2
0.6
0
3
2.4
1.8
1.2
0.6
0
,-95 •>*
3
2.4
1.8
1.2
0.6
0
95 J.
— Model Output
1-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 69
— Model Output
in-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 72 ZPCB ("9"-) '" segment 73
3 ., ,
— Model Output
-94 Jul-94 Dec-94 Jul-95 Dec
ZPCB (ng/L) in segment 76
— Model Output
2,4
1.8
1.2
0.6
— Model Output
0 • 1 1 1 1
.95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
3 -
2.4
1.8 -
1.2
0.6
0 -I-
ZPCB (ng/L) in segment 77
— Model Output
1 1 1 '
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 J'"'M JU'"94 DeC"94 J"''95 DeC'95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCBs in dissolved phase (ng/L)
Lake Michigan sediment segments
396
-------
IPCB (ng/L) in segment 78
IPCB (ng/L) In segment 79
IPCB (ng/L) In segment SO
2.4
1.8
1.2
U.b
— Model Output
2.4
1.8
1.2
0.6
n -
— Model Output
2.4-
1.8
1.2
0.6
— Model Output
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 81
IPCB (ng/L) in segment 82
IPCB (ng/L) in segment 83
2.4
1.8
1.2
— Model Output
2.4
1.8
1.2
D
— Model Output
2.4
1.8
1.2
OR
— Model Output
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
IPCB (ng/L) in segment 84
IPCB (ng/L) in segment 85
IPCB (ng/L) in segment 86
2.4
1.8
1.2-
0.6-
n -
— Model Output
2.4
1.8
1.2
OB
o -
— Model Output
2.4
1.8
1.2
o.e
n .
— Model Output
Jan-94 Jul-94 Dec-94 Jul-96 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCBs in dissolved phase (ng/L)
Green Bay sediment segments
397
-------
ZPCB (ng/L) in segment 87
ZPCB (ng/L) in segment 88
ZPCB (ng/L) in segment 89
3
2.4
1.8
1.2
0.8
— Model Output
3
2.4
1.8
1.2
0.6
— Model Output
2.4
1.8
1.2
0.6
— Model Output
Jan-94 Jul-94 Dec-94 Jul-95 Dec-96 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 90
2.4
1.8
1.2
0.6
— Model Output
ZPCB (ng/L) in segment 91
2.4
1.8
1.2
0.6
— Model Output
ZPCB (ng/L) In segment 92
2.4 •
1.8
1.2
0.6
n -
— Model Output
Jan-94 Jul-94 Dec-94 Jul-95 Dec-96 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 93
2.4
1.8
1.2
0.6
0
— Model Output
ZPCB (ng/L) in segment 94
2.4 -
1.8 -
1.2
0.6
n
— Model Output
Jan-94 Jul-94 Dec-94 Jul-96 Dec-96 Jan-94 Jul-94 Dec-94 Jul-9S Dec-95
ZPCBs in dissolved phase (ng/L)
Green Bay sediment segments
398
-------
IPCB (ng/L) in segment 1
— Model Output
» Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) In segment 2
0.3-
0.2 •
— Model Output
» Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 3
0.3
— Model Output
» Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
IPCB (ng/L) in segment 4
0.1
— Model Output
* Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-95 Dec-96
IPCB (ng/L) in segment 5
0.2 •
— Model Output
» Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 6
— Model Output
» Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 11
— Model Output
» Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 12
0.1
— Model Output
» Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 13
— Model Output
• Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
IPCBs in particulate phase (ng/L)
Lake Michigan layers 1, 2, and 3
Upper 30 m water segments
Error bars = standard error
399
-------
ZPCB (ng/L) in segment 14
ZPCB (ng/L) in segment 15
ZPCB (ng/L) in segment 16
0.3-
— Model Output
» Cruise Mean
0.2
— Model Output
» Cruise Mean
0.2
— Model Output
* Cruise Mean
• * » »
Jan-94 Jul-94 Dec-94 Jul-9S Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
IPCB (ng/L) in segment 20
ZPCB (ng/L) in segment 21
ZPCB (ng/L) in segment 22
0.4
— Model Output
» Cruise Mean
0.1
— Model Output
» Cruise Mean
— Model Output
» Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 23
ZPCB (ng/L) in segment 24
ZPCB (ng/L) in segment 25
0.3
0.2
— Model Output
* Cruise Mean
— Model Output
» Cruise Mean
0.2
0.1
— Model Output
» Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-96 Dec-95 Jan-94 Jul-94 Dec-94 Jul-96 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
IPCBs in particulate phase (ng/L)
Lake Michigan layers 1, 2, and 3
Upper 30 m water segments
Error bars = standard error
400
-------
IPCB (ng/L) in segment 29
ZPCB (ng/L) in segment 30
ZPCB (ng/L) in segment 31
IPCB (ng/L) in segment 32
0.3
0.2
0.1
0
Ja
0.4
0.3
0.2
0.1
0
Jan
0.4
0.3
0.2
0.1
0
— Model Output
* Cruise Mean
*
i^]U>>-^-*--*A>_Jk_^J«J
1-94 Jul-94 Dec-94 Jul-95 De
ZPCB (ng/L) in segment 33
— Model Output
* Cruise Mean
^T-^-^
-94 Jul-94 Dec-94 Jul-95 Dec
ZPCB (ng/L) in segment 38
— Model Output
* Cruise Mean
<— *-*-*-. ^^—rr~
0.3
0.2
0.1
0
c-95 Ja
0.4
0.3
0.2
0.1
0
-95 Jar
0.4
0.3
0.2
0.1 •
n
— Model Output
* Cruise Mean
1-94 Jul-94 Dec-94 Jul-95 De
ZPCB (ng/L) in segment 34
— Model Output
* Cruise Mean
* »
1-94 Jul-94 Dec-94 Jul-95 Del
ZPCB (ng/L) in segment 39
— Model Output
» Cruise Mean
*-»-5~7~- . -r^
0.3
0.2
0.1
0-
c-95 Jar
0.4
0.3 •
0.2
0.1
0
:-95 Jan
0.4
0.3
0.2
0.1 -
n -
— Model Output
* Cruise Mean
^^^-r-
-94 Jul-94 Dec-94 Jul-95 Dec
ZPCB (ng/L) in segment 36
— Model Output
» Cruise Mean
• t . *
-94 Jul-94 Dec-94 Jul-95 Dec
ZPCB (ng/L) in segment 40
— Model Output
* Cruise Mean
* » * *
0.4
0.3
0.2
0.1
0
-95 Jar
0.4
0.3
0.2
0.1
0
.95 Jan
0.4
0.3
0.2
0.1 -
0 -
— Model Output
» Cruise Mean
•
-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) In segment 37
— Model Output
* Cruise Mean
^^^-
-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (ng/L) in segment 41
— Model Output
* Cruise Mean
*~* » ^ *- «—-
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
IPCBs in particulate phase (ng/L)
Lake Michigan layers 4 and 5
30 m to bottom water segments
Error bars = standard error
401
-------
EPCB (ng/L) in segment 7
ZPCB (ng/L) in segment 8
IPCB (ng/L) in segment 9
IPCB (ng/L) in segment 10
0.4
— Model Output
» Cruise Mean
0.8
0.6
— Model Output
* Cruise Mean
2.4
— Model Output
» Cruise Mean
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan.g4 Ju|_94 Dec-94 Jul-95 Dec-95
IPCB (ng/L) in segment 17
— Model Output
« Cruise Mean
ZPCB (ng/L) in segment 18 ZPCB ("3"-) in segment 19
0.8
0
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95Jan-94 Jul'94 Dec-94 Jul-95 Dec-95
IPCB (ng/L) in segment 26
2.4
— Model Output
« Cruise Mean
0.8
0.6
0.4
0.2-
— Model Output
« Cruise Mean
isn (M ini CM n*>f
-------
ZPCB (Jg/L) In segment 42
— Model Output
ZPCB (JglL) in segment 43
ZPCB (Og/L) in segment 44
30
24
18 -
12 -
— Model Output
24
18
12
— Model Output
Jan-94 Jul-94 DeC-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-9S
ZPCB (Og/L) in segment 45
— Model Output
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (Og/L) in segment 46
ZPCB (Og/L) in segment 47
24
18
12
6
— Model Output
30
24
18
12
6
— Model Output
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (Og/L) in segment 48
ZPCB (Og/L) in segment 49
ZPCB (Og/L) in segment 50
24
18
12
6
0
— Model Output
30 -,
24
18
12 -
6
— Model Output
~- .
24 •
18
12
6
0 -
— Model Output
'
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
IPCBs in particulate phase (pg/L)
Lake Michigan sediment segments
403
-------
EPCB (Og/L) in segment 51
EPCB (Og/L) In segment 52
IPCB (Og/L) in segment 53
30
24
18
12
6
0
Jan
30
24
18
12
e
0
Ja
— Model Output
'
.94 Jul-94 Dec-94 Jul-95 Dec
ZPCB (Og/L) in segment 54
— Model Output
n-94 Jul-94 Dec-94 Jul-95 De
EPCB (Og/L) in segment 57
30 -
24
18 •
12
6
— Model Output
Jan-94 Jul-94 Dec-94 Jul-95 Dec
30
24
18
12
6
0
EPCB (Og/L) in segment 60
— Model Output
30
24
18
12
6
0
-95 Jar
30 -
24
18
12
6 -
0 -
c-95 Jar
30
24-
18
12
6
0
95 Jan
30
24
18
12
6
0
— Model Output
^— -— _^
-94 Jul-94 Dec-94 Jul-95 Dec
EPCB (Og/L) in segment 55
— Model Output
^
-94 Jul-94 Dec-94 Jul-95 Dec
EPCB (Og/L) in segment 58
— Model Output
94 Jul-94 Dec-94 Jul-95 Dec
EPCB (Og/L) in segment 61
— Model Output
24
18 -
12
6
0 -I
-95 Jan
30-
24
18
12 -
6
0
-95 Jan
30
24
18
12
6
0
95 Ja
— Model Output
-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (Og/L) in segment 56
— Model Output
-94 Jul-94 Dec-94 Jul-95 Dec-95
EPCB (Og/L) in segment 59
— Model Output
n-94 Jul-94 Dec-94 Jul-95 Dec-95
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCBs in particulate phase (M9/L)
Lake Michigan sediment segments
404
-------
ZPCB (Og/L) in segment 63
IPCB (Og/L) in segment 64
30-
24
18
12
E
0
Ja
— Model Output
.
1-94 Jul-94 Dec-94 Jul-95 De
24
18
12
6
0
c-95 Ja
— Model Output
24
18
12
6
0
<-94 Jul-94 Dec-94 Jul-95 Dec-95 Jar
— Model Output
-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (Og/L) in segment 65 .. ^B (Og/L) in segment 66 EPCB (Jg/L) in segment 67
30
24
18
12
6
— Model Output
24
18
12
6
0
O-l , 1 1 1 Ja
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
— Model Output
1-94 Jul-94 Dec-94 Jul-95 De
24
18
12
G
— Model Output
0 -I 1 1 1 1
;-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
IPCBOalL) in segment 68 IPCB (VD in segment 69 ZPCB In segment 70
30 -
24
18
12-
6
0
— Model Output
, , ,
24
18
12
6 -
0 -t
— Model Output
24
18
12
6
— Model Output
1
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCBs in paniculate phase (M9/L)
Lake Michigan sediment segments
405
-------
ZPCB (Jg/L) in segment 72
ZPCB (JgIL) in segment 73
30
24
18
12
6
0
Jan
30 -
24 -
18 -
12 -
6
0
Jai
— Model Output
30 -I
- Model Output M . ~ Model Output
18 18
12 • 12 "
6 • 6 "
0 , 1 1 1 " '
-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec- u - ec-
ZPCB (Jg/L) in segment 74
— Model Output
IPCB (Jg/L) in segment 75 £pCB ^ |p segment 76
30
— Model Output
24- —Model Output
18 18
12 12
6
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 ° .
-94 Jul-94 Dec-94 Jul-96 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
IPCB ()glL) in segment 77
24-
18
12
e
n
— Model Output
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
IPCBs in particulate phase (pg/L)
Lake Michigan sediment segments
406
-------
IPCB (jg/L) in segment 79
ZPCB ()g/L) in segment 80
30-
24
18
6
— Model Output
24 •
18
12
6 -
— Model Output
24
18
12
6
0
— Model Output
lan 04 ini Qd nor- QA ini oft nA<- aE Jan-94 Jul-94 Dec-94 Jul-95 Dec
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-96 Dec-95
ZPCB(Jg/L) in segment 81 EPCB (JgIL) in segment 82 ZPCB()glL) in segment 83
24
18
12
e
Jan
JO
24
18
12
6
n
— Model Output
-94 Jul-94 Dec-94 Jul-95 Dec
ZPCB (Jg/L) in segment 84
— Model Output
24 •
18
12
6
— Model Output
-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec
EPCB (jg/L) in segment 85
24
18
12
6
— Model Output
' r • 1 1 1
24 •
18
12
6
0 -
-95 Jan-
30
24
18
12
6
0
— Model Output
94 Jul-94 Dec-94 Jul-95 Dec-9
ZPCB (JglL) in segment 86
— Model Output
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
IPCBs in particulate phase (pg/L)
Green Bay sediment segments
407
-------
ZPCB (jg/L) in segment 87
ZPCB (Jg/L) In segment 88
ZPCB (JgIL) in segment 89
30 -I
24 -
18
E
— Model Output
30 T
24
18
12
6
— Model Output
— —- ^__
24
18 -
12-
6 -
0 -
-Model Output
^— ^__
--— .
0 iii UH , , , 1
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan.94 Ju|.94 Dec.94 ju(.95 De(;.95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-S
ZPCB (J g/L) in segment 90 ZPCB () g/L) in segment 91
30
24
18 -
1?
6
A
— Model Output
-_
24
18
12
6
n
— Model Output
""--— ^____^^
24
18
12
E
0
ZPCB (JgIL) in segment 92
— Model Output
1 i 1
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95 Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
ZPCB (Jg/L) in segment 93
ZPCB (JgIL) In segment 94
24
18
12
6 •
— Model Output
Jan-94 Jul-94 Dec-94 Jul-96 Dec-95
24
18 •
12 -
6
n •
— Model Output
Jan-94 Jul-94 Dec-94 Jul-95 Dec-95
IPCBs in participate phase (pg/L)
Green Bay sediment segments
408
-------
U.O-
0.4-
"3> 0.3-
o 0.2-
Q-
W
0.1-
segment 1
» »
(L^«^^"%. W^"VA_**^.
• * ^"***+*i\f **-W
segment 2
*
» t «
**j^*^*^*^*^ /^^\*
^^ * ^^^^NX^*^'^^ ^vv
segment 3
* *
\ x*V»".™.* <^1
^v^"^ T^^^Xfcv^**-/*^^^^^*
segment 4
4
^ *
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
05
0.4.
ra 0.3.
0 0.2-
Q_
W
0.1-
o
segment 5
»
L*-^*~*vu^. » J-VJ»A
*v— ^v^*
» ^~~ «
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
0.4-
"5 0.3-
0 0.2-
Q.
W
0.1-
segment 20
»
jiL*-^,.^ *.
'V^Y^AA^J1 N^W
segment 21
,*
, »
^""^ v^
segment 22
*
^-^» « ^ "V*_^^. — -^— s
segment 23
«
* »
^—-^T^*1^— ^" — "^v
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Ju!94 Dec94 Jul95 Dec95 Jan94 Ju)94 Dec94 Jul95 Dec95
n **
0.4-
3
g> 0.3-
CO
0 0.2-
w
01-
0-
segment 24
»
»J«w-<«i * * M
*
segment 25
*^~«vs *
^ — ^V; — -> — ^v-
•^ Model Output
» Cruise Mean
error bars = standard error
SPCB (ng/L)
Lake Michigan layers 1 , 2, 3
upper 30 meter
water segments
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
409
-------
0.5-1
0.4-
1 0.3-
m
0 0.2-
a
w
0.1-
segment 29
it *
ilijUt- ii »Ju[ 1 __*JLJ
« • *vvw^~-r-
segment 30
*
4* ££ rm" *y* 1
segment 31
*
r~> ^
^— *»» X*'^* ^^
segment 32
^ ,
, rM 4
^v^*<7 ^*X*- "V.
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec'
n ^
0.4.
°> 0.3-
CO
o 0.2-
CL
W
0.1.
0-
segment 33
** *S — "frlfSn
^^Js+ W^-vv^-^ ^»V^
segment 34
*
.
^^**" ^ ^^^
^<— *" v^-wv^^~^"^pvl
.
segment 36
•
^^»^*^-i
-J "^^ . ^\^ ^s^~" "^*
« *
.
segment 37
r-^7v
i -^^| ^V^ ^s**+ ,-^«
* ^>jp*'^
*
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec£
«*. Model Output
Q 5 * nrnica tiflemn
0.4.
2"
D) 0.3-
0 0.2J
0.
w
0.1-
0-
segment 38
^-**^^N^ —
^^f^^^ ^*^*~ • -^**~ ~*
f1^ * *
^ * 4 * *
-cament 30 error bars = standard error tinmpnt 4D
,cgmont30 vPCB (ng/L) segment 40
Lake Mich gan layers 4 & 5
30 meters to bottom
water segments
^^T^^
-^"^ ^^^^^ ^^^~ ' ~ '*
f^^ * *
*»* *
-
--""—V
/^*^ 1^^ ^ *>^^^^
_i**^^ * ^^^^**^^^ * A
^ ' »
-
segment 41
^-**~\
f^^ *w_ ^**^^^X
f~^^^ ~^ ^ ^" * * T
f • *
• » »
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Deo94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec9S
410
-------
m
o
a.
1.0'
0.8-
0.6
0.4-
0.2
CO
O
Q.
W
segment 7
»
Vv\wW^*s^
2.4-
1.6-
0.8-
segment 9
Jan94 Jul94 Dec94 Jul95 Deo95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
1.0-
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
I
03
W
«>» Model Output
* Cruise Mean
error bars = standard error
SPCB (ng/L)
Green Bay
water segments
Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95 Jan94 Jul94 Dec94 Jul95 Dec95
411
-------
ZPCB (jig/L) in segment 42
ZPCB (ug/L) in segment 43
ZPCB (ug/L) in segment 44
30 -
24 -
18
12
6
0
Jar
30
24
18 -
12
6
0 -
Jar
30
24
— Model Output
• ^
^
1.94 Jul.94 Dec.94 Jul.95 De
ZPCB (ug/L) in segment 45
— Model Output
^^^ '
.94 Jul.94 Dec.94 Jul.95 Dec
ZPCB (ug/L) in segment 48
-
18
12
6
0
30 -
24 -
18 -
12
6
0
;.9S Jai
30 -
24 -
18 •
12
6
0
.95 Jan
— Model Output
30
24
18
12
6
0
— Model Output
---
— Model Output
18
12
.94 Jul.94 Dec.94 Jul.95 Dec. 95 Jan.94 Jul.94 Dec.94 Jul.95 Dec
ZPCB (ug/L) in segment 46 ZPCB (ug/L) in segment 47
— Model Output
' •
— Model Output
18
12
6
94 Jul.94 Dec.94 Jul.95 Dec. 95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.9
ZPCB (ug/L) in segment 49 IPCB (" 9"-) in segment 50
*n
— Model Output
' .
— Model Output
18
12 _____________^
6
0 1 1 1
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
IPCBs (dissolved + particulate) (|jg/L)
Lake Michigan sediment segments
412
-------
ZPCB (ug/L) in segment 51
— Model Output
ZPCB (ng/L) in segment 52
-Model Output
ZPCB (|ig/L) in segment 53
30
24
18
12 •
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
ZPCB (ug/L) in segment 54
ZPCB (ug/L) in segment 55
ZPCB (ug/L) in segment 56
30-i
24
18
12
6 •
0 -I
— Model Output
"^ '
24
18 -
12
6
— Model Output
_______
24
18
12
6
— Model Output
Jan.94 Jul.94 Dec.94 Jul.96 Dec.95
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
ZPCB (ug/L) in segment 57
ZPCB (ug/L) in segment 58
ZPCB (ug/L) in segment 59
30 -I
24
18
12
6
0 -I
— Model Output
24
18
12
6
0
— Model Output
24
18
12
6
0
— Model Output
Jan.94 Jul.94 Dec 94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
ZPCB (ug/L) in segment 60
ZPCB (ug/L) in segment 61
24
18
12
ft
— Model Output
24
18
12
6
n
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.96
iPCBs (dissolved + participate) (|jg/L)
Lake Michigan sediment segments
413
-------
ZPCB (fig/L) in segment 62
1
24
18
12
b
— Model Output
ZPCB (|ig/L) in segment 63
ZPCB (fig/L) in segment 64
30 -1
24
18
12
6
n -
— Model Output
24
18
12
6 •
0 -I
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
ZPCB ((ig/L) in segment 65
ZPCB (jig/I) in segment 66
EPCB (|ig/L) in segment 67
— Model Output
24
18
12 -
6
0 -I
— Model Output
30 -I
24
18
12
6
0 -I
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan 94 Ju, 94
Ju| 95
ZPCB (fjg'L) in segment 68
ZPCB (ug/L) in segment 69
ZPCB (|ig/L) in segment 70
30 -I
24
18
12 •
6
0 -
— Model Output
30 -I
24
18
12
6
0
— Model Output
I- 1 1 1 1
24
18
12-
6
0
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
ZPCBs (dissolved + particulate) (|jg/L)
Lake Michigan sediment segments
414
-------
IPCB (ug/L) in segment 71
IPCB (ng/L) in segment 72
IPCB (»ig/L) in segment 73
24
18
12
6
0
Jan
30
24
18
12
6
0
Ja
— Model Output
.94 Jul.94 Dec.94 Jul.95 Dec
ZPCB (ug/L) in segment 74
— Model Output
-
-
24
18
12
6
0 -
.95 Jar
30 -
24
18 -
12 -
6
— Model Output
.94 Jul.94 Dec.94 Jul.95 Dec
ZPCB (ug/L) in segment 75
— Model Output
n.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan-94 Jul-94 Dec 94 Jul'95 De<
ZPCB (ug/L) in segment 77
24
18
12
6
n -
— Model Output
24
18
12
6
0 -\
.96 Jar
30
24
18
12
6
0
.95 J
— Model Output
.94 Jul.94 Dec.94 Jul.96 Dec.95
•
•
ZPCB (ug/L) in segment 76
— Model Output
in. 94 Jul.94 Dec.94 Jul.95 Dec.95
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
IPCBs (dissolved + particulate) (pg/L)
Lake Michigan sediment segments
415
-------
ZPCB (ug/L) in segment 78
IPCB (ug/L) in segment 79
ZPCB (ug/L) in segment 80
3D
24
18
6 •
— Model Output
JO
24 •
18
12
6
n .
— Model Output
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.9S Dec.95
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
ZPCB (ug/L) in segment 81
— Model Output
ZPCB (M9/L) in segment 82
ZPCB (ug/L) In segment 83
24
18
12
6
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
24
18
12
6
ft
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
ZPCB (ug/L) in segment 84
30
24
18 •
12
6
— Model Output
ZPCB (ug/L) in segment 85
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95 Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
ZPCB (ug/L) in segment 86
24
18
12
G
n -
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
IPCBs (dissolved + participate) (Mg/L)
Green Bay sediment segments
416
-------
ZPCB (ug/L) in segment 87
24
18
17
6
— Model Output
ZPCB (fig/L) in segment 88
30
24
18
12
6 •
— Model Output
Jan.94 Jul.94 Dec.94 Jul.96 Dec.95
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
ZPCB (ng/L) in segment 89
Jan.94 Jul.94 Dec.94 Jul,95 Dec.95
ZPCB (ug/L) in segment 90
ZPCB (fig/L) in segment 91
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
Jan.94 Jul.94 Dec.94 Jul.96 Dec.95
ZPCB (ug/L) in segment 92
24-
18
12
6
n .
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
24-
18
12
6 -
ZPCB (|ig/L) in segment 93
— Model Output
ZPCB (ug/L) in segment 94
24
18
12
6
— Model Output
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
Jan.94 Jul.94 Dec.94 Jul.95 Dec.95
ZPCBs (dissolved + particulate) (pg/L)
Green Bay sediment segments
417
-------
Appendix 4.5.4
Simulation Results From Chloride
418
-------
Segment 1
Segment 2
en
c
QJ
O
_C
O
15
12
9
6
3
0
-
-
,
en
c
O)
"0
o
o
15
12
9
6
3
0
. ^ t • • - * •
-
-
0 120 240 359 479 599
Time (Day 0 = 4/1/94)
0 120 240 359 479 599
Time (Day 0 = 4/1/94)
12
0)
o 6
_c
O
3
0
Segment 3
0 120 240 359 479 599
Time (Day 0 = 4/1 /94)
en
c
0)
"O
-
O
15
12
9
6
3
0
Segment 4
0 120 240 359 479 599
Time (Day 0 = 4/1/94)
15
12
en
c 9
o 6
_c
O
3 -
0
Segment 5
0 120 240 359 479 599
Time (Day 0 = 4/1/94)
15
12
en
c
CL>
"O
O
0
Segment 6
0 120 240 359 479 599
Time (Day 0 = 4/1/94)
Chloride
419
-------
1 5
1 2
9
6
3
O
Seg men t 7
12O 24O 359 479 599
Time (Day O = 4/1/94)
1 5
1 2
9
6
3
O
Seg ment
O 12O 24-O 359 479 599
Time (Day O = 4/1/94)
1 5
1 2
9
6
3
O
Segment
O 1 2O 24O 359 479 599
Time (Day O = 4/1/94)
1 5
1 2
1r 9
CD
~O
| 6
-------
15
-12
c 9
CD
-a
'§ 6
-C
0 3
0
Segment 1 1
-
15
\ 12
c 9
*
15
O2
c 9
-------
CD
T3
15
O2
en
c 9
V
-a
'o 6
_c
u 3
0
Segment 20
• . • • • •
15
^12
cr>
c 9
1 6
_c
" 3
0
Segment 21
j . • • •
15
O2
CT>
c 9
0)
"Q
1 6
.c
0 3
0
Segment 22
. . • • • •
_o
6
0 120 240 359 479 599
Time (Day 0 = 4/1/97)
120 240 359 479 599
Time (Day 0 = 4/1/97)
120 240 359 479 599
Time (Day 0 = 4/1/97)
Segment 23
Segment 24
\
c
0)
u
\_
o
_c
U
15
12
9
6
3
0
• i • • • •
\
c
CD
•a
0
.c
O
15
12
9
6
3
0
-* • " •
*
-
-
120 240 359 479 599
Time (Day 0 = 4/1/97)
120 240 359 479 599
Time (Day 0 = 4/1/97)
Segment 25
\
c
CD
0
_O
O
1b
12
9
6
3
0
['- ' ' - t -
-
-
CM
C\J
0 120 240 359 479 599
Time (Day 0 = 4/1/97)
Segment 26
Segment 27
Segment 28
15
\ 12
c 9
0)
"D
o 6
.c
u 3
0
15
\ 12
c 9
CD
-a
o 6
(~
0 3
0
— — . r • •
-
-
,
15
-12
c 9
CD
~U
o 6
JZ
0 3
0
^>^-_- . _^
-
-
0 120 240 359 479 599
Time (Day 0 = 4/1/97)
120 240 359 479 599
Time (Day 0 = 4/1/97)
120 240 359 479 599
Time (Day 0 = 4/1/97)
-------
Segment 29
15
12 -
0
0 120 240 359 479 599
Time (Day 0 = 4/1/94)
15
12
en
c 9
CD
TD
o 6
-C
O
3
0
Segment 30
0 120 240 359 479 599
Time (Day 0 = 4/1/94)
Segment 31
Segment 32
en
c
cu
~o
o
O
1b
12
9
6
3
0
-
-
r
-
15
12
en
c 9
CD
TD
_° 6
o
3
0
-
-
-
-
0 120 240 359 479 599
Time (Day 0 = 4/1/94)
0 120 240 359 479 599
Time (Day 0 = 4/1/94)
Segment 33
Segment 34
15
12
If 9
CD
TD
^ 6
-C
O
3
0
• i i r- —
• • . . • . •
-
1 1 1 i
]5
12
en
c 9
0)
"O
i_
.2 6
o
3
0
i i i r~ ' '~
• » • •- B » •
-
0 120 240 359 479 599
Time (Day 0 = 4/1/94)
0 120 240 359 479 599
Time (Day 0 = 4/1/94)
Chloride
423
-------
en
c
cu
~g
' i_
_o
_c
O
15
12
9
6
3
0
Segment
0 120 240 359 479 599
Time (Day 0 = 4/1/97)
Chloride
424
-------
Segment 36
15
12 -
S1 9
c 9
OJ
XI
v_
o 6
o
3
0
Segment 40
0 120 240 359 479 599
Time (Day 0 = 4/1/94)
15
12
cr>
c
CD
ID
_c
O
0
Segment 4'
0 120 240 359 479 599
Time (Day 0 = 4/1/94)
Chloride
425
-------
PART 4
LM2-TOXIC
Appendix 4.5.5. Primary Production for
the LM2-Toxic
Amy Anstead
Welso Federal Services, LLC
Large Lakes Research Station
9311 Groh Road
Grosse lie, Michigan 48138
The LM2-Toxic hindcast model (1940-1995) relies on
external measurements or calculations to provide the
internally produced (primary production) carbon
loads. There have been very few studies published
that estimated lake-wide primary productivity for Lake
Michigan (Fee, 1973). It is, therefore, difficult to put
together a primary productivity history from
1940-1995 as required for the model hindcast.
However, historical total phosphorus loads and
concentration values were available, and a
relationship between total phosphorus and primary
productivity was used to estimate primary
productivity. Lake Michigan total phosphorus loads
were estimated from 1974-1991 by the International
Joint Commission (IJC) (Great Lakes Water Quality
Board, 1989; Pauer era/., 2006), and for 1994-1995
as part of Lake Michigan Mass Balance Project
(LMMBP) (described in Part 2). Prior to 1974, there
were only a few reports published that documented
measured total phosphorus loads (Patalas, 1972;
Lee 1974). However, total phosphorus loads from
1800-1970 were estimated using a model that
incorporated phosphorus sources and sinks for the
Great Lakes (Chapra, 1977). There were reliable
lake-wide total phosphorus concentration data from
1976 to present and measurements dating back to
the 1950s (Risley and Fuller, 1965; Rockwell et al,
1980). In most cases the data prior to 1976 were
location specific, and it was difficult to extrapolate the
results to obtain a representative lake-wide total
phosphorus concentration.
Modeled loads from 1800-1850 (Chapra, 1977) were
higher than measured total phosphorus loads from
1994-95 (Figure 4.5.5.1). It was curious that
measured total phosphorus loads in recent times of
higher anthropogenic phosphorus input (1994-1995)
were lower than modeled total phosphorus loads
during the pre-Western civilization period (1800-
1850). There seemed to be a disconnect between
the measured loads of recent times and the loads
modeled for the 1800s. The model simulation was
recreated and extended until 2000. As expected, the
model overpredicted total phosphorus loads when
compared with the measured data (Figure 4.5.5.1).
Improvements were made to the model parameters
which resulted in a better fit to the measured data
from 1974-1995 (Figure 4.5.5.1). Modifications to the
model included decreasing phosphorus export
coefficients for land use, decreasing per capita
detergent phosphorus export and decreasing
atmospheric phosphorus loads. All modifications
made fell within reported literature values for the
Great Lakes system (Reckhow and Simpson, 1980;
Rast and Lee, 1983; U.S. Environmental Protection
Agency, 1976). The model also simulated annual
average surface water total phosphorus
concentration for Lake Michigan by simply dividing
modeled total phosphorus load by lake volume.
Modeled total phosphorus concentration compared
well to measured total phosphorus concentration
values for Lake Michigan (Rockwell et al., 1980;
Risley and Fuller,1965; Beeton and Moffett, 1964;
426
-------
16000
original model
modified model
n measured TP loads from IJC and LMMBP
1800
1825
1975 2000
Figure 4.5.5.1. Model data versus measured total phosphorus loads.
Holland,1969; Rousar and Beeton, 1973) (Figure
4.5.5.2). The adjusted model output provided a
complete annual average total phosphorus
concentration history from which primary productivity
was calculated.
Vollenweider etal. (1974) established a relationship
between total phosphorus load and primary
productivity but it was limited by few values from a
time when phosphorus loads and primary productivity
were at their highest in Lake Michigan's history. The
relationship was not as strong for years with lower
total phosphorus loads. We established a total
phosphorus-primary productivity relationship (Figure
4.5.5.3) from the output of a 1976-1995 hindcast of
the eutrophication model, LM3-Eutro. This
relationship was used to calculate lake-wide annual
primary productivity from 1940-1995 Total
phosphorus concentration history. Final annual
primary production (as organic carbon) was provided
as a spreadsheet for incorporation into LM2-Toxic.
References
Beeton, A.M. and J.W. Moffett. 1964. Lake
Michigan Chemical Data, 1954-55, 1960-61.
U.S. Department of the Interior, U.S. Fish and
Wildlife Service, Ann Arbor, Michigan. Data
Report 6.
Chapra, S.C. 1977. Total Phosphorus Model for the
Great Lakes. J. Environ. Engin., 103(EE2):147-
161.
Fee, E.J. 1973. A Numerical Model for Determining
Primary Production and Its Application to Lake
Michigan. J. Fish. Res. Board Canada,
30(10): 1447-1468.
Great Lakes Water Quality Board. 1989. 1987
Report on Great Lakes Water Quality, Appendix
B: Great Lakes Surveillance, Volume I.
International Joint Commission, Windsor, Ontario,
Canada. 287 pp.
427
-------
o
c
o
c
-------
Holland, R.E. 1969. Seasonal Fluctuations of Lake
Michigan Diatoms. Limnol. Oceanogr.,
14(3):423-436.
Lee, G.F. 1974. Phosphorus, Water Quality and
Eutrophication of Lake Michigan. Proceedings of
the Fourth Session of the 1972 Conference on
Pollution of Lake Michigan and Its Tributary Basin
by Illinois, Indiana, Michigan, and Wisconsin.
U.S. Environmental Protection Agency, Region V,
Chicago, Illinois.
Patalas, K. 1972. Crustacean Plankton and the
Eutrophication of St. Lawrence Great Lakes. J.
Fish. Res. Board Canada, 29(11): 1451-1462.
Pauer, J.J., A.M. Anstead, K.W. Taunt, W. Melendez,
and R.G. Kreis, Jr. 2006. The Lake Michigan
Eutrophication Model, LM3-Eutro: Model
Development and Calibration. Submitted for
publication to the Canadian J. Fish. Aquat. Sci.
Rast, W. and G.F. Lee. 1983.
Estimates for Lakes. J.
109(2):502-517.
Nutrient Loading
Environ. Engin.,
Reckhow, K.H. and J.T. Simpson. 1980. A
Procedure Using Modeling and Error Analysis for
the Prediction of Lake Phosphorus Concentration
From Land Use Information. Canadian J. Fish
Aquat. Sci., 37(9): 1439-1448.
Risley, C. and F. Fuller. 1965. Chemical
Characteristics of Lake Michigan. In:
Proceedings of the Eighth Conference on Great
Lakes Research, pp. 168-174. Great Lakes
Research Division, The University of Michigan,
Ann Arbor, Michigan.
Rockwell, D.C., C.V. Marion, M.F. Palmer, D.S.
DeVault.andR.J.Bowden. 1980. Environmental
Trends in Lake Michigan. In: R.C. Loehr, C.S.
Martin, and W. Rast (Eds.), Phosphorus
Management Strategies for Lakes, Chapter 5, pp.
91-132. Ann Arbor Science Publishers,
Incorporated, Ann Arbor, Michigan.
Rousar, D.C. and A.M. Beeton. 1973. Distribution of
Phosphorus, Silica, Chlorophyll a, and
Conductivity in Lake Michigan and Green Bay.
In: E. McCoy (Ed.), Transactions of the
Wisconsin Academy of Sciences, Arts, and
Letters, pp. 117-140. University of Wisconsin,
Madison, Wisconsin.
U.S. Environmental Protection Agency. 1976.
Areawide Assessment Procedures Manuals.
Volume 2. U.S. Environmental Protection
Agency, Municipal Environmental Research
Laboratory, Cincinnati, Ohio. EPA/600/9-76/014,
427 pp.
Vollenweider, R.A., M. Munawar, and P. Stadelmann.
1974. A Comparative Review of Phytoplankton
and Primary Production in the Laurentian Great
Lakes. J. Fish. Res. Board Canada, 31 (5):739-
762.
429
-------
PART 4
LM2-TOXIC
Chapter 6. The LM2-Toxic Application
and Interpretation
One of the important goals during the development of
the LM2-Toxic was to quantitatively understand
polychlorinated biphenyl (PCB) dynamics (i.e.,
transport and fate of PCBs) in the Lake Michigan
system and delineate the relationship between PCB
external loads and its concentrations in the system.
For the following discussion, Lake Michigan refers to
the main part of the lake (main lake) only, excluding
Green Bay. Any references to the Lake Michigan
system should be considered to include both Green
Bay and Lake Michigan. Lake Michigan system =
Lake Michigan + Green Bay. Main lake = Lake
Michigan. After calibration of organic carbon and
PCB congener dynamics and model confirmation, the
LM2-Toxic was used as a mass budget diagnostic
tool to identify the critical contaminant sources and
sinks and key environmental processes in Lake
Michigan and Green Bay. The model was also
applied for forecasting the long-term responses of the
Lake Michigan system to a variety of forcing
functions and load reduction scenarios for PCBs.
The mass budget analysis and the long-term
forecasts under the specified load reductions were
intended to provide information useful in making
management decisions for the Lake Michigan
system. Long-term PCB exposure concentrations
predicted from the LM2-Toxic for various forcing
functions and load reduction scenarios were provided
to LM Food Chain as forcing functions to compute
PCB concentrations in fish tissue.
4.6.1 Conversion of PCB Congener
Results to Total PCBs
LM2-Toxic is a PCB congener-based model. It was
developed to compute the concentrations, total mass,
and mass movement (fluxes) of 54 PCB congeners
in each compartment of the Lake Michigan system.
These 54 PCB congeners account for about 63% to
85% of the total PCB mass in the various media (see
Table 3.1.1 for the list of ratios between total PCBs
and the summed modeling congeners in all media).
There were enormous amounts of information related
to the model inputs and outputs from the model on a
basis of each PCB congener. For the efficiency and
effectiveness of presenting the input information and
results from the model, and for the convenience of
the reviewers and readers, all numbers in this
chapter are presented as total PCBs. The model
outputs for each Lake Michigan Balance Project
(LMMBP) selected PCB congener were first summed
as IPCBs, and the IPCBs was then converted to
total PCBs using a regression between total PCBs
and IPCBs. The regression analysis for different
media was done by Computer Sciences Corporation
(CSC)/Large Lakes Research Station (LLRS)
personnel and is detailed in Part 1, Chapter 3 of this
report. The regression equations used in this chapter
for the specified media are listed in Table 4.6.1.
More information on regression equations for all
media can be found in Table 1.3.1.
4.6.2 Mass Budget Diagnosis of the LM2-
Toxic for the LMMBP Period
A mass budget diagnostic tool was developed within
the LM2-Toxic in order to quantitatively analyze the
430
-------
Table 4.6.1. Regression Equations Used for Converting ZPCBs to Total PCBs for the LM2-Toxic
Results
Media
Regression Equation
Dissolved Water
Particulate Water
Dissolved + Particulate Water
Surficial Sediment
y = 1.2738x + 0.0268
y = 1.2251 x + 0.0051
y = 1.2427x + 0.0347
y = 1.1668x + 0.6125
0.9413
0.9992
0.9829
0.9970
behavior of PCBs in the Lake Michigan system. This
tool has the ability to estimate very detailed PCB
mass fluxes in the lake, mass inventories in different
compartments of the lake, phase distributions, and
contaminant residence times in the system.
Therefore, the results of the mass budget diagnosis
were used to demonstrate the most significant PCB
sources and sinks and to identify key environmental
processes in the Lake Michigan system.
A mass budget diagnosis was performed for each
selected PCB congener modeled in the LM2-Toxic
for the two-year LMMBP period (1994-1995). The
final results of the mass budget diagnosis are
presented as the annual total PCBs only. Figures
4.6.1 and 4.6.2 provide a summary of the results of
the total PCB mass budget diagnosis and analysis in
Lake Michigan and Green Bay. Figure 4.6.1 depicts
the masses transported and inventories for the entire
Lake Michigan system that includes Green Bay.
Figure 4.6.2 depicts the mass transported and
inventories for Green Bay separately from the main
lake. Table 4.6.2 lists more detailed results of the
total PCB mass budget analysis, including total PCB
mass distributions in different phases and residence
times in the system. The diagrams and table also
give an indication of the importance for each
environmental process conceptualized in the Lake
Michigan system. The unit of the annual average
mass fluxes (average of the two-year LMMBP period
- 1994-1995) in the mass budget diagrams is in
kg/year. The mass inventories in the diagrams for
both water column and surficial sediment (0-4 cm)
are the average mass at any time over the two-year
LMMBP period and in units of kg. Due to seasonal
variations in the concentration of both the water
column and the surficial sediments in the Lake
Michigan system, the numbers for inventories can be
different on any given day in the LMMBP period. The
average mass of total PCBs in the water column of
the Lake Michigan system during 1994-1995 was
1,216 kg. About 30% (370 kg) of the total PCB mass
in the water column was in the particulate phase
(particulate detrital carbon [PDC] bounded + biotic
carbon [BIG] bounded). Dissolved phase (dissolved
organic carbon [DOC] bounded + unbounded)
accounted for approximately 70% (846 kg) of the
average mass of total PCBs in the water column.
The average mass of total PCBs in the surficial
sediments (0-4 cm) during the LMMBP period was
13,085 kg, and virtually all of the mass in the surficial
sediment was bound to PDC. Based on the volumes
of the water column (4.8148 x 1012 m3) and the
surficial sediment layer (1.0871 x 109 m3) of the Lake
Michigan system, the average concentration of total
PCBs in the water column was 0.253 ng/L, and the
average concentration of total PCBs in the surficial
sediment layer was 12,037 ng/L. These
concentrations were consistent with the average
concentrations (0.259 ng/L for the water column and
650-25,000 ng/L for the surficial sediment layer)
derived from the LMMBP field data.
Compared with Lake Michigan, the total PCB mass
distributions in dissolved and particulate phases were
quite different in the Green Bay water column. The
inventories of particulate and dissolved PCBs in
Green Bay were almost equal. The higher PCB
mass in the particulate phase in Green Bay was due
to the dominant tributary load from the Fox River. In
the river, the particulate PCB concentrations were
much higher than the dissolved PCB concentrations.
In Lake Michigan, the particulate PCBs were less
than half of the dissolved PCBs in its water column.
431
-------
volatilization
3439
gas absorption
1507
resuspension
1393
m
export to
Lake Huron
12
atmospheric
deposition
980
Chicago,
River {
export /
8
water column =1216
active sediment = 13,085
(0-4 cm interval)
sediment
burial
1284
monitored and unmonitored
tributary loading
(Lake Michigan watershed)
Figure 4.6.1. Mass budget average for 1994-1995 total PCBs in the Lake Michigan system (including
Green Bay). Unit of the masses transported (arrows) is in kg/year.
432
-------
input from
Lake-Huron
4
Green Bay
volatilization
721
Green Bay
gas absorption
111
Green Bay
monitored and
unmonitored
tributary loading
230 C
Green Bay
resuspension
i453
-rr-r •= J'
Green Bay
sediment burial
68
settling i
355 •
atmospheric
deposition
267
Green Bay Mass Budget
main lake
volatilization
2718
Green Bay
export
128
mam lake
gas absorption
1395
atmospheric
deposition
714
export to
Lake Huron
12
resuspension
939
PCB Inventory kg
export via
Chicago
Diversion
Main Lake:
water column = 1165
active sediment = 10380
(0-4 cm interval)
mam lake monitored and
Linmonitored tributary loading
(Lake Michigan watershed
excluding Green Bay)
151
Green Bay:
water column = 51
active sediment = 2704
(0-4 cm interval)
sediment
burial
1216
Figure 4.6.2. 1994-1995 total PCB Lake Michigan and Green Bay mass budget (averaged). Unit of the
masses transported are in kg/year.
433
-------
Table 4.6.2. Results of Total PCB Mass Budget Analysis for Lake Michigan and Green Bay (Fluxes are
Annual Averages of the Two-Year LMMBP -1994-1995; Inventories are the Average Inventories of the
Two-Year LMMBP Period)
Mass Budget Component
Lake Michigan
+ Green Bay
Lake Michigan Green Bay
Fluxes (kg/year)
Total Loads
(Monitored Tributary)
(Unmonitored Tributary)
(Atmospheric Dry)
(Atmospheric Wet)
Settling
(PDC Bounded)
(BIG Bounded)
Resuspension
Burial (Sedimentation)
Diffusion
Absorption
Gross Volatilization
Net Volatilization
Input From Lake Huron
Export to Lake Huron
Net Output to Lake Huron
Chicago Diversion
Net Flux From Green Bay to Lake Michigan
Inventory (kg)
Water Column
(PDC Bounded)
(BIG Bounded)
(DOC Bounded)
(Unbounded)
Surficial Sediment (0-4 cm)
Residence Time in Water Column (Days)
Residence Time in Sediment (Days)
1362
348
33
767
214
1136
1076
61
1393
1284
154
1507
3439
1932
4
12
8
1
128
1216
297
73
36
810
13085
97
1688
865
124
26
514
200
782
764
17
939
1216
137
1395
2718
1323
4
12
8
1
128
1165
277
67
35
786
10380
121
1653
496
224
7
253
13
355
311
44
453
68
16
111
721
610
128
51
19
6
1
24
2704
17
1837
Note: The fluxes represent the masses transported across the total area of an interface between adjacent
compartments of the Lake Michigan system. Residence time for the water column was calculated by dividing
the annual average inventory in the compartment by the annual total output (sum of gross volatilization,
settling, export to Lake Huron, and Chicago diversion) from the water column. Residence time for the surficial
sediment layer was derived by dividing the annual averaged sediment inventory by the sum of the losses
(burial, resuspension, and diffusion).
434
-------
The information on the individual fluxes of total PCBs
in Figures 4.61 and 4.6.2 and Table 4.6.2 on an
annual average basis for the LMMBP period provide
further quantitative diagnosis on the importance of
each source, sink, and the key environmental
process in the Lake Michigan system. The single
largest flux leaving the lake was gross volatilization
(3,439 kg/year). This flux was countered by the flux
from gas absorption (1,507 kg/year) as the largest
source to the lake. The air-water exchange was the
most important process for the Lake Michigan
system. It accounted for the largest net loss (1,932
kg/year) from the Lake Michigan system of which
31.6% was from Green Bay (610 kg/year).
Resuspension (1,393 kg/year) was a major flux into
the water column of the Lake Michigan system. This
was offset by the flux from settling (1,136 kg/year).
The processes associated with the interaction
between the water column and the surficial
sediments (resuspension and settling) were very
important processes in the Lake Michigan system.
The results of these processes led to the second
largest net source (257 kg/year) next to the total
external load (1,362 kg/year) for the water column of
the Lake Michigan system. About 40% of this net
gain from the sediment-water interactions was
contributed from Green Bay (98 kg/year). Green Bay
received more than one-third (496 kg/year) of the
total external load. The principal loss in the surficial
sediment layer was burial (1,284 kg/year). The flux
contributed by diffusion from the surficial sediment
layer to the water column was 154 kg/year. The high
value for the mass transported by the diffusion from
the surficial sediment layer was not unexpected. The
reason for the high value could lie in the selection of
the diffusion coefficient used in the LM2-Toxic (1.73
x 10'4 m2/day) (DePinto et al., 1993) for sediment-
water diffusion process. Compared with the
coefficient used in the level 1 model (1.8 x 10'5
m2/day, Table 3.3.2) and the coefficient defined
under low-flow conditions for river sediment (1.5 x
10'4 m2/day) (Ortiz et al., 2004), the coefficient used
in the LM2-Toxic was a bit higher. This could be
another reason for the higher PCB concentrations for
the bottom water segments output from the model
during the LM2-Toxic PCB dynamics calibration (see
Chapter 5 for details).
For the Lake Michigan system, the total external PCB
'oad (monitored tributary, unmonitored tributary,
atmospheric dry, and atmospheric wet loads) and
input from Lake Huron was 1366 kg/year. The total
output or loss was equal to 3,229 kg/year due to net
volatilization, sediment burial, export to Lake Huron,
and Chicago diversion. Therefore, there was a net
loss of 1,863 kg/year of total PCBs from the entire
Lake Michigan system. This indicated that both the
water column and the surficial sediment layer of the
lake were not at steady-state during the LMMBP
period. By examining the mass fluxes of total PCBs
in the water column alone, the annual average total
gain (1,777 kg/year) during the project period was the
sum of total external load, net resuspension
(resuspension flux - settling flux) from the surficial
sediment layer, diffusion from the surficial sediment
layer, and input from Lake Huron. The annual
averaged total loss for the water column during the
same period was 1,945 kg/year and was equal to the
sum of net volatilization, export to Lake Huron and
Chicago diversion. Thus the water column
experienced a total net annual loss of 168 kg/year.
The annual average net export of total PCBs from
Green Bay to Lake Michigan was equal to 128
kg/year during 1994-1995. The number is very close
to the value (122.3 kg/year) estimated from the 1989
Green Bay Mass Balance Project (GBMBP) (DePinto
et al., 1993). A net loss of 1,695 kg PCB per year
from sediment was due to burial below the surficial
sediment (1,284 kg/year); net resuspension to the
water column (257 kg/year = resuspension flux 1,393
kg/year - settling flux 1,136 kg/year); and diffusion
(154 kg/year) from the surficial sediment layer to the
water column.
Residence time for the water column was calculated
by dividing the annual average inventory in the
compartment by the annual total output (sum of
gross volatilization, settling, export to Lake Huron,
and Chicago diversion) to the water column.
Therefore, the water column total PCB residence
times for Lake Michigan including Green Bay, Lake
Michigan only, and Green Bay only were
approximately 97, 121, and 17 days, respectively.
Similarly, the residence time for the surficial sediment
layer was derived by dividing the annual averaged
sediment inventory by the sum of the losses (burial,
resuspension, and diffusion). Thus the total PCB
residence times in the surficial sediment layer for
Lake Michigan including Green Bay, Lake Michigan
only, and Green Bay only were about 1,688, 1,653,
and 1,837 days, respectively. Thus, PCBs reside in
435
-------
the surficial sediment layer much longer than in the
water column of Lake Michigan.
4.6.3 LM2-Toxic Application for Long-
Term Forecast and Sensitivity Scenarios
Predictions of long-term PCB dynamics under a
variety of external forcing conditions were made
using LM2-Toxic for seven PCB forecast and
sensitivity scenarios. The model-predicted PCB
concentrations were then used as the time-
dependent exposure concentrations in the LM Food
Chain to calculate PCB concentrations in lake trout.
The simulation period for each scenario was 62
years, starting on January 1, 1994 and ending on
December 31, 2055.
All scenarios used the same LMMBP-generated field
data as input for the first two years (1994-1995) of
the simulations. Then, each scenario began on
January 1, 1996 and ran for a period of 60 years.
The observed PCB total load for the LMMBP period
(1994-1995) was adjusted upward by a factor of 1.98.
The adjusted PCB load is consistent with the 1994
total load used in the PCB hindcast (see 1.7.3 of this
report for details on the derivation of the PCB
hindcast loading function) and is a reasonable
estimate when considering the possibility of missing
atmospheric loads during the LMMBP period (see
1.7.3 of this report for a detailed discussion).
The seven PCB forecast and sensitivity scenarios
were:
A) Constant Conditions - The measured PCB loads
(tributary load plus atmospheric dry and wet
deposition) for the LMMBP period (1994-1995),
but adjusted upward by a factor of 1.98. The
adjusted loads followed the same spatial
distribution and monthly variation patterns
established by the LMMBP measured PCB loads.
The adjusted loadings, the 1994-1995 vapor-
phase concentration, Lake Huron boundary
conditions, and all other forcing functions as
observed in 1994 and 1995 were repeated
throughout the simulation period. Sediment
burial was active as well as all other model
processes.
B) Continued Recovery (Fast) - This was the same
as Scenario "A", but atmospheric components
(vapor phase concentration, wet and dry
deposition) declined with a six-year half-life
(Hillery etal., 1997; Schneider et at, 2001), and
tributary loads declined with a 13-year half-life
(Endicott, 2005; Marti and Armstrong, 1990).
The boundary conditions at the Straits of
Mackinac declined at a rate of 0.17/year (a four-
year half-life) (Schneider et al., 2001). These
rates were applied starting on January 1,1996.
C) Continued Recovery (Slow) - This was the same
as Scenario "A", but atmospheric components
(vapor phase concentration, wet and dry
deposition) declined with a 20-year half-life
(Buehler etal., 2002) and tributary loads declined
with a 13-year half-life. The boundary conditions
at the Straits of Mackinac declined with a four-
year half-life. These rates were applied starting
on January 1,1996.
D) No Atmospheric Deposition - This was the same
as Scenario "A", but starting on January 1,1996,
the atmospheric loads (dry and wet deposition)
were set to zero. All other forcing functions as
observed in the LMMBP period were repeated
throughout the simulation period.
E) No Tributary Loadings - This was the same as
Scenario "A", but starting on January 1,1996, all
tributary loads were set to zero. All other forcing
functions as observed in the LMMBP period were
repeated throughout the simulation period.
F) Lakewide Sediment Cleanup - This was the
same as Scenario "A", but starting on January 1,
1996, the lake-wide sediment PCB concentration
was instantaneously set to zero. All other
sediment properties remained as existed prior to
sediment clean-up. All other forcing functions as
observed in the LMMBP period and processes
were repeated throughout the simulation period.
G) No Atmospheric Deposition and No Tributary
Loadings - The loading cuts of Scenarios "D" and
"E" were combined. All other forcing functions as
observed in the LMMBP period were repeated
throughout the simulation period.
436
-------
The results of the above seven scenarios will be
presented in two separate groups. These are
forecast scenarios (Scenarios A, B, and C) and
sensitivity or engineering scenarios (Scenarios D, E,
F, and G).
4.6.4 Results of the Forecast and
Sensitivity Scenarios and Discussion
Figures 4.6.3 and 4.6.4 show the annual and monthly
average long-term responses of total PCBs in the
water column of the Lake Michigan system for the
seven forecast and sensitivity scenarios. Compared
to the annual lake-wide total PCB concentrations
(Figures 4.6.3a and 4.6.3b), the monthly lake-wide
total PCB concentrations show a much wider
variation with high concentrations in the summer
months and low concentrations in the winter months
(Figures 4.6.4a and 4.6.4b).
The model results were compared to measured data
and water quality criteria. The lake-wide average
total PCB concentration (0.259 ± 0.172 ng/L) for the
LMMBP period (1994-1995) was based on 298 field
measurements from various water depths at 41 water
column sampling stations. The total PCB
concentration (0.165 ± 0.029 ng/L) from the Episodic
Events-Great Lakes Experiment (EEGLE) project
represents the average PCB concentration for
southern Lake Michigan in 2000 (Miller, 2003). The
focus of EEGLE was to investigate the potential
impact of major sediment resuspension events on
persistent organic pollutants (POPs) in southern Lake
Michigan. Field sampling for EEGLE was conducted
in 1998, 1999, and 2000. The average PCB
concentration for 2000 from the EEGLE project was
a better representation than the earlier years for the
open-water concentration of PCBs in the lake and
was used for post-audit comparisons to the results of
the LM2-Toxic PCB long-term forecast and sensitivity
scenarios. USEPA water quality criteria for the
protection of wildlife is 0.12 ng/L and for human
health is 0.026 ng/L; which is a human cancer value
(HCV) that is still under review and development
(U.S. Environmental Protection Agency, 2005; U.S.
Environmental Protection Agency, 1997).
Scenario A - Constant Conditions serves as the
upper bound of the range of possibilities resulting
from the specified PCB forecast and sensitivity
scenarios. The long-term response to the Constant
Conditions Scenario clearly demonstrated that,
during the LMMBP period (1994-1995), the Lake
Michigan system was not at steady-state with respect
to the 1994-1995 loads, vapor phase concentrations,
and the level of sediment total PCB inventory. As the
mass budget analysis indicated in the previous
section and in Table 4.6.2, the mass losses from net
volatilization and sediment burial were the major
contributors to the decline of total PCB
concentrations in both the water column and the
surficial sediment layer. The spatially averaged
steady-state value for the water column under this
scenario was about 0.145 ng/L and was reached
around 2024. This value represents a 44% reduction
in the annual average concentration (0.259 ng/L) in
the water column for the LMMBP period. The
steady-state concentration of the Constant Condition
Scenario will still be approximately 20% higher than
the most recent (U.S. Environmental Protection
Agency, 2005) USEPA water quality criteria for the
protection and wildlife and five to six times higher
than the USEPA water quality criteria for the
protection of human health in the Great Lakes
system (U.S. Environmental Protection Agency,
1997).
Among the forecast scenarios, the outcome from
Scenario A (Constant Conditions) may not be a
realistic prediction of long-term PCB concentrations
in the lake. Because PCB production was phased
out in the 1970s, PCB inputs into the Lake Michigan
system through the atmosphere (via dry and wet
deposition, and absorption of vapor phase) and
tributaries have been decreasing significantly due to
regulatory policies and remediation efforts made by
federal and state agencies. Therefore, it is
reasonable to assume that PCB inputs into the lake
should continue decreasing under current regulatory
policies and clean-up efforts. The decline rates used
in Scenarios B and C for PCB inputs from the
atmosphere and tributaries were the result of
analyzing observed data collected for the past 25
years. These rates were subject to a certain degree
of uncertainty (see Section 1.7.2). The variation of
the estimated decline rates for atmospheric
components (dry and wet deposition and vapor
phase concentration) was quite large with half-lives
ranging from 6 to 20 years. It appears that the rate
of decline decreased with the addition of more recent
data (Hillery et al., 1997; Simcik et al., 1999;
Schneider era/., 2001; Buehler et al., 2002, 2004).
437
-------
0.30-
Constant conditions - scenario A
Continue recovery (slow) - scenario C
Continue recovery (fast) - scenario B
LMMBP data -1994-95; EEGLE data 2000
EPA water quality criteria for protection of
wildlife (2005)
EPA water quality criteria for protection of
human health (1997)
1994 2004 2014 2024 2034 2044 2054
year
Figure 4.6.3a. Annual long-term responses to total PCB concentrations in the water column of Lake
Michigan for the forecast scenarios and USEPA water quality criteria for the protection of wildlife (U.S.
Environmental Protection Agency, 2005) and human health (U.S. Environmental Protection Agency,
2997) in the Great Lakes system.
0.30
0.25
~ 0.20
§ 0.151
c
o
0 0.10-
DO
O
°- 0.05-
Lakewide sediment cleanup - scenario F
No tributary loadings - scenario E
No atmospheric deposition - scenario D
No atmospheric deposition & tributary
loadings - scenario G
LMMBP data -1994-95; EEGLE data 2000
1994
2004
2014
2024
year
2034
2044
2054
Figure 4.6.3b. Annual long-term responses to total PCB concentrations in the water column of Lake
Michigan for the sensitivity scenarios.
438
-------
0.30
~— Constant conditions - scenario A
Continue recovery (slow) - scenario C
Continue recovery (fast) - scenario B
» LMMBP data -1994-95; EEGLE data 2000
— EPA water quality criteria for protection of
wildlife (2005)
EPA water quality criteria for protection of
human health (1997)
0.00 -I
1994
2004
2044
2054
Figure 4.6.4a. Monthly long-term responses to total PCB concentrations in the water column of Lake
Michigan for the forecast scenarios and USEPA water quality criteria for the protection of wildlife (U.S.
Environmental Protection Agency, 2005) and human health (U.S. Environmental Protection Agency,
1997) in the Great Lakes system.
0.30-
0.25
CD
o
C
o
o
CD
O
a.
43
"o
0.20
0.15-
0.10-
0.05-
1994
No atmospheric deposition - scenario D
No tributary loadings - scenario E
Lakewide sediment cleanup - scenario F
No atmospheric deposition & tributary
loadings - scenario G
LMMBP data -1994-95; EEGLE data 2000
2044
2054
Figure 4.6.4b. Monthly long-term responses to total PCB concentrations in the water column of Lake
Michigan for the sensitivity scenarios
439
-------
The purpose of forecast Scenarios B and C was to
provide bounds for predicted long-term PCB water
column concentrations by assuming six-year half-life
and 20-year half-life decline rates in PCB
atmospheric components, respectively.
For Scenario B (Fast Continued Recovery Scenario),
it takes about five years (starting January 1,1996) for
PCB concentrations in the water column to meet the
USEPA water quality criteria for the protection of
wildlife (U.S. Environmental Protection Agency, 2005)
and more than two decades to reach the USEPA
water quality criteria for the protection of human
health (U.S. Environmental Protection Agency, 1997)
in the Lake Michigan system. The water column
PCB concentrations predicted in Scenario C (Slow
Continued Recovery Scenario) declined at a much
slower speed. The model results indicated that it
would take about 12 years for the water column PCB
concentrations in the lake to reach the USEPA water
quality criteria for the protection of wildlife in the
Great Lakes (U.S. Environmental Protection Agency,
2005). Figure 4.6.3a also shows that the PCB
concentrations in the water column reached the
USEPA water quality criteria for the protection of
human health (U.S. Environmental Protection
Agency, 1997) around 2046 (five decades after
1996). In both forecast Scenarios B and C, the PCB
inventory in the surficial active sediment layer plays
an important role in sustaining the water column PCB
concentrations.
It is important to point out that the decline rate used
in Scenarios B and C may not necessarily be realistic
rates for the Great Lakes in the future. There is a
chance that the actual rate could be slower,
especially if there is no further regulatory and
remediation actions taken to reduce the PCB sources
from the atmosphere of the entire Lake Michigan
watershed.
The sensitivity Scenarios D, E, F, and G were
designed for the purpose of demonstrating how
sensitive the long-term responses of the Lake
Michigan system could be by hypothetically
eliminating atmospheric deposition (dry and wet),
tributary loads, total sediment inventory, and total
external loads (dry and wet atmospheric deposition,
and tributary loads altogether), respectively, starting
on January 1, 1996. It is very important to mention
that for these sensitivity scenarios, the PCB vapor
phase concentrations were kept the same as the
LMMBP (1994-1995) measured concentrations.
Long-term PCB concentrations in the water column
were more sensitive to the atmospheric deposition
(dry and wet) than the load from tributaries (Figure
4.6.3b). The steady-state concentration predicted
from Scenario E (No Tributary Loadings) was 0.131
ng/L which was equivalent to less than a 10%
decrease in the steady-state concentration (0.145
ng/L) from Scenario A (Constant Conditions). The
steady-state concentration predicted from Scenario
D (No Atmospheric Deposition) experienced a much
larger drop to 0.094 ng/L, with a 35% reduction
compared to the steady-state concentration of
Scenario A. When eliminating both atmospheric
deposition (wet and dry) and tributary load (Scenario
G), the steady-state concentration decreased to
0.080 ng/L. By eliminating PCB total inventory in the
lake sediments, starting on January 1,1996, the PCB
concentration in the water column experienced a
steep drop initially and then gradually increased and
reached a steady-state concentration of 0.145 ng/L.
Notice that this value was the same as the one
predicted from Scenario A (Constant Condition).
It should be emphasized that Scenarios D, E, F, and
G are hypothetical and not realistic. Because LM2-
Toxic was not coupled with an air quality model to
dynamically compute PCB vapor phase
concentration, the atmospheric concentrations in the
sensitivity scenarios were kept constant as measured
during the 1994-1995 LMMBP period. In reality, the
PCB water column concentrations should be
significantly lower than the steady-state
concentrations resulting from these four sensitivity
scenarios. As demonstrated in Figure 4.6.1, the
gross volatilization flux is the largest flux moving PCB
mass out of the lake. The decrease in the water
column PCB concentrations after the actions taken
for these four sensitivity scenarios would reduce the
gross volatilization fluxes. As a result, the PCB vapor
phase concentration over the watershed should
decrease accordingly. This could lead to the
reduction of PCB absorption flux to the lake and
further reduce the PCB water column concentrations.
Figure 4.6.5 shows the long-term responses of PCB
concentrations in the lake sediments to the seven
forecast and sensitivity scenarios. In general, the
PCB concentrations in the sediments followed
440
-------
Constant conditions - scenario A
Continue recovery (slow) - scenario C
Continue recovery (fast) - scenario B
2044 2054
Figure 4.6.5a. Annual long-term responses to
total PCB concentrations in the surf icial sediment
of Lake Michigan for the forecast scenarios.
O>
'10000-
8000
6000-
4000-
2000
No atmospheric deposition - scenario D
— No tributary loadings - scenario E
132 Lakewide sediment cleanup - scenario F
— No atmospheric deposition & tributary loadings
scenario G
0-1-
1994 2004 2014 2024 2034 2044 2054
Year
Figure 4.6.5b. Annual long-term responses to
total PCB concentrations in the surf icial sediment
of Lake Michigan for the sensitivity scenarios.
similar long-term trends predicted for the water
column. The steady-state PCB sediment
concentration predicted from Scenario E (No
Tributary Loads) was 1,347 ng/L which was 18% less
than the steady-state PCB sediment concentration
(1,665 ng/L) from Scenario A (Constant Conditions).
The steady-state concentration predicted from
Scenario D (No Atmospheric Deposition Wet and
Dry) was 909 ng/L which was 45% less than the
concentration from Scenario A.
A few differences were noted when comparing
responses of PCBs in sediment with the long-term
responses of PCBs in the water column. First, there
was no apparent seasonal variation in the long-term
temporal profiles of the sediment PCB
concentrations. The large PCB inventory and the
mixing processes within the surficial sediment layer
may have been factors smoothing out any seasonal
variations caused by atmospheric components and
tributary loads. Secondly, the long-term responses
of PCB concentrations in sediment were more
sensitive than those in the water column for the
sensitivity Scenarios D (No Tributary Loads), E (No
Atmospheric Depositions), and G (No Tributary and
Atmospheric Loads). LM2-Toxic was developed for
simulating 54 PCB congeners. Based on the data
analysis results for PCBs in different media (see Part
1.6 for details), the heavier PCB congeners (higher
molecular weight) were more abundant in tributary
loads and atmospheric deposition (dry and wet) than
in the vapor phase. In the water column, more than
two-thirds of PCB mass was in the dissolved phase,
and PCB concentrations were significantly influenced
by PCB vapor phase concentration. Unlike in the
water column, particulate PCBs in the lake sediments
were the dominant phase, while PCBs in the
dissolved phase were negligible. In addition, the
heavier PCB congeners were more abundant in the
sediments due to their higher partitioning coefficients.
The larger percentage of heavy congeners in
tributary loads and atmospheric deposition and the
strong influence of vapor PCB concentrations on the
water column PCB concentrations might be the
factors making the PCB concentrations in sediment
more sensitive to the tributary loads and atmospheric
deposition than in the water column.
Reducing the PCB vapor phase concentration was
critical to the level of long-term PCB concentration in
the Lake Michigan system. Both Figures 4.6.3 and
4.6.5 (Scenarios B and C compared to Scenario A)
demonstrate that the long-term response of PCBs in
the lake system is very sensitive to PCB vapor
concentrations. Figure 4.6.1 also shows that the
volatilization flux and absorption flux are the number
one and the number two fluxes for Lake Michigan.
Again, the future declining rates for PCB atmospheric
components (including vapor concentration, wet and
dry deposition) may be slower than the rate used in
Scenario C if there is no action taken in the future to
continue reducing PCB vapor concentrations and
441
-------
particulate deposition from the atmosphere of the
Lake Michigan watershed.
4.5.5 Results Provided for the LM Food
Chain Model
The exposure concentrations generated from LM2-
Toxic for the seven scenarios were provided as
forcing functions to the LM Food Chain model to
predict the long-term concentration changes for PCB
congeners in lake trout tissue. Sets of exposure
concentrations were generated for the Saugatuck
biota box, segment 2, the Sturgeon Bay biota box,
and segment 3 (see Figure 5.4.1 in Part 5 LM Food
Chain for the locations of the biota zones). Each
data set contained water column dissolved PCB
concentrations (ng/L), water column particulate PCB
concentrations (ng/g organic carbon), sediment
dissolved PCB concentrations (ng/L), and sediment
particulate PCB concentrations (ng/g organic
carbon). Because there were multiple LM2-Toxic
sediment segments under each biota box and water
column segment, the sediment PCB concentrations
provided to the LM Food Chain were computed using
area-weighted averaging based on the segment-
specific concentrations generated by the LM2-Toxic.
Field data for water column hypolimnetic total
particulate PCB concentrations for the Saugatuck
biota box was higher than for the much larger
hypolimnetic level 2 segments (Segments 21,30,37)
within which the Saugatuck biota box resided.
Therefore, a factor of 1.5 was used to scale
hypolimnetic water column total particulate PCB
concentrations from the larger level 2 segments
(Segments 21,30,37) to the hypolimnetic Saugatuck
biota box. This factor was calculated based upon the
LMMBP data collected from hypolimnetic sampling
locations near the biota box and sampling locations
within hypolimnetic segments 21, 30, and 37 as a
whole (Table 4.6.3).
References
Buehler, S.S., I. Basu, and R.A. Hites. 2002. Gas-
Phase Polychlorinated Biphenyl and
Hexachlorocyclohexane Concentrations Near the
Great Lakes: A Historical Perspective. Environ.
Sci. Technol., 36(23):5051-5056.
Buehler, S.S., I. Basu, and R.A. Hites. 2004.
Causes of Variability in Pesticide and PCB
Concentrations in Air Near the Great Lakes.
Environ. Sci. Technol., 38(2):414-422.
DePinto, J.V., R. Raghunathan, P. Sierzenga, X.
Zhang, V.J. Bierman, Jr., P.W. Rodgers, andT.C.
Young. 1993. Recalibration of GBTOX: An
Integrated Exposure Model for Toxic Chemicals
in Green Bay, Lake Michigan. Final Report. U.S.
Environmental Protection Agency, Office of
Research and Development, ERL-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
132pp.
Table 4.6.3. Mean and Median Particulate PCBs/Organic Carbon and Field Data and Scaling Factor for
Hypolimnetic Level 2, Segments 21, 30, and 37, and for Saugatuck Biota Box Hypolimnion
Field Data
Segments 20, 29, 36
(Hypolimnion)
Saugatuck Biota
Box, Hypolimnion Factor
Mean Particulate PCBs/Organic Carbon
(ngPCB/mgC)
Median Particulate PCBs/Organic Carbon
(ngPCB/mgC)
0.560
0.509
0.817
0.747
1.5
1.5
442
-------
Endicott, D.D. 2005. 2002 Lake Michigan Mass
Balance Project: Modeling Total PCBs Using the
MICHTOX Model. In: R. Rossmann (Ed.),
MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan, Part 2. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse He, Michigan.
EPA/600/R-05/158, 140 pp.
Hillery, B.L, I. Basu, C.W. Sweet, and R.A. Hites.
1997. Temporal and Spatial Trends in a Long-
Term Study of Gas-Phase PCB Concentrations
Near the Great Lakes. Environ. Sci. Technol.,
Marti, E.A. and D.E. Armstrong. 1990.
Polychlorinated Biphenyls in Lake Michigan
Tributaries. J. Great Lakes Res., 16(3):396-405.
Miller, S.M. 2003. The Effects of Large-Scale
Episodic Sediment Resuspension on Persistent
Organic Pollutants in Southern Lake Michigan.
Ph.D. Thesis, The University of Iowa, Iowa City,
Iowa. 194pp.
Ortiz, E., R.G. Luthy, D.A. Dzombak, and J.R. Smith.
2004. Release of Polychlorinated Biphenyls
From River Sediment to Water Under Low-Flow
Conditions: Laboratory Assessment. J. Environ.
Engin., 130(2):126-135.
Schneider, A.R., H.M. Stapleton, J. Cornwell, and
J.E. Baker. 2001. Recent Declines in PAH,
PCB, and Toxaphene Levels in the Northern
Great Lakes as Determined From High
Resolution Sediment Cores. Environ. Sci.
Technol., 35(19):3809-3815.
Simcik, M.F., I. Basu, C.W. Sweet, and R.A. Hites.
1999. Temperature Dependence and Temporal
Trends of Polychlorinated Biphenyl Congeners in
the Great Lakes Atmosphere. Environ. Sci.
Technol., 33(12)1991-1995.
U.S. Environmental Protection Agency. 1997.
Revocation of the Polychlorinated Biphenyl
Human Health Criteria in the Water Quality
Guidance for the Great Lakes System. Federal
Register, October 9, 1997, Volume 62, Number
196. [DOCID:fr09oc97-9]. From the Federal
Register Online via GPO Access
[wais.access.gpo.gov].
U.S. Environmental Protection Agency. 2005. Water
Quality Guidance for the Great Lakes System.
Code of Federal Regulations, Title 40, Volume
21, Chapter 1, Part 132. Http://www.access.
gpo.gov/nara/.
443
-------
PART 4
LM2-TOXIC
Chapter 7. LM2-Toxic Sensitivity Analysis
Sensitivity analysis is a very efficient tool that can
semi-quantitatively demonstrate the uncertainties in
the outputs from a water quality model. These
uncertainties in the model outputs could result from
the uncertainties associated with the forcing time
functions; the water and sediment transport; the
numerical algorithms used in the model; or the
parameters for the chemical and biochemical
processes defined in the model. This chapter
summarizes the results of a series of sensitivity
analyses completed only for the forcing time
functions such as primary production and
polychlorinated biphenyl (PCB) loads. The PCB
atmospheric components (PCB vapor concentration,
dry and wet deposition) and tributary PCB load half-
lives are discussed in Part 4, Chapters 3 and 6.
Model sensitivity analyses were performed over both
a short-term (two years, 1994-1995) and long-term
(62 years -1994 to 2055) periods.
4.7.7 Primary Production Sensitivity
Due to the affinity of PCBs for organic carbon, the
internal organic carbon load (primary production) is
very important in understanding the transport and
fate of PCBs in the Lake Michigan system. The
internal organic carbon load used in the LM2-Toxic
was generated by the eutrophication model, LM3-
Eutro (See Part 2 for details). Because there are
uncertainties associated with the LM3-Eutro
generated internal organic carbon load, the variations
in the load on the LM2-Toxic PCB model output
concentrations (including solids dissolved organic
carbon (DOC), biotic carbon (BIG), paniculate
detrital carbon (PDC), and PCBs) were explored.
The results of the sensitivity analysis for PCBs are
illustrated with the PCB28+31 congener pair because it
is the most abundant PCB congener pair in Lake
Michigan. In LM2-Toxic, this congener pair is
modeled as a single state variable. The internal
organic carbon load generated from the LM3-Eutro
for the LMMBP period (1994-1995) was increased
50% for one analysis and decreased 50% for a
second analysis. The model simulations for the
analyses were conducted for both a short-term (two-
year period: 1994 and 1995) period and a long-term
(62-year period: 1994-2055) periods. The results
from the sensitivity analysis were compared to the
results from the LM2-Toxic model base runs (i.e.,
calibration run for the 1994-1995 period and long-
term Constant Condition Scenario, see Part 4,
Chapters 4 and 6 for detailed descriptions of both of
these base runs).
Below is a summary of the results from the sensitivity
analysis:
1. As shown in Figures 4.7.1 and 4.7.2, a 50%
increase or decrease in the primary production
has a noticeable effect on the solid
concentrations (DOC, BIG, PDC) in the water
column compared to the base run concentrations
for both short-term and long-term simulations.
Table 4.7.1 lists annual average concentrations
for the water column carbon solids and the
percentage change of the water column carbon
solids concentrations due to the increase and
decrease in the primary production for both short-
term and long-term simulations. Primary
production has significant and almost instant
444
-------
3- 2.00
~5>
E 1.60
.2o
•S >
£ « 1 20
c w
is
u 3
§| 0.80
o n
O
O 0.40
O
— increased primary production by 50%
— base run - constant condition
decreased primary production by 50%
Jan-94 May-94 Aug-94 Dec-94 Apr-95 Aug-95 Dec-95
~ 0.07-
jfa* 0.06-
i-DJ
o S 0.05-
ns ro
i
-------
DOC concentration (mg/L)
lakewide average
oooo^-'-^-'--^
KJ 4^ cr> co b K) ji. b) be
3OOOOOOOOC
^1
V_
. . . r-on/
increased primary production t>y oU%
•~«-3 decreased primary production by 50%
1994 2004 2014 2024 2034 2044 2054
year
BIC concentration (mg/L)
lakewide average
ppppopppc
b b b b b b b b c
0^->-lsJ|sJWW4i4:
DUiocnocnocnoc
— — —
1994 2004 2014 2024 2034 2044 2054
year
n 1 °
ra 0.16-
£o>
•^•ra 0.14-
•B | 0.12-
fQ on
•£« 0.10-
05 0.08-
c 5
8m 0.06-
^~ 0.04-
°- 0.02-
n-
^__.,
1994 2004 2014 2024 2034 2044 2054
year
Figure 4.7.2. Long-term (1994-2055) variations of lake-wide (Green Bay included) organic carbon
concentrations for ± 50% primary production changes without adjusting settling and resuspension
rates.
446
-------
Table 4.7.1. Annual Average Concentrations of Water Column Carbon Solids and Annual Average
Change in Percentage for Water Column Carbon Solids Concentrations Resulting From the LM2-Toxic
Model Runs for Both the Short-Term (1994-1995) and the Long-Term (1994-2055) Simulations With 50%
Increase and 50% Decrease of the LM3-Eutro Produced Primary Production
Carbon Solids
(Short-Term
Simulations)
DOC
BIG
PDC
Concentration
(mg/L) From
Original Base
1.48
0.034
0.129
Concentration
(mg/L) From
the Run With
50% Reduction
1.42
0.025
0.095
Annual
Average
Percentage
(%) Change
-4
-27
-26
Concentration
(mg/L) From
the Run With
50% Increase
1.55
0.042
0.159
Annual
Average
Percentage
(%) Change
5
23
23
Carbon Solids (Long-Term Simulations)
DOC
BIG
PDC
1.23
0.034
0.129
0.96
0.025
0.088
-22
-27
-32
1.47
0.042
0.162
20
23
26
impact on the level of BIG and PDC concentration
in the water column. Because of much larger
initial DOC inventory in the lake and slower
degradation process from PDC to DOC, the
impact on DOC concentration in the lake due to
the changes of primary production was not as
evident as on BIG and PDC in the early portion of
the long-term simulation period.
2. Figure 4.7.3 shows that, compared with the base
run, there is very little difference in the PCB water
column concentrations generated from the runs
with 50% decrease and a 50% increase in
primary production for the short-term simulation.
Table 4.7.2 provides detailed information on
individual mass fluxes in Lake Michigan and PCB
inventories of both water column and surficial
sediment for the two-year (1994-1995) sensitivity
analyses. The decrease and increase in the
settling PCB mass flux due to the 50% decrease
and the 50% increase in primary production were
compensated by the increase and decrease in
the gross volatilization mass flux, respectively.
This keeps the PCB water column inventories
predicted from these two runs very close to the
inventory generated from the short-term base
runs.
3. The long-term, steady-state PCB concentrations
in the water column was not significantly different
from the base run concentrations (Figure 4.7.4).
Interestingly, PCB concentrations from the long-
term sensitivity runs started out almost identical
and deviated with each other toward the end of
the simulation. Table 4.7.3 provides detailed
information on individual PCB mass fluxes in
Lake Michigan and inventories of both the water
column and the surficial sediment for the last two
years of the long-term (62-years) model
simulation. Similar to the short-term simulation,
PCB settling and net volatilization fluxes were
affected the most by the increase and the
decrease of the primary production. The
increase or decrease of the settling fluxes due to
the increase or decrease in primary production
was countered by the decrease or increase of the
net volatilization fluxes.
In conclusion, the sensitivity analyses illustrate that,
under the 1994-1995 PCB loading/boundary
conditions/other forcing functions, a 50% increase or
decrease in primary production generated from the
LM3-Eutro does not have a significant influence on
PCB concentrations in Lake Michigan for both short-
447
-------
J 0.020
Ul
5.016-
•££0.0121
-------
;J 0.020
O)
c
§0)0.016-1
£ I 0.012-1
o re
o a,
g 1 0.008-
to ~ 0.004-
m
2 o
i
increase primary production by 50%
base run - constant condition
decrease primary production by 50%
1994 2004
2014
2024
year
2034 2044 2054
Figure 4.7.4. Long-term (1994-2055) variations of lake-wide (Green Bay included) PCB28+31 (dissolved
+ particulate) concentrations for ± 50% primary production changes without adjusting settling and
resuspension rates.
Table 4.7.3. PCB28+31 Mass Fluxes and Inventories for Lake Michigan System Results From the LM2-
Toxic Sensitivity Analysis on Primary Production For the Last Two Years of the Long-Term (62-Year
Period: 1994-2055) Simulations
PCB Mass Fluxes, kg/(Two Years) and
Inventories, kg
Original Base Run 50% Reduction 50% Increase
Loads
Settling
Resuspension
Burial
Water Column Inventory
Sediment Inventory
Diffusion
Absorption
Gross Volatilization
Net Volatilization
Export to Lake Huron
Chicago Diversion
113.97
43.80
25.17
21.98
33.78
130.74
0.52
263.07
355.33
92.26
0.69
0.07
113.97
32.47
23.42
22.39
31.21
133.13
0.70
263.07
365.30
102.23
0.58
0.06
113.97
51.94
24.68
21.81
35.71
129.71
0.46
263.07
346.47
83.40
0.77
0.07
Mass Gain/Loss in Water Column
2.84
2.73
2.93
449
-------
term and long-term forecast scenarios. Table 4.7.4
lists the model-generated average PCB inventories
resulting from the sensitivity runs for both the water
column and the surficial sediment layer and the
percentage change of the average PCB inventories
in both compartments due to the increase and
decrease in primary production. The inventories are
the average PCB inventories for the two-year period
of the short-term (1994-1995) simulation and the last
two-year period of the long-term simulation. The
largest percentage change of PCB inventories due to
the 50% increase or decrease in primary production
goes to the PCB inventory in the water column at the
end of the long-term simulation period and is in the
range of 5-8%. The impact on PCB inventory of the
surficial sediment layer due to the changes of primary
production is very small for both the short-term and
the long-term periods.
4.7.2 PCB Loads Sensitivity
The variation in the outcomes of PCB concentrations
in both the water column and sediment from the LM2-
Toxic model can be very significant due to the
uncertainty of PCB loads used as input. The
uncertainty of the LMMBP-generated PCB loads
could be due to sampling approach, analytical
method, interpolation algorithm used for estimating
the loads, and loads that were missed or not
considered.
There is evidence (Wethington and Hornbuckle,
2005) that an additional input of PCBs was
contributed from the local Milwaukee atmosphere
through vapor-water exchange and wet and dry
deposition to Lake Michigan that were not accounted
for in the LM2-Toxic model. The combined additional
PCB source from the Milwaukee regional atmosphere
was estimated to be at least 120 kg per year
(Wethington and Hornbuckle, 2005). It is possible
that loads in other areas of the basin could have
been missed, such as Green Bay.
Another potential unaccounted PCB source to the
lake is the load associated with very large
atmospheric particles. These are particles with a
diameter greater than 10 urn and settling velocities
greater than 7.4 cm s'1. Although there is some
disagreement among experts in the field regarding
the magnitude of PCB loads to the lake via the large
particles, studies indicate that PCB dry deposition
associated with large particles could be a significant
PCB source to the lake (Miller era/., 2001; Franz et
al., 1998; Holsen et al., 1991). The annual PCB
inputs from the atmosphere through the coarse
particles could be in a range of 320 kg/year to 5,500
kg/year (data provided from the LMMBP atmospheric
working group; Wethington and Hornbuckle, 2005;
Franz era/., 1998; and Holsen et al. 1991) during the
period of 1989-1995. However, the science and
technique is not well-developed enough to make
reliable over-lake estimates of these fluxes. Much of
the uncertainty in measuring large particle fluxes
comes from the difficulty in quantifying how far these
large particles travel from their source to the lake.
Model runs were designed to evaluate the impact of
potential missing loads on the model outputs. A
sensitivity analysis was completed for the Milwaukee
load by adding 120 kg/year of PCB load into segment
1 in the LM2-Toxic model. Additional simulations
were run to gain insight into how the model would
respond to increasing the total PCB load (tributary
load + atmospheric load) by 50% and 100%. The
results from the sensitivity analysis were then
compared with the LM2-Toxic model long-term (62
years) base run results (Figure 4.7.5). Compared to
the steady-state concentration from the long-term
base run, the simulation showed total PCBs in the
water column increased less than 5% for the
suggested 120 kg/year missing PCB source from the
Milwaukee atmosphere. An increase of 15% and
30% was found for the simulations where the PCB
load was increased by 50% and 100%, respectively.
The amount of increase in total PCB concentrations
was much less during the first five years of the
simulation than during the steady-state period. This
indicates that, under the current conditions in the
Lake Michigan system, the LMMBP-generated PCB
loads were not the dominant PCB flux controlling the
concentration of PCBs in the lake. When the load
was doubled, the PCB concentrations in the water
column in the first few years only increased about
15%. Part 4, Chapter 6 provides quantitative
analyses and in-depth discussions on the critical
sources and sinks and important environmental
processes for PCBs in Lake Michigan.
450
-------
Table 4.7.4. PCB28+31 Average Inventories of Water Column and Surficial Sediment Results From the
LM2-Toxic Simulations for the Primary Production Sensitivity Analysis, and Changes in Percentage
for These Inventories Compared to the Inventories From the Original Base Runs
Simulations)
Water Column
Surficial Sediment
PCBs (Long-
Term
Simulations)
Water Column
Surficial Sediment
Inventory
Original Base
70.79
808.76
33.78
130.74
Inventory (kg) From
Reduction in
Primary Production
69.67
797.63
31.21
133.13
Percent (%)
Change
-1.6
-1.4
-7.6
1.8
Inventory (kg) From
the Run With 50%
Increase in Primary
Production
71.41
819.53
35.71
129.71
Percentage
(%) Change
0.88
1.3
5.7
-0.79
0.30
^0.25-
w 0.20-
03
8 o.isH
c
o
o
CD 0.1 OH
O
0.
I 0.05 -\
100% more total PCB load
50% more total PCB load
120 kg/yr more atmospheric load from Milwaukee
base run - constant condition
LMMBP data -1994-95; EEGLE data 2000
•<- '-} v V V V V" V" V* V"
0
1994
2004
2014
2024
year
2034
2044 2054
Figure 4.7.5. Sensitivity analysis of the LM2-Toxic predictions to varying PCB loads. Note: LMMBP
data - 1994-1995 = 0.259 ± 0.172 ng/L; EEGLE data 2000 = 0.165 ± 0.029 ng/L. The model output
concentrations and field data in this graph are lake-wide average concentrations.
451
-------
References
Franz, T.P., SJ. Eisenreich, and T.M. Holsen. 1998.
Dry Deposition of Paniculate Polychlorinated
Biphenyls and Polycyclic Aromatic Hydrocarbons
to Lake Michigan. Environ. Sci. Technol.,
32(23):3681 -3688.
Holsen, T.M., K.E. Noll, S. Liu, and W. Lee. 1991.
Dry Deposition' of Polychlorinated Biphenyls in
Urban Areas. Environ. Sci. Technol., 25(6):1075-
1081.
Miller, S.M., M.L. Green, J.V. DePinto, and K.C.
Hornbuckle. 2001. Results From the Lake
Michigan Mass Balance Study: Concentrations
and Fluxes of Atmospheric Polychlorinated
Biphenyls and frans-Nonachlor. Environ. Sci
Technol., 35(2):278-285.
Wethington, D.M. and K.C. Hornbuckle. 2005.
Milwaukee, Wl, as a Source of Atmospheric
PCBs to Lake Michigan. Environ. Sci. Technol
39(1):57-63.
452
-------
PARTS
LM FOOD CHAIN
Xin Zhang
Welso Federal Services, LLC
Large Lakes Research Station
9311 Groh Road
Grosse lie, Michigan 48138
Chapter 1. Executive Summary
This part of the report documents the food web
bioaccumulation model developed for the Lake
Michigan Mass Balance Project (LMMBP). The
model established dynamic relationships between
polychlorinated biphenyl (PCB) concentrations in the
exposure environments and result PCB levels in the
fish food webs of Lake Michigan. The primary
objective of this work was to provide practical
modeling tools to predict toxic PCB levels in lake
trout and coho salmon in response to projected water
quality improvements for the lake.
The model was based upon available theory and data
characterizing the bioaccumulation of toxic chemicals
in fish and other aquatic organisms. A detailed
description of the model development is provided in
the document. Extensive collections of data on lake
trout and coho salmon food webs in Lake Michigan
were conducted to facilitate refinements of model
parameters to site-specific conditions. Forty
congeners or congener groups of PCBs were
targeted for the model calibration or parameter
refinements. These PCB congeners represented
toxic chemicals covering a wide range of
hydrophobicity.
The food web model was calibrated with PCB data
collected in 1994 and 1995 for three lake trout food
webs at Sturgeon Bay, Sheboygan Reef, and
Saugatuck. The lake trout sub-populations in these
three biota zones were believed to be appropriate
representations of lake trout in Lake Michigan.
Model calibration was also performed for a lake-wide
coho salmon food web. During the model calibration,
model parameters were refined to achieve an
adequate agreement between model calculationsand
observed PCB data for a food web. In this study, the
focus of model calibration was not limited to top
predators nor to toxics with certain hydrophobicity.
The model parameters were systematically optimized
for all species at various trophic levels and for PCB
congeners of a wide range of hydrophobicity. Extra
care was taken to ensure the refined parameter
values were consistent with the hydrophobicity of
individual PCB congeners and with the trophic
position of individual species. Satisfactory calibration
results were achieved for the lake trout food webs at
Sturgeon Bay and Saugatuck. Although no formal
validation was possible due to additional
requirements of large amount of PCB congener-
specific field data, the calibrated food web models for
Sturgeon Bay and Saugatuck were confirmed in
some degree by the identical values of calibrated
parameters between these two models.
453
-------
The availability of a complete account of observed
data for each food web made this model calibration
probably the most thorough process among similar
efforts. Although PCB concentrations in lake trout or
coho salmon was the endpoint of the model
computation and the focus of most model
applications, we believe that the food web model with
parameters "fine-tuned" for species at all trophic
levels can be used to target any desirable species in
the food web with a high degree of confidence. Also,
the food web model was capable of assimilating
toxics with various hydrophobicities. In fact, no food
web model intended to simulate as many toxic
chemicals with diverse hydrophobicity has been
previously developed.
We believe that the food web model capable of
simulating congener-based PCB dynamics in fish
food webs provided a useful tool for the development
of more effective load reduction plans or total
maximum daily loads (TMDLs) targeted to priority
PCB congeners, instead of traditional category-based
load reduction plans targeted to various contaminant
sources. The results of the PCB congener-based
model simulations also help to better understand
toxic chemical behavior in food webs.
The calibrated food web model was used to perform
several model simulations for PCBs in lake trout food
webs at Sturgeon Bay and Saugatuck. These model
simulations depicted dynamic responses of individual
PCB congeners in the food webs to different PCB
exposure input. Hypothetical long-term PCB
exposure scenarios for the food webs at the
Sturgeon Bay and Saugatuck biota zones were
generated by the water quality model LM2-Toxic
corresponding to different management choices for
the reduction of PCBs into the Lake Michigan
ecosystem.
For each long-term PCB exposure input, similar
model responses were observed for these two biota
zones. As an example, the temporal responses of
individual PCB congener-based concentrations in
adult lake trout at Sturgeon Bay associated with
constant external PCB loadings are presented for
discussion. The expected total PCB concentrations
in adult lake trout at Saugatuck in response to
various hypothetical PCB exposure inputs are also
presented. Given the exposure PCB concentration
time functions provided by the LM2-Toxic, the model
simulations suggested that without further
intervention, the total PCB concentrations in adult
lake trout (5.5 year-old) was expected to reach the
target level of 0.075 ppm in 2026 for Saugatuck biota
zone and in 2032 for Sturgeon Bay biota zone.
454
-------
PARTS
LM FOOD CHAIN
Chapter 2. Recommendations
Through the course of the development and
calibration of the food web models, every effort was
made to make full use of available information and to
make necessary simplifications and assumptions to
achieve a balanced progress without sacrificing the
overall quality in the food web modeling. The models
described in this part was believed to represent the
best available knowledge for the management of
congener-specific PCB contaminants in Lake
Michigan food webs. However, additional works are
needed to further improve the validation and to test
the applicability of the models to other hydrophobic
chemicals. There are also research needs that we
believe are essential for the improvement of the
performance of the fish models.
5.2.7 Additional Model Validations
Additional model validation should be conducted
once sufficient new field data sets become available.
One of the additional model validation would be to
test the validity of the models for PCB concentrations
in Lake Michigan food webs after 1994-1995. This
exercise would include calculating PCB
concentrations in fish using the new field data
collections of PCB concentrations in water and
sediment, comparing the model calculated data with
actual measured PCB concentrations in fish, and
evaluating model performance in reproducing future
PCB concentrations in Lake Michigan fish.
Apparently, model validation is a continuous process.
Each validation exercise only addresses model
performance under a set of specific conditions or to
a type of application. For example, the
aforementioned validation exercise, once achieved,
indicates that the models can be used as a practical
tool to make quantitative prognoses about PCB
contaminants in particular Lake Michigan food webs.
In order to test models' applicability to other
chemicals, model validation can also be extended to
frans-nonachlor in Lake Michigan. There are
adequate field data sets available to carry out this
validation study. The values of chemical-specific
model parameters for frans-nonachlor can be
deduced from the correlations between parameter
values and log Kow values of the contaminants, which
were established based on model calibrations for
PCB congeners.
5.2.2 Model Applications
It is desirable to examine the capability of the fish
models to reproduce archived fish PCB data for Lake
Michigan. This can be done with the reconstruction
of historical PCB concentrations in water and
sediments by a fate and transport model (such as the
LM2-Toxic) based on dated sediment core profiles for
PCBs. Model estimates of past PCB concentrations
in various fish species can then be made from the
reconstructed concentrations in water and sediments.
Once completed satisfactorily, the results of this
model study may provide a better baseline for PCB
load reduction analysis.
455
-------
5.2.3 Future Improvements
The following were some of the research needs
identified to be important for the improvement of the
model quality.
A. Additional studies for the refinement of
bioenergetic information and growth data for
species in lower trophic levels of the food webs,
such as zooplankton, Mysis, deepwater sculpin,
and slimy sculpin in Lake Michigan.
B. Field measurement of the moisture contents of
zooplankton, Mysis, and Diporeia at each biota
zone in Lake Michigan. These data will help
reduce model uncertainty associated with
erroneous estimates of water content while
converting dry weight-based data to wet weight-
based.
C. More studies on correlations between the
chemical assimilation efficiency and Kow values of
the chemicals, and on species dependency of the
chemical assimilation efficiency.
D. Investigation of possible PCB metabolism in Lake
Michigan fish, and the kinetics of the processes.
E. More data collections for possible refinement of
the model descriptions for fish dietary
compositions in Lake Michigan.
456
-------
PARTS
LM FOOD CHAIN
Chapter 3. Model Description
A food web bioaccumulation model usually consists
of a series of submodels characterizing chemical
bioaccumulation in each of the aquatic organisms in
the food web. The individual submodels are linked
together through feeding interactions among the
organisms. The food web model (LM Food Chain)
constructed for the Lake Michigan Mass Balance
Project (LMMBP) is based on Version 5.2 of the
computer model FDCHAIN which was originally
developed by HydroQual, Inc., Mahwah, New Jersey.
The original model and its early versions have been
previously applied in numerous projects including the
Green Bay Mass Balance Food Chain Modeling
project (Connolly et al., 1992). Several additions
and modifications have been made to enhance the
performance of the food web model. They include
the introduction of a multi-compartment approach to
better accommodate the spatially variable conditions
in Lake Michigan, refinements of certain parameters
to reflect advancements of knowledge in related
disciplines, the incorporation of a new submodel for
chemical bioaccumulation of benthic invertebrates,
and the integration of alternative modeling equations
for species-specific parameters that are not readily
obtainable. The following is a detailed description of
submodels used in the LM Food Chain for simulating
organic chemicals in individual fish and in organisms
of lower trophic levels of fish food webs.
5.3.1 Chemical Bioaccumulation in Fish
The model is a set of equations derived using the
principle of mass conservation. It is generally
accepted that the primary processes of chemical
exchange between a fish and its exposure
environment are: 1) chemical uptake from water, 2)
chemical uptake from food sources, 3) chemical
elimination due to respiration and excretion, and 4)
chemical concentration reduction by growth dilution
(Figure 5.3.1). The submodel for chemical
bioaccumulation in fish can then be derived based on
a simple mass balance equation for chemicals in the
fish. The general form of the mass balance equation
is well-defined. The rate of change in chemical
concentration in a fish (dCF/dt) is equal to the sum of
the relevant chemical fluxes into and out of the fish.
dCF/dt = F
where
w
Fp - F0 - Fg
(5.3.1)
dCp/dt = chemical increment in fish per unit time
(ug/kg/day)
Fw = flux of chemical uptake from water
(ug/kg/day)
Fp = flux of chemical uptake from prey items
(Mg/kg/day)
Fe - flux of chemical elimination via respiration
(ug/kg/day)
Fg = flux of chemical reduction by growth
dilution (ug/kg/day)
In some cases, other chemical fluxes, such as flux
associated with the chemical elimination through
metabolism transformation in the organism, may also
need to be included in Equation 5.3.1. In this study,
457
-------
dilution by growth
uptake from prey
\
F
elimination
uptake from water (via respiration and excretion)
Figure 5.3.1. Primary chemical exchange processes between a fish and its environment.
we assumed that metabolism transformation of
polychlorinated biphenyl (PCB) contaminants was
negligible (Gobas, 1993; Stapleton etal., 2001; U.S.
Environmental Protection Agency, 1999).
After construction of the mathematical description for
each of the chemical fluxes in the mass balance
equation, the chemical concentration in the fish CF
(ug-chem/kg-body) at time t + At was then calculated
by numerical integration:
= CF(0 + [dCF(t)/dt]
(5.3.2)
To predict chemical bioaccumulation for top predator
fish, the mass balance equation was repeatedly
applied to organisms at each trophic level to simulate
chemical biomagnification from forage species to top
predators.
Several methods have been developed to describe
chemical dynamics in fish and to estimate related
chemical fluxes in the mass balance equation. In this
food web bioaccumulation model, the chemical
dynamics were described based on fish bioenergetics
(Lantry and Stewart, 1993; Rudstam, 1989; Rudstam
ef a/., 1994; Stewart et ai, 1983; Stewart and
Binkowski, 1986). The mathematical equation used
to estimate the chemical fluxes in the mass balance
equation are described below.
5.3.1.1 Chemical Uptake From Water
The chemical flux entering an organism from water
via gill ventilation (Fw) is expressed as a product of
the fish's ventilation rate and the dissolved chemical
concentration in water. The extent to which
chemicals that enter the gill compartment by gill
ventilation and are actually absorbed by the fish is
usually expressed by the chemical gill transfer
coefficient, Ec, which is included in Fw.
= E. • AC.
(5.3.3)
where
Ec = chemical gill transfer coefficient
Kv = gill ventilation rate (L-water/kg-fish/day)
Cw - dissolved chemical concentration in water
(ug-chem/L-water)
The gill ventilation rate of a fish (Kv) is dependent on
the amount of oxygen required by the fish to sustain
its normal respiration (R0) and the oxygen content in
the water that passes through the gill membrane.
• [O2]) (5.3.4)
458
-------
where
R0 = rate of oxygen uptake from water, or fish
respiration rate (mg-O2/kg-fish/day)
E0 = oxygen gill transfer coefficient
[OJ = oxygen content in water (mg-O2/L-water)
Similarly, the oxygen gill transfer coefficient, E0
reflects the extent to which oxygen that enters the gill
compartment by gill ventilation is actually absorbed
by the fish. The value of R0, which is expressed in
terms of oxygen consumption, can usually be
calculated using a bioenergetics model (Hewett and
Johnson, 1989). Oxygen content in water [O2] was
estimated as a function of water temperature based
on an empirical equation for oxygen saturation in
water (Greenberg etal., 1992).
Substituting Equation 5.3.4 into Equation 5.3.3, the
chemical flux via gill uptake from water (FJ then
follows as
Fw = (EC/E0) • (R0/[02]) • Cw
5.3.1.2 Chemical Uptake From Prey
(5.3.5)
The chemical flux absorbed by fish from diet (Fp) via
the gastrointestinal tract is expressed using the food
ingestion rate of the fish (Kf) and chemical
concentration in its diet (Cp). The extent to which
chemicals in the diet are actually absorbed by the
fish can be expressed by the chemical assimilation
efficiency a, which is included in F .
FP =
where
(5.3.6)
Of = chemical assimilation efficiency
K, = food ingestion rate (g-prey/g-body/day)
Cp = chemical concentration in prey (ug-chem/g-
food)
The chemical concentration in the diet (Cp) is based
on diet composition and chemical content in each
prey component. The food ingestion rate is
determined by an energy balance. The energy intake
from food sources is equal to the energy expenditure
of the fish for respiration and growth:
(K, • Dp) • p = R • DF + G • DF (5.3.7)
where
Dp = energy density of prey (kJ/kg-prey)
DF = energy density of the fish (kJ/kg-body)
R = fish respiration rate (kg-fish/kg-body/day)
G = fish growth rate (kg-fish/kg-body/day)
/? = fraction of ingested energy that is assimilated
R can usually be calculated using a fish bioenergetics
model (Hewett and Johnson, 1989). G can be
estimated by individual fish weight-age relationships.
The energy density (DF and Dp) can be estimated
from the lipid and protein content of the fish and prey.
Substituting Equation 5.3.7 into Equation 5.3.6, the
flux of chemical uptake v/afood consumption, Fp, can
be formulated as follows:
(DF/Dp) • (fl+G) • Cp
5.3.1.3 Chemical Elimination Via Gills
The flux of chemicals eliminated by a fish via the gills
is expressed as a product of gill elimination rate
constant, Ke, and chemical concentrations in the
organism, CF:
CF
(5.3.9)
where
Ke = gill elimination rate constant (1/day)
CF = chemical concentration in organism (ug-
chem/kg-body)
Because the elimination is, in essence, the reverse
process of gill uptake, the gill elimination rate
constant can be related to the gill uptake rate
constant. If we view the ratio of gill uptake and
elimination rate constants as the chemical partition
coefficient between the body tissue and aqueous
459
-------
phases of the organism, the gill elimination rate
constant can then be derived as
fL • n) (5.3.10)
where
Ec = chemical gill transfer coefficient
Kv = gill ventilation rate (L-water/kg-body/day)
p = aqueous phase density of the organism (kg/L)
fa = non-lipid fraction of the fish
fL = lipid fraction of the fish
n = chemical partition coefficient between lipid
and non-lipid phases of the organism
Substitution of Equation 5.3.10 into Equation 5.3.9
yields an equation for estimating the flux of chemicals
eliminated from the fish via gill ventilation:
= CF • (Ec • Kv)
/i. • n) (5.3.11)
For most organic chemicals, gill elimination is a major
mechanism of chemical discharge from fish (Gobas
ef a/., 1989). Fecal elimination and excretion of
chemicals are not specifically modeled in this mass
balance equation. Their contribution can be viewed
as having been factored into the food of chemical
assimilation efficiency and gill transfer coefficient.
5.3.1.4 Chemical Dilution by Growth
Fish growth results is an increase of the fish volume
and a reduction of chemical mass per fish volume.
The equivalent flux of chemical loss due to fish
growth (Fg) is expressed as a product of the fish
growth rate (G) and chemical concentration in the
fish (CF).
Fg = G • CF (5.3.12)
where
G = growth rate of organism (1/day)
CF = chemical concentration in fish (ug-chem/kg-
body)
The fish growth rate (G) was estimated based on fish
weight-age relationships established for each fish
species.
5.3.2 Chemical Bioaccumulation in the
Base of Food Webs
The aquatic species at the base of the Lake Michigan
food web are zooplankton (pelagic) and Diporeia
(benthic). The modeled equations discussed above
for individual fish can not be applied to zooplankton
and Diporeia due to the lack of species-specific
bioenergetics data. Alternative submodels are
needed for chemical bioaccumulation in the base of
the food webs.
5.3.2.1 Chemical Bioaccumulation in
Zooplankton
Zooplankton in the Lake Michigan food webs are a
mixture of a wide variety of species. The species
composition of the zooplankton is not fixed. It varies
with season depending on the optimal temperature
for the growth of individual species. It is also
dependent on prey selections of its predators in a
given food web. At this stage, it is unfeasible to
develop a kinetic submodel for this species group
due to the lack of appropriate information.
For simplicity, a steady-state model was adapted in
our food web models to calculate concentrations in
Lake Michigan zooplankton. In this chemical
bioaccumulation submodel, zooplankton were
assumed to be a homogeneous pseudo-species.
Under steady-state, the chemical mass balance
Equation 5.3.1 can then be expressed as
FW + FP - Fe ~ F9 = ° (5.3.13)
The parameters in this equation have the same
definition as those in the fish submodel. Substituting
Equations 5.3.3, 5.3.6, 5.3.9, and 5.3.12 into
Equation 5.3.13, the chemical concentration in
zooplankton Cz can then be calculated by the
following equation.
G) (5.3.14)
460
-------
5.3.2.2 Chemical Bioaccumulation in Diporeia
There are several experimental studies on chemical
uptake from sediments by Diporeia (Landrum, 1989;
Landrum etal., 1985). Because most of the studies
were conducted under controlled laboratory
conditions, the kinetics of chemical exchange
between Diporeia and its environment derived from
these studies can not be readily transformed into a
kinetic model applicable to a real system. The lack
of information on site-specific growth data and the
difficulty in characterizing the surface sediment
portion that is actively selected by Diporeia as a food
source also hindered the development of a kinetic
model for chemical bioaccumulation in Diporeia.
The submodel for chemical bioaccumulation in
Diporeia used in this food web model was based on
a published steady-state model for benthic animals.
This model, introduced by Morrison et al. (1996),
assumes that under a steady-state condition the total
chemical intake flux from water (Uw) and food (Ud) by
a benthic animal equals the total chemical elimination
flux from the animal via gill (Dw), faeces (D(), and
metabolism (DJ:
Uw+Ud=Dw+ Df + Dm (5.3.15)
For detrivores, this assumption yields the equation:
(V g = [Ew • Gw • (fw/fs) + Ed • Gd • (fd/fs)
DSd-OCd
I[EW-GW
-3)
+ vb'Kn,'Ktw\ (5.3.16)
where
fb = chemical fugacity in benthos, Pa
fs = chemical fugacity in sediment, Pa
fa = chemical fugacity in diet (sediment or
suspended particles), Pa
Gw = gill ventilation rate, L/day
Ed = chemical assimilation efficiency from diet
Gd = food ingestion rate, L (wet volume)/day
DSd = density of diet (wet), kg/L
OCd = organic carbon fraction of diet on wet
weight base
Koc = organic carbon-water partition coefficient,
L7kg
cr = organic carbon assimilation efficiency
/? = fraction of ingested diet absorbed
Vb = volume of benthic animal, L
km = chemical metabolic transformation rate in
benthic animal, 1/day
Kbw = benthos-water partition coefficient of
chemicals, L/L
With mathematical manipulation and necessary unit
conversion, the chemical fugacity terms (fb/fs), (fw/fs)
and (fyfs) in this model equation can be replaced by
some more readily available chemical parameters.
The submodel for chemical bioaccumulation in
benthic animals can then be expressed as:
CB - (EW-GW-CW + Ed- GD-CD) • Lb-Kow/
/[EW-GW-1000 + Ed-(1 -a)(1 - P)
•
-------
Wb = body weight of fresh benthic animal, gram
Lb = lipid fraction in fresh benthic animal
5.3.3 Model Description of Exposure
Environment
Calculations in the submodels discussed above
require information which characterizes the
environmental conditions for individual organisms,
such as environmental temperature, oxygen content,
and the contaminant levels in water (for pelagic
species) and sediment (for benthic species). These
data are essential for application of a food web
bioaccumulation model.
However, among all existing aquatic food web
models the environmental condition of a food web is
typically defined with a single spatial compartment.
This makes no distinction of preferred living condition
among individual organisms and implies that ail
organisms in a food web live in an uniform
environment. This simplified model approach is
adequate for food webs in shallow and small water
bodies where gradients are relatively small, and thus
the exposure environments are expected to be
similar among organisms in different trophic levels on
a seasonal basis. However, for food webs in a large
aquatic system, such as Lake Michigan, the single
spatial compartment approach for defining exposure
environment of a food web may not be adequate.
In Lake Michigan, the spatial variation in water
temperature can be substantial, especially during
summer stratification (Ayers, 1962; Brandt et al.,
1991; Carr, 1973; Sommers ef al., 1981). As a
result, organisms in the lake are exposed to different
temperatures depending on individual temperature
preferences (Brandt et al., 1980; Otto et al., 1976).
Species living in surface water are exposed to a
temperature that varies dramatically from season to
season. Species living in deep water are exposed to
a relatively stable temperature. There are also
species that prefer different environments at different
life stages. The exposure temperatures of these
species are expected to vary by age (Lantry and
Stewart, 1993; Stewart and Binkowski, 1986). It is
therefore, possible for a food web to consist of
predators and prey that have different exposure
temperatures. It appears that existing food web
model frameworks are not adequately formulated to
accommodate the differential exposure temperatures
among organisms in Lake Michigan food webs.
To better represent the exposure environment for
each component of a food web and thus, to reduce
the associated uncertainties in model estimates, a
multi-compartment approach was introduced in the
food web model framework. Unlike the original single
compartment modeling approach which models the
exposure condition as a homogeneous one for the
whole food web, the multi-compartment approach
allows modelers to define the exposure conditions
individually for each organism with separate spatial
compartments. Each compartment can be assigned
organism-specific parameters which reflect the
environmental condition of the preferred location of
the associated organism. The temporal variation of
the preferred location of the organism can be
represented by the corresponding change in the
parameters of the compartment over time. Figure
5.3.2 provides the conceptual diagrams for both the
original single compartment approach and the new
multi-compartment modeling approach. For the
modified model approach, the differential exposure
temperatures among the organisms in a food web
can be easily described by defining each organism
with an independent spatial compartment.
References
Ayers, J.C. 1962. Great Lakes Water, Their
Circulation and Physical and Chemical
Characteristics. In: H.J. Pincus (Ed.), Great
Lakes Basin. American Association for the
Advancement of Science, Washington, D.C.
Brafield, A.E. and M.J. Llewellyn. 1982. Animal
Energetics. Blackie and Son, Ltd., Glasgow,
Scotland. 168 pp.
Brandt, S.B., J.J. Magnuson, and L.B. Crowder.
1980. Thermal Habitat Partitioning by Fishes in
Lake Michigan. Canadian J. Fish. Aquat. Sci.,
37(7): 1557-1564.
Brandt, S.B., D.M. Mason, E.V. Patrick, R.L. Argyle,
L. Wells, P.A. Unger, and D.J. Stewart. 1991.
Acoustic Measures of the Abundance and Size of
Pelagic Planktivores in Lake Michigan. Canadian
J. Fish. Aquat. Sci., 48(5):894-908.
462
-------
forage fish 1
zooplankton
predator fish
forage fish 2
forage fish 3
phytoplankton
D = Ta(t)
Homogeneous compartment
with an average seasonal
temperature profile for whole
food web
benthos
A. Single compartment approach for exposure temperature in
existing food chain models.
Separate compartments
with individualized
seasonal temperature
profile for each food
web component
B. Multi-compartment approach for exposure temperature in food
chain models.
Figure 5.3.2. Comparison of modeling
approaches for exposure temperatures in food
web models.
Carr, J.F., J.W. Moffett, and J.E. Gannon. 1973.
Thermal Characteristics of Lake Michigan, 1954-
1955. U.S. Bureau of Sport Fisheries and
Wildlife, Washington, D.C. Technical Paper 69,
143pp.
Connolly, J.P., T.F. Parkerton, J.D. Quadrini, S.T.
Taylor, and A.J. Turmann. 1992. Development
and Application of PCBs in the Green Bay, Lake
Michigan Walleye and Brown Trout and Their
Food Webs. Report to the U.S. Environmental
Protection Agency, Office of Research and
Development, ERL-Duluth, Large Lakes
Research Station, Grosse He, Michigan. 300 pp.
Gobas, F.A.P.C., K.E. Clark, W.Y. Shiu, and D.
Mackay. 1989. Bioaccumulation of
Polybrominated Benzenes and Biphenyls and
Related Superhydrophobic Chemicals in Fish:
Role of Bioavailability and Fecal Elimination.
Environ. Toxicol. Chem., 8(3):231-247
Gobas, F.A.P.C. 1993. A Model for Predicting the
Bioaccumulation of Hydrophobic Organic
Chemicals in Aquatic Food-Webs: Application to
Lake Ontario. Ecol. Model., 69(1/2):1-17.
Greenberg, A.E., L.S. Clesceri, and A.D. Eaton
(Eds.). 1992. Standard Methods for the
Examination of Water and Wastewater, 18th
Edition. American Public Health Association,
Washington, D.C. 982 pp.
Hewett, S.W. and B.L. Johnson. 1989. A General
Bioenergetics Model for Fishes. American
Fisheries Society Symposium, 6:206-208.
Landrum, P.F., B.E. Eadie, W.R. Faust, N.R.
Morehead, and M.J. McCormick. 1985. The
Role of Sediment in the Bioaccumulation of
Benzo(a)pyrene by the Amphipod Pontoporeia
hoyi. In: M. Cooke and A.J. Dennis (Eds.),
Polynuclear Aromatic Hydrocarbons:
Mechanisms, Methods and Metabolism, Eighth
International Symposium, pp. 799-812. Batteile
Press, Columbus, Ohio.
Landrum, P.F. 1989. Bioavailability and
Toxicokinetics of Polycyclic Aromatic
Hydrocarbons Sorbed to Sediments for the
Amphipod, Pontoporeia hoyi. Environ. Sci.
Technol., 23(5):588-595.
Lantry, B.F. and D.J. Stewart. 1993. Ecological
Energetics of Rainbow Smelt in the Laurentian
Great Lakes: An Interlake Comparison. Trans.
Amer. Fish. Soc., 122(5):951-976.
Morrison, H.A., F.A.P.C. Gobas, R. Lazar, and G.D.
Haffner. 1996. Development and Verification of
a Bioaccumulation Model for Organic
Contaminants in Benthic Invertebrates. Environ.
Sci. Technol., 30(11):3377-3384.
463
-------
Otto, R.G., M.A. Kitchel, and J.O. Rice. 1976.
Lethal and Preferred Temperatures of the Alewife
(Alosa pseudoharengus) in Lake Michigan.
Trans. Amer. Fish. Soc., 105(1):96-106.
Rudstam, L.G. 1989. A Bioenergetic Model for
Mysis Growth and Consumption Applied to a
Baltic Population of Mysis mixta. J. Plankton
Res., 11(5):971-983.
Rudstam, L.G., P.P. Binkowski, and M.A. Miller.
1994. A Bioenergetic Model for Analysis of Food
Consumption Patterns by Bloater in Lake
Michigan. Trans. Amer. Fish. Soc., 123(3):344-
357.
Sommers, L.M., C. Thompson, S. Tainter, L. Lin,
T.W. Colucci, and J.M. Lipsey. 1981. Fish in
Lake Michigan. Michigan Sea Grant Advisory
Program, Ann Arbor, Michigan. 38 pp.
Stapleton, H.M., R.J. Letcher, and J.E. Baker. 2001.
PCB Metabolism in a Freshwater Fish. Environ
Sci. Technol., 35(12):4747-4752.
Stewart, D.J., D. Weininger, D.V. Rottiers, and T.A.
Edsall. 1983. An Energetics Model for Lake
Trout, Salvelinus namaycush: Application to the
Lake Michigan Population. Canadian J. Fish.
Aquat. Sci., 40(6):681-698.
Stewart, D.J. and F.P Binkowski. 1986. Dynamics
of Consumption and Food Conversion by Lake
Michigan Alewives: An Energetics-Modeling
Synthesis. Trans. Amer. Fish. Soc., 115(5):643-
661.
U.S. Environmental Protection Agency. 1999.
Polychlorinated Biphenyls (PCBs) Update:
Impact on Fish Advisories. U.S. Environmental
Protection Agency, Office of Water, Washington,
D.C. 7pp.
464
-------
PARTS
LM FOOD CHAIN
Chapter 4. Description of Data,
Constants, and Other Information
Necessary to Run Model
5.4.7 Chemical
Contaminants
Properties of PCB
Polychlorinated biphenyls (PCBs) have been
recognized as significant environmental contaminants
since 1966 (Mullin et ai, 1984). Their impact is
particularly evident in the Great Lakes basin
(Neidermeyer and Mickey, 1976; Hesselberg et ai,
1990; Oliver et ai, 1989; Eisenreich et ai, 1989). In
this modeling project, 40 PCB congeners or co-
eluters were targeted for simulation of their individual
bioaccumulation by fish in the lake. Most of the PCB
congeners were selected for their abundance and
bioaccumulative tendency in the lake ecosystem.
Other PCB congeners were included to make the
targeted PCB group cover the full range of PCB
hydrophobicity, and thus, a better representative
subset of all existing 209 PCB congeners.
Hydrophobicity of a PCB congener is measured by its
octanol-water partition coefficient (Kow) which is the
most important chemical property governing
bioaccumulation of the congener in organisms.
Another important chemical property involved in
modeling PCB contaminants is the organic carbon
partition coefficient (Koc) whose value can often be
correlated to that of Kow. In this work, the following
empirical relationship (Eadie et ai, 1990) was used:
= 1-94 + 0.72 logK0.
(5.4.1)
The targeted PCB congeners or co-eluter congeners
are listed in Table 5.4.1 with their octanol-water
partition coefficients Kow. The values of Kow are
those of Hawker and Connell (1988). The molecular
weight (MW) for each PCB congener is also listed for
additional reference.
5.4.2 Site-Specific Data
5.4.2.1 Fish Food Web Structures
The structure of a food web shows how individual
organisms in the food web are related to each other
through feeding interactions. This dietary information
is necessary for establishing appropriate linkages
among individual submodels of a food web model
and is important to the accurate simulation of
chemical bioaccumulation in the food web.
The fish food webs of interest are those of two top
predators in Lake Michigan, lake trout and coho
salmon. These two species were selected for their
important economic value. It is desirable to have a
better understanding of the present and future
concentrations of PCB contaminants in these two fish
populations with the help of model simulations.
5.4.2.1.1 Lake Trout Food Web
It is believed that the lake trout in Lake Michigan are
represented by three subpopulations at Sturgeon
Bay, Sheboygan Reef, and Saugatuck (Figure 5.4.1).
Movements of lake trout in Lake Michigan are
believed to be considerably restricted in range
(Brown et ai, 1981). Each of the lake trout
subpopulations has a site-specific food web
structure.
465
-------
Table 5.4.1. Targeted PCB Congeners and Their K0
Congener
4
2,3
2,4'
3,4
3,4'
4,4'
2,4',4
2,2', 3
2,4,6
2,2', 5
2,3', 5
2,4,4'
2,4', 5
2',3,4
3,4,4'
2,2',3,4
2,2', 3,5'
2,24,5'
2,2',5,5'
2,3,3',41
2,3,4,4'
2,3',4,4'
2,3',4',5
2',3,4,5
2,4,4',5
3,3', 4,4
2,3>3',4',6
3,4,4',5
2,2', 3,4,5'
2,2', 3,3', 6
2,2', 3,5,5'
2,2',3,4,61
2,2',3,4,4'
2,2', 4,4, 5
2,2', 4,5,5'
2,3',4,4',5
2',3)4,41,5
2,2'I3,4',5',6
2,3,3', 4,4
2,2', 3,3', 4,6'
2,2',4,4,5,5'
2,2',3,5,5'J6
2,2',3,4,4I,5'
2,3,3',4',5,6
2,21,3,4',5,51
2,21,3,31,4,41,5
IUPAC
0
3
5
8
12
13
15
17
16
32
18
26
28
31
33
37
42
44
49
52
56
60
66
70
76
74
77
110
81
87
84
92
89
85
99
101
118
123
149
105
132
153
151
138
163
146
170
Homolog
0
1
2
2
2
2
2
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
4
5
4
5
5
5
5
5
5
5
5
5
6
5
6
6
6
6
6
6
7
Molecular Weight
154
188
223
223
223
223
223
257
257
257
257
257
257
257
257
257
292
292
292
292
292
292
292
292
292
292
292
326
292
326
326
326
326
326
326
326
326
326
361
326
361
361
361
361
361
361
395
log Kow
4.09
4.69
4.97
5.07
5.22
5.29
5.3
5.25
5.16
5.44
5.24
5.66
5.67
5.67
5.6
5.83
5.76
5.75
5.85
5.84
6.11
6.11
6.2
6.2
6.13
6.2
6.36
6.48
6.36
6.29
6.04
6.35
6.07
6.3
6.39
6.38
6.74
6.74
6.67
6.65
6.58
6.92
6.64
6.83
6.99
6.89
7.27
466
-------
Table 5.4.1. Targeted PCB Congeners and Their Kow (Continued)
Congener
IUPAC
Homolog Molecular Weight
logK0
2,3,3',4)41,5,61
2,2', 3,3', 4,5,5'
2,2', 3,3', 4,4', 6,6'
2,2', 3,4,4', 5,5'
2,2, 3,4,4', 5,&
2,2, 3,4', 5,5', 6
2,2', 3,3', 4,4', 5,6
2,2',3,3',4,5,51,6,6I
2,21,3,3',4,4',5',6
2>2',3,4>4',5,5',6
2,21,3,31,41,5,5'>6
190
172
197
180
182
187
195
208
196
203
201
7
7
8
7
7
7
8
9
8
8
8
395
395
430
395
395
395
430
464
430
430
430
7.46
7.33
7.3
7.36
7.2
7.17
7.56
7.71
7.65
7.65
7.62
Figure 5.4.1. Biota zones in Lake Michigan.
467
-------
For each lake trout subpopulation, the food web was
constructed using dietary data compiled from field
sampling of lake trout and associated forage fish
population. Lake trout (Salvelinus namaycush)
were caught at the three locations during the spring,
summer, and fall of 1994 and 1995. They were
primarily captured via gill netting at depths ranging
from 9 to 40 m. A minor portion of trout was
captured by bottom trawling. Bottom trawling was
used at depths of 10 to 50 m to obtain forage fish.
Prey fish included alewife (Alosa pseudoharengus),
rainbow smelt (Osmerus mordax), bloater
(Coregonus hoyi), slimy sculpin (Cottus cognatus),
and deepwater sculpin (Myoxocephalus thompsoni).
The diets of lake trout and forage fish were
determined by stomach analysis following a standard
operating procedure established for the Lake
Michigan Mass Balance Project (LMMBP) (U.S.
Environmental Protection Agency, 1997a). For lake
trout, the diet components were further classified into
age classes.
The organisms in the base of Lake Michigan fish food
webs are zooplankton, Mysis, and Diporeia. Their
dietary information was obtained from literature
sources. Mysis are reported to feed on zooplankton,
phytoplankton, and "fresh" detrital material at the
sediment surface and suspended in the water column
(Beeton and Bowers, 1982; Grossnickle, 1982).
Zooplankton are believed to feed on organic-rich
particles, mainly phytoplankton in the water column
(Peters and Downing, 1984). Diporeia are reported
to feed on relatively "fresh" detrital material at the
sediment surface (Evans etal., 1990; Gardner etal.,
1990; Johnson, 1987; Lydy and Landrum, 1993;
Marzolf, 1965; Quigley, 1988; Quigley and
Vanderploeg, 1991).
Annual average dietary data for lake trout and its
forage populations in the three biota zones of the
lake are summarized in Tables 5.4.2a through 5.4.7.
These data were used to construct a complete food
web structure for each of the three lake trout
populations in Lake Michigan.
5.4.2.1.2 Co/70 Salmon Food Web
The coho salmon in Lake Michigan are believed to
move around large portions of the lake during the
fish's lifetime (Patriarche, 1980). They were modeled
as a single lake-wide population. The dietary
information of the coho salmon was compiled from
field sampling. Coho salmon (Oncorhynchus
kisutch) were sampled from angler's catches at
various locations of the lake from May to November
in 1994 and April to November in 1995.
The diet of coho salmon was determined by stomach
analysis following a standard operating procedure
established for the LMMBP (Elliott etal., 1996; Elliott
and Holey, 1998; U.S. Environmental Protection
Agency, 1997a). The prey species were further
classified into age classes. The results are
presented in Table 5.4.8.
Due to their extensive movement, coho salmon in the
lake may encounter site-specific forage populations
in different regions. This means that a given forage
species in the coho salmon diet may belong to
different subpopulations. The forage fish may have
a location-dependent dietary history. Therefore, the
food web structure below the top trophic level can
vary with the movement of coho salmon. In order to
construct an accurate food web structure for coho
salmon in Lake Michigan, information on its migration
pattern and food web structures of its forage
populations in related locations is needed. The
migration pattern of the coho salmon was established
based on a general index of fish density, catch-per-
unit-of-effort (CPE), in various locations on a monthly
basis. In general, the fish aggregate in southern
Lake Michigan during spring and travel to the
southwestern region of the lake in summer. In the
late summer and early autumn, most of the coho
salmon are found in the northeastern region of the
lake. They move back to the southeastern region
during the winter. However, dietary information for
forage fish in these locations were not readily
available. Therefore, it was not possible to construct
a comprehensive food web structure for coho salmon
that reflects the seasonal or spatial variation of its
forage food webs.
The most complete dietary information for forage fish
was that collected from the Sturgeon Bay,
Sheboygan Reef, and Saugatuck lake trout biota
zones (Tables 5.4.3 through 5.4.7). In this study,
these dietary data were used to construct three local
food web structures for the coho salmon by linking
each of them with the dietary data of the coho
salmon as presented in Table 5.4.8.
468
-------
Table 5.4.2a. Annual Dietary Composition of Lake Trout at Saugatuck (1994-1995)
Lake
Trout
Age
Age1
Age 2
Age3
Age 4
Age 5
Age 6
Age?
Age 8
Age 9
Age 10
Forage
Fish
Age
Age1
Age 2
Age3
Age 4
Age 5
Age 1
Age 2
Age 1
Age 2
Age 3
Age 4
Age 2
Age 3
Age 4
Age 2
Age3
Age 4
Age 2
Age 3
Age 4
Age 5
Age 6
Age 3
Age 4
Age 5
Age 6
Age 3
Age 4
Age 5
Age 6
Age 7
Age 4
Age 5
Age 6
Age 7
Age 2
Age 3
Age 4
Age5
Age 6
Alewife Rainbow Bloater Slimy Deepwater Diporeia Mysis
Smelt Sculpin Sculpin
20 20
20
20
20
35
5 20 40
10 20 10
20
30
10
5 25 10
10 25
25
5
10 20 15
10 40
10 5
20
10 5 20
20 10
15
15 30
10
30
10
20 15
20
20 10
5
20
30
20 10
20
10
10
15 15 10
10
30
469
-------
Table 5.4.2a. Annual Dietary Composition of Lake Trout at Saugatuck (1994-1995) (Continued)
Lake
Trout
Age
Age 11
Age 12
Table 5.4.2b.
Lake
Trout
Age
Age1
Age 2
Age 3
Age 4
Age 5
Age 6
Forage
Fish
Age
Age 3
Age 4
Age 5
Age 6
Age?
Age1
Age 2
Age 3
Age 4
Age 5
Age 6
Age 7
Annual
Forage
Fish
Age
Age 1
Age1
Age1
Age 2
Age 1
Age 2
Age 3
Age 4
Age 5
Age 1
Age 2
Age 3
Age 4
Age 5
Age 6
Age 2
Age 3
Age 4
Age 5
Alewife Rainbow Bloater Slimy Deepwater Diporeia
Smelt Sculpin Sculpin
10
30
25
10
25
5
10
20 15
10 30
10
Dietary Composition of Lake Trout at Sheboygan Reef (1994-1995)
Alewife Rainbow Bloater Slimy Deepwater Diporeia
Smelt Sculpin Sculpin
85 15
80 10 5
55
20
20 10
10
10
10
20
15
15
10 20
10
30 10
20
10
10 20
Mysis
Mysis
5
45
20
10
470
-------
Table 5.4.2b. Annual Dietary Composition of Lake Trout at Sheboygan Reef (1994-1995)
(Continued)
Lake
Trout
Age
Age?
Age8
Age 9
Age 10
Age 11
Age 12
Forage
Fish
Age
Age 2
Age3
Age 4
Age 5
Age 2
Age 3
Age 4
Age 5
Age 2
Age3
Age 4
Age 5
Age 6
Age 2
Age 3
Age 4
Age 5
Age 6
Age 2
AgeS
Age 4
Age 5
Age 6
Age 2
AgeS
Age 4
Age 5
Age 6
Age 7
Alewife
35
25
10
15
20
5
20
15
10
15
30
20
5
20
40
5
20
20
10
10
15
10
10
Rainbow Bloater Slimy Deepwater Diporeia Mysis
Smelt Sculpin Sculpin
15
20
20
10
15
15
10
10
15
20
20
20
25
471
-------
Table 5.4.2c. Annual Dietary Composition of Lake Trout at Sturgeon Bay (1994-1995)
Lake
Trout
Age
Age1
Age 2
Age 3
Age 4
Age 5
Age 6
Age?
Age 8
Age 9
Forage
Fish
Age
Age 1
Age 1
Age1
Age 2
Age 3
Age1
Age 2
AgeS
Age 4
Age1
Age 2
Age 3
Age 4
Age1
Age 2
Age 3
Age 4
Age 5
Age 6
Age?
Age 2
Age3
Age 4
Age 5
Age 6
Age?
Age 2
Age 3
Age 4
Age5
Age 6
Age?
Age3
Age 4
Age5
Age 6
Age?
Alewife
85
80
45
10
30
10
10
30
15
10
10
20
30
15
30
20
20
10
10
10
20
25
10
5
10
30
20
10
Rainbow Bloater Slimy Deepwater Diporeia Mysis
Smelt Sculpin Sculpin
15
10 55
5
5 10
5 20
20
30
15
15
15
5
10
10
5
5
15
5
10
10
10 10
472
-------
Table 5.4.2c. Annual Dietary Composition of Lake Trout at Sturgeon Bay (1994-1995) (Continued)
Lake
Trout
Age
Age 10
Age 11
Age 12
Forage
Fish Alewife Rainbow Bloater Slimy Deepwater
Age Smelt Sculpin Sculpin
Age 2 5
Age 3 15 5
Age 4 20
Age 5 25 5 5
Age 6
Age 7 20
Age 2 5
Age 3 1 5
Age 4 20
Age 5 35 5
Age 6
Age 7 20
Age 2 15
Age 3 25
Age 4 10
Age 5 25 25
Diporeia Mysis
Table 5.4.3. Dietary Composition of Alewife in Lake Michigan (1994-1995)
Small:
Age 1-2
Large:
Age 3-7
Saugatuck Sturgeon Bay
Prey (0-<75m) (0-~100m)
Fish Length < 120 mm
Diporeia 10 45
Mysis
Zooplankton 90 55
Fish Length > 120 mm
Diporeia 10 75
Mysis
Zooplankton 90 25
Sheboygan Reef
(50 - 75 m)
40
60
20
50
30
473
-------
Table 5.4.4. Dietary Composition of Bloater in Lake Michigan (1994-1995)
Small:
Age 1-3
Large:
Age 4-7
Prey
Fish Length <= 160 mm
Diporeia
Mysis
Zooplankton
Fish Length (g) > 160 mm
Diporeia
Mysis
Zooplankton
Saugatuck
(0 - < 75 m)
80
20
75
25
Sturgeon Bay
(0-~100m)
100
70
30
Sheboygan Reef
(50 - 75 m)
35
35
30
25
75
Table 5.4.5. Dietary Composition of Rainbow Smelt in Lake Michigan (1994-1995)
Prey
Saugatuck
(0 - < 75 m)
Sturgeon Bay
(0-~100m)
Sheboygan Reef
(50 - 75 m)
All Ages
Diporeia
Mysis
Zooplankton
65
35
10
90
60
40
Table 5.4.6. Dietary Composition of Slimy Sculpin in Lake Michigan (1994-1995)
Prey
Saugatuck
(0 - < 75 m)
Sturgeon Bay
(0-~ 100m)
Sheboygan Reef
(50 - 75 m)
All Ages
Diporeia
Mysis
90
10
80
20
90
10
474
-------
Table 5.4.7. Dietary Composition of Deepwater Sculpin in Lake Michigan (1994-1995)
Prey
Saugatuck
(0 - < 75 m)
Sturgeon Bay
(0-~100m)
Sheboygan Reef
(50 - 75 m)
All Ages
Diporeia
Mysis
70
30
45
55
80
20
Table 5.4.8. Dietary Composition of Coho Salmon in Lake Michigan (1994-1995)
Coho
Salmon Age
Age 1
Age 2
Forage Fish
Age
Age 1
Age 2
Age 1
Age 2
Age3
Age 4
Age 5
Age 6
Age 7
Alewife Rainbow Smelt Bloater Diporeia Mysis
40 10 10
40
25
10
20 5
20
10
10
5.4.2.2 Fish Growth Rates
At a given body weight, W, fish growth rate, G, can
be written as:
where
G = (dw/dt)/W
where
(dw/dt) = the derivative of fish weight W with
respect to fish age t
With a set of weight-age data of a fish available, the
average value for the fish growth rate for a given
period of time can then be estimated by the following
equation:
(5.4.3)
W1 = fish weight (g) at age t, (day)
W0 = fish weight (g) at age t0 (day)
(5.4.2) G = fish average growth rate during age t0 to t1
The weight-age data for fish species in the food webs
were obtained from field sampling conducted in
1994-1995 by the Great Lakes National Program
Office (GLNPO) for the LMMBP. The methods offish
collection are described in Section 4.2.1. Each fish
was weighed to the nearest gram. The lake trout and
coho salmon were aged based on either decoding
the information on a coded-wire tag (if found) or
enumeration of annuli on scales in conjunction with
use of fin clip information. More details on the fish
aging procedure can be found in Lake Michigan
Mass Balance Study Methods Compendium (U.S.
475
-------
Environmental Protection Agency, 1997a) and
Madenjian et al. (1998a, 1999). Forage fish were
aged based on lengths and weights taken from the
literature, and compared to the length and weight
data collected for each of the fish species in this
study.
A general relationship between age and weight for
each fish was established through regression of the
large amount of field data. The age-weight
relationships for the lake trout in three biota zones,
the migratory coho salmon, and their forage fish
populations are presented in Tables 5.4.9a through
5.4.9c. Age-weight relationships for forage fish
exhibit no regional variation, and a lake-wide average
was obtained for each forage species. The results in
Tables 5.4.9a, 5.4.9b, and 5.4.9c were used to
estimate fish growth rates in the food web models.
The weight-age relationship for Mysis was estimated
based on information from literature sources (Brafield
and Llewellyn, 1962; Pothoven et al., 2000). The
results are presented in Table 5.4.9d.
A constant value of 0.10 (1/day) was adapted as the
average growth rate for zooplankton in the lake
(Connolly et al., 1992).
5.4.2.3 Energy Density of Food Web Components
In a bioenergetics-based food web model, energy
balance is the basis for estimating chemical fluxes
between fish and its prey species. It is, therefore,
important to have a good knowledge of the energy
content of the fish and its prey items.
Energy densities, D, of all fish species in this study
were estimated based on lipid and protein fractions
in individual organisms (Lucas, 1996).
D = 35.5 fL + 20.08 fpr
(5.4.4)
The terms fL and fpr are lipid and protein fractions in
the fish body, respectively. The energy equivalents
of lipid components (kj/g) is 35.5, and the energy
equivalents of protein components (kJ/g) is 20.08.
The standard value of energy equivalent for protein
is 23.4 kJ/g-protein (Cho et al., 1982). It was
adjusted to a lower value of 20.08 kJ/g-protein
because after digestion, a portion of energy in the
assimilated protein is lost by nitrogenous excretion
and is not available for further respiration. Energy
contributions from other body components of a fish,
such as carbohydrates, are negligible (Diana, 1995).
Fish lipid content was analyzed by extracting
homogenized fish composite with 100 mL of 90/10
(v:v) petroleum ether/ethyl acetate. The extract was
then evaporated and the residue was weighed as
extractable lipid. Detailed procedures for fish lipid
separation and determination are available in the
Lake Michigan Mass Balance Study Methods
Compendium (U.S. Environmental Protection
Agency, 1997b) and Madenjian et al (2000). The
values of protein fraction in the lake trout, coho
salmon, and the other fish were compiled from or
estimated based on various literature sources (Flath
and Diana, 1985; Foltz and Norden, 1977; Gardner
et al., 1985; Rottiers and Tucker, 1982; Schindler ef
al., 1971; Vijverberg and Frank, 1976). The lipid and
protein fractions used for estimating energy content
for all organisms in this study are compiled in Tables
5.4.1 Oa through 5.4.1 Oh.
5.4.2.4 Exposure Conditions
Environmental conditions to which fish are exposed
play an important part in determining chemical
exchange fluxes between a fish and its environment.
Among the model parameters which characterize the
environmental conditions for food webs, contaminant
levels in water and sediment have direct influence on
the contaminant level in exposed fish food webs, and
temperature and oxygen content of the exposure
environment regulate the chemical kinetics in fish
food webs.
Due to the variation in Lake Michigan water
characteristics, the exposure condition is different
among fish food webs in different biota zones. To
facilitate model calculations for fish food webs at
Sturgeon Bay, Sheboygan Reef, and Saugatuck,
exposure information for each of these three biota
zones was required. Exposure data used are
summarized here. All data for the LMMBP are
available upon request to the GLNPO.
476
-------
Table 5.4.9a. Average Weight-Age Relationships for Lake Trout in Lake Michigan (1994-1995)
Sheboygan Reef
Age Weight (g)
1 20
2 128
3 244
4 490
5 900
6 1378
7 1900
8 2600
9 3400
10 4000
1 1 4400
12 4700
13 4900
14 5200
Saugatuck
Weight (g)
90
180
550
1100
2050
2850
3400
4000
4500
5400
6500
6900
7100
7100
Sturgeon Bay
Weight (g)
98
120
350
800
1500
2700
.3200
3700
4400
5000
5500
5600
5800
6000
Table 5.4.9b. Average Weight-Age Relationships for Coho Salmon in Lake Michigan (1994-1995)
Age Day
1 90
122
152
183
214
244
274
304
335
366
2 30
60
90
121
151
183
214
244
274
304
Weight (g)
30
80
140
220
322
450
620
878
880
885
890
895
900
1400
1850
2190
2450
2670
2860
3050
477
-------
Table 5.4.9c. Average Weight-Age Relationships of Forage Fish in Lake Michigan (1994-1995)
Age
1
2
3
4
5
6
7
8
9
10
11
Alewife
Weight (g)
3
15
27
37
45
50
53
55
Bloater
Weight (g)
3.7
12
26
38
50
65
88
110
Rainbow Smelt
Weight (g)
5.3
8
13
19
22
25
28
30
32
34
Slimy Sculpin
Weight (g)
0.6
1.2
2.2
4.6
8.4
10
10.6
Deepwater Sculpin
Weight (g)
0.6
1.8
3.5
7
13
19
24
29
34
38
40
Table 5.4.9d. Estimated Weight-Age Relationships of Mysis in Lake Michigan
Month
Weight (g-wet)
Sturgeon Bay
Weight (g-wet)
Sheboygan Reef
Weight (g-wet)
Saugatuck
0 0.00019
4 0.00194
8 0.00893
12 0.01691
16 0.03336
0.00001
0.00061
0.00330
0.00910
0.01860
0.00001
0.00095
0.00537
0.01706
0.04123
Table 5.4.10a. Average Lipid and Protein Fractions (%) of Lake Trout in Lake Michigan (1994-1995)
Age
Sheboygan Reef Sturgeon Bay
Saugatuck
Protein %
1 2.3
2 3.66
3 7.9
4 9.36
5 12.48
6 15.56
7 18.6
8 19.36
9 19.34
10 19.1
1 1 20.73
12 22.4
13 20.2
14 20.1
15
4.8
4.68
9.21
11.81
17.04
18.3
19.13
20.52
20.15
22.63
22.5
20.53
20.9
21.4
22.4
2.3 17.37
3.66
7.13
9.52
14.77
18.96
21.05
18.56
19.12
20.68
22
23
21.7
19.7
30.6
478
-------
Table 5.4.1 Ob. Average Lipid and Protein Fractions (%) of Coho Salmon in Lake Michigan (1994-1995)
Age Day Lipid % Protein %
1 90
122
152
183
214
244
274
304
335
366
2 30
60
90
121
151
183
214
244
274
304
5.14 20.00
5.25
5.37
5.54
5.75
6.01
6.36
6.90
6.90
6.91
6.92
6.93
6.94
7.98
8.91
9.61
10.15
10.61
11.00
11.39
Table 5.4.1 Oc. Average Lipid and Protein Fractions (%) of Alewife in Lake Michigan (1994-1995)
Age Sheboygan Reef Saugatuck Sturgeon Bay Protein %
1 7.2
2 8.5
3 9
4 10.5
5 11.5
6 12
7 12.2
8 12.5
5.5
5.5
6
7.5
9
10
11
12
4
6
6
6
6
6
6
6
16.7
479
-------
Table 5.4.1 Od. Average Lipid and Protein Fractions (%) of Bloater in Lake Michigan (1994-1995)
Age Sheboygan Reef
1 5
2 5.5
3 8
4 11
5 12
6 12.5
7 13
8 13.5
Saugatuck
4
4.5
5.5
6.5
7.5
8.5
10.5
11
Sturgeon Bay
5
7
8.5
9.5
12.5
13.5
14.5
15.5
Protein %
16.3
Table 5.4.1 Oe. Average Lipid and Protein Fractions (%) of Rainbow Smelt in Lake Michigan (1994-1995)
Age Sheboygan Reef
1 4.4
2 4.4
3 4.4
4 4.4
5 4.4
6 4.4
7 4.4
8 4.4
9 4.4
10 4.4
Saugatuck
3.5
3.5
3.5
3.5
3.5
3.5
3.5
3.5
3.5
3.5
Sturgeon Bay
3
3
3
3
3
3
3
3
3
3
Protein %
16.9
Table 5.4.1 Of. Average Lipid and Protein Fractions (%) of Slimy Sculpin in Lake Michigan (1994-1995)
Age Sheboygan Reef
1 6.4
2 6.5
3 6.6
4 6.8
5 7.1
6 7.2
7 7.3
Saugatuck
3.5
4
4.5
5
5.2
5.2
5.2
Sturgeon Bay
8
8.1
8.2
8.3
8.4
8.5
8.5
Protein %
15.9
480
-------
Table 5.4.1 Og. Average Lipid and Protein Fractions (%) of Deepwater Sculpin in Lake Michigan (1994-
1995)
Age
Sheboygan Reef
Saugatuck
Sturgeon Bay
Protein %
1
2
3
4
5
6
7
8
9
10
11
8.8
8.9
9
9.1
9.4
9.7
9.9
10.1
10.3
10.5
10.6
2
3
4
5
5.5
6
7
7.2
7.2
7.5
7.5
7 14.4
7.1
7.2
7.3
7.5
7.7
7.8
7.9
8
8.1
8.2
Table 5.4.1 Oh. Average Lipid and Protein Fractions (%) of Zooplankton, Mysis, and Diporeia in Lake
Michigan (1994-1995)
Species
Sheboygan Reef
Saugatuck
Sturgeon Bay
Protein %
Zooplankton
Mysis
Diporeia
2.91
2.31
3.21
2.79
1.61
1.66
1.57
2.9
4.48
7.1
7
10
5.4.2,4.1 PCB Concentrations in Water
Lake Michigan water and particulate samples were
collected at several stations within the Sturgeon Bay,
Sheboygan Reef, and Saugatuck biota zones.
Information regarding the sampling stations,
collection procedures, sample preparation, and
methods for PCB analysis are available in detail (U.S.
Environmental Protection Agency, 1997a, 1997b).
The organic carbon fraction in the suspended
particles was also analyzed. The analysis
procedures can also be found in the above
documents.
No temporal variation of PCB concentrations was
found for samples collected during 1994 and 1995.
PCB concentrations in suspended particles were
organic carbon normalized. There was substantial
variation of PCB concentrations in suspended
particles among samples collected from different
water depths. No substantial vertical variation was
found for PCBs in the dissolved form. PCBs in
suspended particles were divided into those collected
at depth < 20 m and those collected at depth > 20 m.
For this study, it was assumed that the fish food
webs were exposed to particulate PCB
concentrations in the deeper layer. Median values
for dissolved PCBs and those associated with
suspended particles were used for model calibration.
The PCB concentrations in the water column of the
three biota zones are given in Table 5.4.11.
5.4.2.4.2 PCB Concentrations in Sediment
Sediment sampling was not specifically conducted
within the three biota zones. Sediment PCB
concentrations in the three biota zones were
481
-------
Table 5.4.11. PCB Concentrations in Lake Michigan Water Column (1994-1995)
Sturgeon Bay
PCB
Congeners
3
8+5
12
13
15+17
16
32
18
26
28+31
33
37+42
44
49
52
56+60
66
70+76
74
77+110
81
87
92+84
89
85
99
101
118
123+149
132+153+105
151
163+138
146
170+190
172+197
180
1 87+1 82
208+1 95
196+203
201
Dissolved
(ng/L)
0
0
0.002831
0.001163
0.003063
0
0
0.00333
0.000941
0.008012
0.004408
0.008967
0.003189
0.002259
0.005627
0.00134
0.001664
0.002179
0.00103
0.00291
7.68E-05
0.00227
0.005722
0.00068
0.000507
0.006156
0.001328
0.001236
0.000705
0.000724
0
0.002134
0.00059
4.74E-05
0
0
0.002588
4.34E-05
3.28E-05
0.000168
Particulate
(ng/g-OC)
0
0
0
0.63374
4.02012
1.00157
1 .37860
3.57836
0.22132
12.42481
2.28478
14.5969
5.38999
3.96632
9.48455
7.20351
18.39126
7.46113
4.18880
13.79423
1.52913
4.43503
15.83466
0.15860
4.76618
25.2633
10.76926
10.49375
6.59078
18.7597
2.11833
20.59195
5.19236
2.54427
1 .03453
1.91204
4.91753
0.88921
1 .50532
3.05836
Sheboygan Reef
Dissolved
(ng/L)
0
0
0.002265
0.00122
0.002608
0
0
0.00377
0.001258
0.007067
0.005054
0.009517
0.002878
0.002054
0.005518
0.001344
0.001893
0.0021
0.001039
0.002586
0
0.002572
0.007226
0
0.000569
0.004236
0.00278
0.001156
0.000862
0.000958
0
0.002948
0.000583
7.36E-05
0
5.13E-05
0.000984
0
2.75E-05
7.34E-05
Particulate
(ng/g-OC)
0
0
6.94990
2.09185
7.54759
1 .56798
1 .58024
5.47443
0.37498
17.84289
3.4361 1
15.35747
7.30135
6.86181
16.12783
13.76338
29.53261
16.42939
5.84207
28.8021 1
2.09813
8.03297
32.15896
0
8.63774
36.02048
17.31631
19.16489
13.43283
31.21532
3.88177
37.87159
7.64438
5.30883
1.85133
5.94020
7.07428
1.91721
3.83510
6.59986
Saugatuck
Dissolved
(ng/L)
0
0
0.003126
0.0009
0.004061
0.001473
0
0.004623
0.001582
0.009846
0.006045
0.008866
0.00581
0.003302
0.008475
0.00198
0.002783
0.003036
0.001371
0.004342
0.000147
0.002373
0.01356
0
0.000681
0.004228
0.004522
0.001713
0.001331
0.001451
0
0.002877
0.000572
0.000131
0
0
0.000683
0
0
0.00018
Particulate
(ng/g-OC)
0
0
0
2.11576
11.95844
3.22824
4.22044
10.4442
2.96423
55.58153
9.97024
16.86476
20.89396
14.05582
35.36909
30.47388
63.51781
33.74864
14.10166
49.83354
2.60703
13.37302
74.07366
1 .59772
13.30061
49.72043
34.31707
35.9058
21 .52096
58.52275
6.37438
55.57444
9.26933
8.24745
2.81659
18.31716
12.63331
2.42335
5.10087
9.01875
482
-------
estimated based on samples collected at several
nearby stations. These stations were selected for
their closeness to a specific biota zone in distance,
depth, and sediment characteristics. Because
organic carbon normalized sediment PCB data
showed limited horizontal variation, the estimate of
sediment PCB exposure by using data from nearby
stations was appropriate. Information regarding the
sampling stations, collection procedures, sample
preparation, and methods for PCB analysis are
available in detail (U.S. Environmental Protection
Agency, 1997a, 1997b). Organic carbon and dry
fraction of sediment samples were also analyzed.
The analysis procedures can also be found in the
above documents.
Sediment data analysis revealed no significant
temporal variation in PCB concentrations for samples
collected during 1994 and 1995. PCB concentrations
in sediment were organic carbon normalized.
Median values for PCBs in sediment carbon were
used for model calculations. The concentrations of
PCBs dissolved in sediment pore water were
estimated based on measured PCB data, organic
carbon content, dry fraction in the sediment samples,
and organic carbon-water partition coefficients for
individual PCB congeners. The results of PCB
concentrations in the sediment solids and pore water
for the three biota zones are given in Table 5.4.12.
5.4.2.4,3 Exposure Temperature
Lake Michigan is a vast water body with a volume of
4,920 km3. It has a surface area of 57,800 km2, and
its deepest point is 282 m (Coordinating Committee
on Great Lakes Basic Hydraulic and Hydrologic Data,
1992). Physical characteristics of the lake vary with
region and depth (Environment Canada and U.S.
Environmental Protection Agency, 1997). To better
reflect this reality, the model was constructed to
simulate the exposure environment for each species,
rather than as a whole for all species in a food web.
The prevailing annual cycles of exposure
temperature for a lake-wide coho salmon population
and for three lake trout and their forage populations
at Sturgeon Bay, Sheboygan Reef, and Saugatuck
were established and are presented in Figures 5.4.2a
through 5.4.2c. The results were compiled based on
site-specific information, such as annual water
temperature profiles (U.S. Environmental Protection
Agency, 1995), species optimal temperature and
depth at different life stages (Otto et al., 1976;
Peterson et al., 1979; Stewart era/., 1983; Wismer
and Christie, 1987; Wells, 1968), prey availability
(Crowder and Crawford, 1984; Eck and Wells, 1986;
Janssen and Brandt, 1980), spawning season
(Janssen and Brandt, 1980), and spawn site
preference (Jude et al., 1986; Rice, 1985). For
simplicity, the exposure temperatures for different
age groups in certain species were aggregated and
average annual temperature cycles were determined
for the species. The seasonal variation of surface
water temperatures (U.S. Environmental Protection
Agency, 1995) in the lake is also presented in the
first panel of Figures 5.4.2a 5.4.2b and Figure
5.4.2c for reference.
5.4.2.4.4 Oxygen Concentration in Water
The oxygen concentration in water that organisms
vent through their gill membranes was determined by
water temperature. In this study, the dissolved
oxygen content in water [O2] was estimated
according to an empirical correlation between oxygen
solubility (mg/L) and water temperature (Greenberg
etal., 1992).
Ln[O2] = -139.34411 + (1.575701 x 105/T)
- (6.642308 x107/T*) + (1.2438 x ID10/!3)
- (8.621949 x 1011/74) (545)
where
7 = temperature (°K)
5.4.3 Physiological Data of Fish and
Other Organisms
5.4.3.1 Species-Specific Respiration Rates
In the bioenergetics-based food web model (LM Food
Chain), fish respiration (or metabolism) rate is a key
model parameter which determines the dynamics of
chemical uptake from water and food. Fish
respiration rate is dependent on fish weight,
temperature, and degree of fish activity. For most of
the fish species in the Lake Michigan food webs, an
extensive study of respiration as a function of weight,
483
-------
Table 5.4.12. PCB Concentrations in Lake Michigan Surface Sediment (1994-1995)
Sturgeon Bay
PCB
Congeners
3
8+5
12
13
15+17
16
32
18
26
28+31
33
37+42
44
49
52
56+60
66
70+76
74
77+110
81
87
92+84
89
85
99
101
118
123+149
132+153+105
151
163+138
146
170+190
172+197
180
187+182
208+195
1 96+203
201
Pore Water
(ng/L)
0
0.0279054
0.0014341
0.0013447
0.0096751
0.0038924
0.0024807
0.0071207
0.0026656
0.049642
0.0125176
0.009989
0.0149115
0.0078881
0.0174969
0.0183856
0.0446862
0.0180978
0.0075816
0.0175742
0.0004445
0.0049453
0.0089757
0.000158
0.00761
0.0065556
0.0116518
0.0104703
0.0034083
0.0133833
0.0011542
0.0099996
0.0014001
0.0009791
0.0002736
0.0017814
0.0010107
0.0001999
0.0006348
0.0006772
Particle
(ng/g-OC)
0
10.00643
0.71635
0.75432
5.29398
1.76031
1.41452
3.67665
2.76099
52.27209
1 1 .73799
13.00778
17.92995
11.19489
24.42368
40.15014
113.2876
43.29467
19.2216
69.76274
1.46919
14.55518
22.56819
0.3229
22.77225
22.77379
39.81201
64.97851
19.95935
79.90554
6.06862
82.25939
11.14243
17.12468
4.51562
30.90044
13.11631
5.46997
17.80964
18.07668
Sheboygan Reef
Pore Water
(ng/L)
0
0.0382893
0
0.0029074
0.0252991
0.0074487
0.0026843
0.018433
0.0035926
0.067541
0.0149408
0.019252
0.023162
0.0098061
0.0227162
0.0281371
0.05182
0.0239207
0.007947
0.026113
0.0006704
0.0062292
0.0178462
0.0011103
0.00995
0.007798
0.0138112
0.0130125
0.0040488
0.0151232
0.0012538
0.0127299
0.0015557
0.0009686
0.0002758
0.0018309
0.000956
0.000171
0.0005772
0.0004869
Particle
(ng/g-OC)
0
13.72811
0
1 .63073
13.84212
3.36831
1 .53063
9.51698
3.72098
71.12439
14.00947
25.29948
27.84947
13,91655
31 .70843
61 .44742
131.3768
57.22649
20.14776
102.5709
2.21588
18.33357
44.87101
2.26910
29.77426
27.0894
47.18955
80.75468
23.71038
95.42261
6.59256
104.7201
12.38059
16.94218
4.53335
31 .75848
12.40633
4.67938
16.19255
12.99829
Saugatuck
Pore Water
(ng/L)
0
0.0908558
0
0.0061499
0.0681406
0.0171774
0.0277008
0.0504009
0.0167924
0.2105729
0.0549273
0.061323
0.06441 97
0.0308258
0.0629656
0.0645118
0.1180745
0.0613122
0.0219209
0.0514251
0.0018093
0.0162384
0.0322442
0.0019127
0.0154561
0.0150233
0.029209
0.0212667
0.0083891
0.02301 93
0.0026042
0.0209239
0.0028539
0.001 6554
0.0005621
0.0030391
0.0016946
0.0002984
0.0011243
0.0012951
Particle
(ng/g-OC)
0
32.58473
0
3.44944
37.28868
7.76701
15.80006
26.02673
17.39422
221.7646
51.50962
80.17273
77.46354
43.75034
87.89781
140.8908
299.3598
146.6866
55.5777
200.6624
5.98041
47.79412
81 .07549
3.90920
46.2521
52.19077
99.80335
131.9822
49.12844
145.2468
13.69295
172.1294
22.71165
28.95361
9.23983
52.71641
21 .99273
8.16719
31 .54036
34.57076
484
-------
25-i
20-
15-
1
10-
5c
o
surface temperature
trout age class O - 1
O trout age class 2-15
bloater age class 0
- bloater age class 1-7
mysis
25 T
20-
15-
' 10-
5 x— x— x— x-
D
o a
a
x
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
16 -i
14 -
12-
o 10 -
in
£ 8-
Ol
* s-
2 -
n
A alewife age class O - 2
alewife age class 3 - 1O
/"
A A A
A
\ A
\
12-i
10 -
8-
o
1 6-
O)
CD
^ 4C1
i
2 -
n
-"•* — deepwater sculpin
O slimy sculpin
/ ^
/ ^^
0 O O O O O O \
^ \
V. ,/~Y—I— r^i f~{ . - - - - -r
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Figure 5.4.2a. Typical annual cycles of exposure temperature for Lake Michigan food webs at
Saugatuck and Sturgeon Bay.
25 -r
20-
15-
lo-
se
surface temperature
trout age class 0-1
O trout age class 2-15
I I I I 1 1 1 1 1 1—
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
— bloater age class 0
D bloater age class 1-7
mysis
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
1 1 1 1 1 1 1 r
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
T 1 1 1 1 1 1 1 r
Jan Feb Mar Apr May Jun Jul Aug Sep Oct
Nov Dec
Figure 5.4.2b. Typical annual cycles of exposure temperature for Lake Michigan food web at
Sheboygan Reef.
485
-------
25
20-
O
ra
10H
5-
surface temperature
coho salmon
—i 1 1 1 1 1 1 ' 1 i
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Figure 5.4.2c. Typical annual cycles of exposure temperature for coho salmon in Lake Michigan.
temperature, and swimming speed was conducted,
and results were reported (Lantry and Stewart, 1993;
Rudstam, 1989; Rudstam etal., 1994; Stewart etal.,
1983; Stewart and Binkowski, 1986). In general, a
fish's daily respiration rate, in g-O2/day, can be
formulated as:
R=
jPr. &vu
(5.4.6)
where a, P, p, v are species-specific empirical
constants, W is weight, and U is the swimming speed
of the fish, in cm/s.
For a given aquatic species, the swimming speed
can be expressed as a function of body weight and
water temperature:
U = co
e
*7"
(5.4.7)
where co, 5, cj) are species-specific empirical
constants.
The values of the species-specific empirical
constants used to estimate the respiration rate were
collected from literature sources (Lantry and Stewart,
1993; Rudstam, 1989; Rudstam etal., 1994; Stewart
et al., 1983; Stewart and Binkowski, 1986) and are
listed in Table 5.4.13. For slimy and deepwater
sculpin, there was insufficient information available to
generate species-specific respiration rates. As an
alternative, their respiration rates were estimated
using the generalized fish respiration equation. The
constants used for the calculation of their respiration
rates were also given in the table.
In this study, a value of 13.56 kJ/g-O2 (Elliott and
Davison, 1975; Brafield and Llewellyn, 1982; Crisp,
1984) was used as the respiratory energy equivalent,
or oxycalorific coefficient, for converting oxygen
respiration to energy utilized by fish.
For zooplankton, a simple equation was used to
estimate its respiration, in kJ/gwet/day, as a function
of water temperature (Connolly et al., 1992):
fl=0.60epr
(5.4.8)
5.4.3.2 Respiration Rates Adjusted for Specific
Dynamic Action (SDA)
The respiration rate estimated with Equation 5.4.6
represents the average energy requirement for the
resting metabolism of a fish. It has been reported
that there is an increase in respiration rate for a
recently fed fish (Kayser, 1963). The additional
respiration activity is often referred to as Specific
Dynamic Action (SDA). The origin of the extra
respiration is believed to be due to the energy
necessary for the digestion of ingested foods, the
absorption of nutrients, the deaminization of amino
acids, and the synthesis of the products of
nitrogenous excretion. In homothermic animals, it
has been shown that SDA represents 30% of the
caloric content of the ingested protein, 13% for a
lipid, and 5% for a carbohydrate (Lucas, 1996). Due
to the difficulty in experimentally discriminating SDA
from additional respiration associated with
excitement and activity with feeding, different SDA
486
-------
Table 5.4.13. Bioenergetic Parameters of Lake Michigan Fishes
Parameter
a
(gOj/gwet/day)
P
P
0)
5
4>
V
Mysis
0.00182
-0.161
0.0752
0
0
0
0
Slimy
Sculpin
0.043*
-0.3
0.03
1.19
0.32
0.045
0.0176
Deepwater
Sculpin
0.043*
-0.3
0.03
1.19
0.32
0.045
0.0176
Alewife
0.00367
-0.2152
0.0548
5.78
-0.045
0.149
0.03
Rainbow
Smelt
0.0027
-0.216
0.036
0
0
0
0
Bloater
0.0018
-0.12
0.047
7.23
0.25
0
0.025
Lake Trout
0.00463
-0.295
0.059
11.7
0.05
0.0405
0.0232
Coho
Salmon
0.00264
-0.217
0.06818
9.7
0.13
0.0405
0.0234
*With a unit of gwet/gwet/day.
values were cited in the literature that ranged from
9% to 20% of the energy contained in the diet
(Jobling, 1981).
In this study, the SDA is modeled as a portion of a
fish's dietary ingestion. The respiration rate adjusted
for SDA can then be written as:
The final respiration rate, in kJ/day, was then
estimated as:
SDA
where
(RSDA + G • D,)
(5.4.9)
RSDA = SDA adjusted respiration rate, g-O2/day
R = resting respiration rate calculated with
empirical equations, g-O2/day
Qox = respiratory energy equivalent or oxycalorific
coefficient, kJ/g-O2
SDA = fraction of assimilated energy spent on
specific dynamic action
G = fish growth rate, 1/day
DI = energy density of the fish
-G- D,)/(1 -SDA) (5.4.10)
5.4.4 Calibrated Model Parameters
There are several constants and variables in the
model's equation whose values are either not readily
available or inconclusive. Their values were
determined through model calibration to site-specific
conditions. The calibrated parameters include food
assimilation efficiency (P) for each species or age
group, the chemical assimilation efficiency (a) for
each species or age group for each PCB congener,
the chemical relative gill transfer coefficient (EyEJ
for each species (or age group) for each PCB
congener, and the fraction of ingested energy for
SDA for each species or age group.
An acceptable value range for each of the calibrated
model parameters and its general trend for PCB
congeners or species in different trophic levels was
established based on information from the literature
and experience gained in previous modeling work.
Depending upon species and its diet, food
assimilation efficiency has a value ranging from 0.05
to 0.85 (Brocksen era/., 1 968; Brocksen and Brugge,
1 974; Elliott, 1 976; Averett, 1 969). The value for the
487
-------
chemical assimilation efficiency can vary from 0.2 to
0.8 and is reported to be correlated with the Kow value
for the chemical (Gobas, 1988). The chemical
relative gill transfer coefficient (E,/E0) ranges from 0.1
to 1.0 and is also believed to be related to Kow for the
chemical (McKim et a/., 1985). Energy fraction for
SDA has a value ranging from 0.00 to 0.20. These
data were used to guide our model calibrations for
appropriate parameterization.
References
Averett, R.C. 1969. Influence of Temperature on
Energy and Material Utilization by Juvenile Coho
Salmon. Ph.D. Thesis, Oregon State University,
Corvallis, Oregon. 74 pp.
Beeton, A.M. and J.A. Bowers. 1982. Vertical
Migration of Mysis relicta Loven. Hydrobiologia,
93(1-2):53-61.
Brafield, A.E. and M.J. Llewellyn. 1982. Animal
Energetics. Blackie and Son, Ltd., Glasgow,
Scotland. 168pp.
Brocksen, R.W., G.E. Davis, and C.E. Warren.
1968. Competition, Food Consumption, and
Production of Sculpins and Trout in Laboratory
Stream Communities. J. Wildl. Mgmt., 32(1 ):51-
75.
Brocksen, R.W. and J.P. Brugge. 1974. Preliminary
Investigation on the Influence of Temperature on
Food Assimilation by Rainbow Trout, Salmo
gairdneri Richardson. J. Fish. Biol., 6(1):93-97.
Brown, E.H., Jr., G.W. Eck, N.R. Foster, R.M.
Horrall, C.E. Coberly. 1981. Historical Evidence
for Discrete Stocks of Lake Trout (Salvelinus
namaycush) in Lake Michigan. Canadian J. Fish.
Aquat. Sci., 38(12):1747-1758.
Cho, C.Y., S.J. Slinger, and H.S. Bayley. 1982.
Bioenergetics of Salmonids Fishes: Energy
Intake, Expenditure and Productivity. Comp.
Biochem. Physiol., 73B(1):25-41.
Coordinating Committee on Great Lakes Basin
Hydraulic and Hydrologic Data. May 1992.
Coordinated Great Lakes Physical Data. U.S.
Army Corps of Engineers, Detroit, Michigan.
Crisp, D.J. 1984. Energy Flow Measurements in
Methods for the Study of Marine Benthos. In:
N.A. Holm and A.D. Mclntyre (Eds.), Methods for
the Study of Marine Benthos, IPB Handbook 16,
pp. 197-279. Blackwell Scientific Publications'
Boston, Massachusetts.
Crowder, L.B. and H.L. Crawford. 1984. Ecological
Shifts in Resource Use by Bloater in Lake
Michigan. Trans. Amer. Fish Soc., 113(6):694-
700.
Diana, J.S. 1995. Biology and Ecology of Fishes.
Biological Sciences Press, Carmel, Indiana. 441
pp.
Eadie, B.J., N.R. Morehead, and P.F Landrum.
1990. Three-Phase Partitioning of Hydrophobic
Organic Compounds in Great Lakes Waters.
Chemosphere, 20(1/2):161-178.
Eck, G.W. and L. Wells. 1986. Depth Distribution,
Diet, and Overwinter Growth of Lake Trout
(Salvelinus namaycush) in Southeastern Lake
Michigan Sampled in December 1981 and March
1982. J. Great Lakes Res., 12(4):263-269.
Eisenreich, S.J., P.O. Capel, J.A. Robbins, and R.
Bourbonniere. 1989. Accumulation and
Diagenesis of Chlorinated Hydrocarbons in
Lacustrine Sediments. Environ. Sci. Technol.,
23(9):1116-1126.
Elliott, J.M. and W. Davison. 1975. Energy
Equivalents of Oxygen Consumption in Animal
Energetics. Oecologia, 19(3):195-201.
Elliott, J.M. 1976. The Energetics of Feeding,
Metabolism and Growth of Brown Trout (Salmo
trutta L.) in Relation to Body Weight, Water
Temperature, and Ration Size. J. Anim. Ecol.,
45(3):923-948.
Elliott, R.F., P.J. Peelers, M.P. Ebener, R.W. Rybicki,
P.J. Schneeberger, R.J. Hess, J.T. Francis, G.W.
Eck, andC.P. Madenjian. 1996. Conducting Diet
Studies of Lake Michigan Piscivores - A Protocol.
U.S. Fish and Wildlife Service, Green Bay,
Wisconsin. Report Number 96-2, 38 pp.
488
-------
Elliott, R.F. and M.E. Holey. 1998. A Description of
the Diet of Lake Michigan Coho Salmon and
Collections for Contaminants Analysis as Part of
the 1994-95 Lake Michigan Mass Balance
Project. U.S. Fish and Wildlife Service, Green
Bay, Wisconsin. Report Number 98-3.
Environment Canada and U.S. Environmental
Protection Agency. 1997. State of the Great
Lakes The Year of the Nearshore. U.S.
Environmental Protection Agency, Great Lakes
National Program Off ice, Chicago, Illinois. 76pp.
Evans, M.S., M.A. Quigley, and J.A. Wojcik. 1990.
Comparative Ecology of Pontoporeia hoyi
Populations in Southern Lake Michigan: The
Profundal Region Versus the Slope and Shelf
Regions. J. Great Lakes Res., 16(1):27-40.
Flath, L.E. and J.S. Diana. 1985. Seasonal Energy
Dynamics of the Alewife in Southeastern Lake
Michigan. Trans. Amer. Fish Soc., 114(3):328-
337.
Foltz, J.W. and C.R. Norden. 1977. Seasonal
Changes in Food Consumption and Energy
Content of Smelt (Osmerus mordax) in Lake
Michigan. Trans. Amer. Fish Soc., 106(3):230-
234.
Gardner, W.S., T.F. Nalepa, W.A. Frez, E.A.
Cichocki, and P.F. Landrum. 1985. Seasonal
Patterns in Lipid Content of Lake Michigan
Macroinvertebrates. Canadian J. Fish. Aquat.
Sci., 42(11):1827-1832.
Gardner, W.S., M.A. Quigley, G.L Fahnenstiel, D.
Scavia, and W.A. Frez. 1990. Pontoporeia hoyi -
A Direct Trophic Link Between Spring Diatoms
and Fish in Lake Michigan. In: M.M. Tilzer and
C. Serruya (Eds.), Large Lakes: Ecological
Structure and Function, pp. 632-644. Springer-
Verlag, New York, New York.
Gobas, F.A.P.C., D.C.G. Muir, and D. Mackay.
1988. Dynamics of Dietary Bioaccumulation and
Faecal Elimination of Hydrophobic Organic
Chemicals in Fish. Chemosphere, 17(5):943-
962.
Greenberg, A.E., L.S. Clesceri, and A.D. Eaton
(Eds.). 1992. Standard Methods for the
Examination of Water and Wastewater, 18th
Edition. American Public Health Association,
Washington, D.C. 982 pp.
Grossnickle, N.E. 1982. Feeding Habitats of Mysis
relicta An Overview. Hydrobiologia, 93(1-
2):101-107.
Hawker, D.W. and D.W. Connell. 1988. Octanol-
Water Partition Coefficients of Polychlorinated
Biphenyl Congeners. Environ. Sci. Technol.,
22(4):382-387.
Hesselberg, R.J., J.P. Hickey, D.A. Northrupt, and
W.A. Willford. 1990. Contaminant Residues in
the Bloater (Coregonus hoyi) of Lake Michigan,
1969-1986. J. Great Lakes Res., 16(1 ):121 -129.
Janssen, J. and S.B. Brandt. 1980. Feeding
Ecology and Vertical Migration of Alewives (Alosa
pseudoharengus) in Lake Michigan. Canadian J.
Fish. Aquat. Sci., 37(2): 177-184.
Jobling, M. 1981. The Influence of Feeding on the
Metabolic Rate of Fishes: A Short Review. J.
Fish. Biol., 18(4):385-400.
Johnson, R.K. 1987. The Life History, Production
and Food Habits of Pontoporeia affinis Lindstrom
(Crustacea:Amphipoda) in Mesotrophic Lake
Erken. Hydrobiologia, 144(3):277-283.
Jude, D.J., D. Bimber, N. Thurber, F. Tesar, L.
Noguchi, P. Mansfield, H. Tin, and P. Rago.
1986. Impact of the Donald C. Cook Nuclear
Plant on Fish. In: R. Rossmann (Ed.), Impact of
the Donald C. Cook Nuclear Plant, pp. 285-351.
The University of Michigan, Ann Arbor, Michigan.
Great Lakes Research Division Publication 22.
Kayser, C. 1963. Bioenergetique. In: C. Kayser et
al. (Eds.), Physiologie, pp. 51-121. Editions
Medicales, Flammarion, Paris, France.
Lantry B.F. and D.J. Stewart. 1993. Ecological
Energetics of Rainbow Smelt in the Laurentian
Great Lakes: An Interlake Comparison. Trans.
Amer. Fish. Soc., 122(5):951-976.
489
-------
Lucas, A. 1996. Bioenergetics of Aquatic Animals.
Taylor and Francis Publishers, London, England.
169pp.
Lydy, M.J. and P.P. Landrum. 1993. Assimilation
Efficiency for Sediment Sorbed Benzo(a)pyrene
by Diporeia spp. Aquat. Toxicol., 26(3-4):209-
223.
Madenjian, C.P., T.J. DeSorcie, and P.M. Stedman.
1998a. Ontogenic and Spatial Patterns in Diet
and Growth of Lake Trout in Lake Michigan.
Trans. Amer. Fish. Soc., 127(2):236-252.
Madenjian, C.P., R.J. Hesselberg, T.J. DeSorcie, L.J.
Schmidt, P.M. Stedman, R.T. Quintal, L.J.
Begnoche, and D.R. Passino-Reader. 1998b.
Estimate of Net Trophic Transfer Efficiency of
PCBs to Lake Michigan Lake Trout From Their
Prey. Environ. Sci. Technol., 32(7):886-891.
Madenjian, C.P., T.J. DeSorcie, R.M. Stedman, E.H.
Brown, Jr., G.W. Eck, L.J. Schmidt, R.J.
Hesselberg, S.M. Chernyak, and D.R. Passino-
Reader. 1999. Spatial Patterns in PCB
Concentrations of Lake Michigan Lake Trout. J.
Great Lakes Res., 25(1):149-159.
Madenjian, C.P., R.F. Elliott, T.J. DeSorcie, R.M.
Stedman, D.V. O'Connor, and D.V. Rottiers.
2000. Lipid Concentrations in Lake Michigan
Fishes: Seasonal, Spatial, Ontogenetic, and
Long-Term Trends. J. Great Lakes Res.,
26(4):427-444.
Marzolf, G.R. 1965. Substrate Relations of the
Burrowing Amphipod Pontoporeia affinis in Lake
Michigan. Ecology, 46(5):579-592.
McKim, J., P. Schmeider, and G. Veith. 1985.
Absorption Dynamics of Organic Chemical
Transport Across Trout Gills as Related to
Octanol-Water Partition Coefficient. Toxicol.
Appl. Pharmacol., 77(1):1-10.
Mullin, M.D., C.M. Pochini, S. McGrindle, M.
Romkes, S.H. Safe, and L.M. Safe. 1984. High-
Resolution PCB Analysis Synthesis and
Chromatographic Properties of All 209 PCB
Congeners. Environ. Sci. Technol., 18(6):468-
476.
Neidermeyer, W.J. and J.J. Hickey. 1976,
Chronology of Organochlorine Compounds in
Lake Michigan Fish, 1929-1966. Pest. Monit. J.
10(3):92-95.
Norstrom, R.J., A.E. McKinnon, and A.S.W.
deFreitas. 1976. Bioenergetics-Based Model for
Pollutant Accumulation by Fish. Simulation of
PCB and Methylmercury Residue Levels in
Ottawa River Yellow Perch (Perca flavescens).
J. Fish. Res. Board Canada, 33(2):248-267.
Oliver, B.G., M.N. Charlton, and R.W. Durham.
1989. Distribution, Redistribution, and
Geochronology of Polychlorinated Biphenyl
Congeners and Other Chlorinated Hydrocarbons
in Lake Ontario Sediments. Environ. Sci.
Technol., 23(2):200-208.
Patriarche, M.H. 1980. Movement and Harvest of
Coho Salmon in Lake Michigan, 1978-1979.
Michigan Department of Natural Resources,
Lansing, Michigan. Fisheries Research Report
1889.
Peters, R.H. and J.A. Downing. 1984. Empirical
Analysis of Zooplankton Filtering and Feeding
Rates. Limnol. Oceanogr., 29(4):763-784.
Peterson, R.H., A.M. Sutterlin, and J.L. Metcalfe.
1979. Temperature Preference of Several
Species of Salmo and Salvelinus and Some of
Their Hybrids. J. Fish. Res. Board Canada,
36(9): 1137-1140.
Pothoven, S.A., G.L. Fahnenstiel, H.A. Vanderploeg,
and M. Luttenton. 2000. Population Dynamics of
Mysis relicta in Southeastern Lake Michigan,
1995-1998. J. Great Lakes Res., 26(4):357-365.
Quigley, M.A. 1988. Gut Fullness of the Deposit-
Feeder Amphipod, Pontoporeia hoyi, in
Southeastern Lake Michigan. J. Great Lakes
Res., 14(2):178-187.
Quigley, M.A. and H.A. Vanderploeg. 1991-
Ingestion of Live Filamentous Diatoms by the
Great Lakes Amphipod, Diporeia sp.: A Case
Study of the Limited Value of Gut Contents
Analysis. Hydrobiologia, 223(1 ):141-148.
490
-------
Rice, J.A. 1985. Mechanisms that Regulate Survival
of Larval Bloater Coregonus hoyi in Lake
Michigan. Ph.D. Dissertation, University of
Wisconsin, Madison, Wisconsin.
Rottiers, D.V. and R.M. Tucker. 1982. Proximate
Composition and Caloric Content of Eight Lake
Michigan Fishes. U.S. Department of the Interior,
U.S. Fish and Wildlife Service, Washington, D.C.
Technical Paper 108, 8 pp.
Rudstam, L.G. 1989. A Bioenergetic Model for
Mysis Growth and Consumption Applied to a
Baltic Population of Mysis mixta. J. Plankton
Res., 11(5):971-983.
Rudstam, L.G., P.P. Binkowski, and M.A. Miller.
1994. A Bioenergetic Model for Analysis of Food
Consumption Patterns by Bloater in Lake
Michigan. Trans. Amer. Fish. Soc., 123(3):344-
357.
Schindler, D.W., A.S. Clark, and J.R. Gray. 1971.
Seasonal Calorific Values of Freshwater
Zooplankton, as Determined with a Philipson
Bomb Calorimeter Modified for Small Samples.
J. Fish. Res. Board Canada, 28(4):559-564.
Stewart, D.J., D. Weininger, D.V. Rottiers, and T.A.
Edsall. 1983. An Energetics Model for Lake
Trout, Salvelinus namaycush: Application to the
Lake Michigan Population. Canadian J. Fish.
Aquat. Sci., 40(6):681-698.
Stewart, DJ. and F.P. Binkowski. 1986. Dynamics
of Consumption and Food Conversion by Lake
Michigan Alewives: An Energetics-Modeling
Synthesis. Trans. Amer. Fish. Soc., 115(5):643-
661.
U.S. Environmental Protection Agency. 1995. Lake
Michigan Mass Balance Project Database. U.S.
Environmental Protection Agency, Great Lakes
National Program Office, Chicago, Illinois.
U.S. Environmental Protection Agency. 1997a. Lake
Michigan Mass Balance Study (LMMB) Methods
Compendium, Volume 1: Sample Collection
Techniques. U.S. Environmental Protection
Agency, Great Lakes National Program Office,
Chicago, Illinois. EPA/905/R-97/012a, 1,440pp.
U.S. Environmental Protection Agency. 1997b. Lake
Michigan Mass Balance Study (LMMB) Methods
Compendium, Volume 2: Organic and Mercury
Sample Analysis Techniques. U.S.
Environmental Protection Agency, Great Lakes
National Program Office, Chicago, Illinois.
EPA905/R-97012b, 532 pp.
Vijverberg, J. and T.H. Frank. 1976. The Chemical
Composition and Energy Contents of Copepods
and Cladocerans in Relation to Their Size.
Freshwater Biol., 6(4):333-345.
Wells, L. 1968. Seasonal Depth Distribution of Fish
in Southeastern Lake Michigan. Fish. Bull.,
Wismer, D.A. and A. E. Christie. 1987. Temperature
Relationships of Great Lakes Fishes: A Data
Compilation. Great Lakes Fishery Commission,
Ann Arbor, Michigan. Special Publication 87-3,
165pp.
491
-------
PARTS
LM FOOD CHAIN
Chapters. Calibration
5.5.1 Introduction
Calibration is a step of model development
necessary for accurate parameterization and
simulation. Before a food web bioaccumulation
model is used to predict future contamination levels
in fish or to address other related environmental
issues, it needs to be calibrated to refine certain
species- and chemical-specific parameters to site-
specific conditions. The extensive collection in 1994
and 1995 of data on congener-specific
polychlorinated biphenyl (PCB) concentrations in fish
food webs and in water and sediment of Lake
Michigan provided an excellent data set for model
calibration.
5.5.2 Description of Process
The food web model was calibrated with site-specific
conditions for lake trout in three biota zones and for
a lake-wide coho salmon population. The
calibrations were conducted for 40 PCB congeners or
co-eluters individually. For each lake trout food web,
the 1994-1995 measured data of PCBs in water and
sediment, and temperature profiles in the associated
biota zone, were used as model inputs. They were
assumed to be representative of life-long average
exposure condition. This assumption is appropriate
because there are no congener-specific PCB
exposure concentrations available prior to 1994, the
decline in PCB concentrations in the lake has slowed
down in recent years, and post-exposure input has a
limited impact on the model output for a recent date.
The dynamic food web model was run continuously
until a steady-state was reached for model outputs.
The obtained model outputs were considered to be
the model estimates of PCB concentrations in the
fish food web in response to the exposure inputs.
The model predicted concentrations of individual
PCB congeners were then compared to the observed
PCB concentrations in the biota zone for species in
each trophic level of the food web. During the
calibration process, selected parameters (i.e., food
assimilation efficiency (P), chemical assimilation
efficiency (a), chemical relative gill transfer coefficient
(Ec/Eo), and specific dynamic action (SDA)) were
adjusted to improve agreement between model
results and measured PCB data for the food web.
The adjustments of the calibrated parameters were
constrained within the limits defined by the accepted
range of the parameters. Starting at the bottom of
the food web, parameter adjustment and refinement
was conducted for each species to identify the
optimal combination of the parameters which yielded
the best agreement between model results and field
data for all PCB congeners. This process was
repeated for all trophic levels in the food web.
The resulting calibrated parameters were then
examined for all species across trophic levels to
ensure that the parameter values among trophic
levels and among chemical hydrophobicities were
internally consistent and that their trends over trophic
levels and hydrophobicities were in agreement with
those reported in the literature. If necessary, the
calibration process was repeated by altering the
optimal combination of parameters until the
calibrated parameters agreed with generally
accepted trends.
492
-------
Similar calibrations were done for PCBs in the Lake
Michigan coho salmon. Given the available
information, it was not possible to construct a
dynamic food web structure to reflect variable diets of
the migratory coho salmon in the lake. The coho
salmon model was calibrated with three fixed food
web structures. They were constructed by combining
an average dietary composition of coho salmon with
one of the forage food web structures from the three
lake trout biota zones.
The model calibration described in this chapter was
based on observed PCB data at a single point in time
(1994-1995). This model calibration focused on
individual PCB congeners (rather than total PCBs
alone) in all age classes of the top predator as well
as their entire supporting forage base. The use of
constant exposure history, as represented by the
1994-1995 field data for PCBs in water and
sediment, in this model calibration was an
appropriate approximation. Our model test indicated
that, within a certain range, the variation in past
exposure concentrations had only minor impacts on
the model output for current contaminant levels in
fish. The uncertainty in the model calibration
associated with the constant exposure history was
well below the uncertainty from other sources, such
as variability in food web structures and PCB field
data.
5.5.3 Calibration Results
The parameter values that generated the best
agreement between modeled and measured PCB
data were considered to be the best estimates of the
calibrated parameter set for modeling PCBs in each
food web. The calibrated results for the Diporeia
submodel are listed in Table 5.5.1. Other calibrated
parameter values for each lake trout food web are
given in Tables 5.5.2 and 5.5.3. A range of values
was given in the tables for the chemical assimilation
efficiencies of fish and Mysis. They were treated as
functions of hydrophobicity of individual PCB
congeners. The correlation of the chemical
assimilation efficiency a to the hydrophobicity (or
Kow) of a PCB congener was adopted from the work
ofGobasefa/. (1988):
For
< 6: a =0.5
For logKow > 6: -1 = 5.3 (± 1 .5) • 1(r8 • Kow
a
+ 2.3 (±0.3) (5.5.1)
This relationship was selected because it offered the
best overall calibration results for congeners with
different hydrophobicity.
There are considerable variations in the reported
values of chemical assimilation efficiency at a given
logKow value (Buckman et a/., 2004; Gobas et a/.,
1988; Muir and Yarechewski, 1988; Niimi and Oliver,
1983; Stapleton et at., 2004; Thomann era/., 1992).
There are also indications that chemical assimilation
efficiency may be a function of species (Gobas era/.,
1988; Muir and Yarechewski, 1988). However,
adequate information was not available to support
derivation of a species-specific assimilation
efficiency. For simplicity, they were assumed to be
independent of the species. The chemical
assimilation efficiency used in this study are at the
low end of the literature reported values (Buckman et
al., 2004; Gobas et a/., 1988; Stapleton era/., 2004).
We believe that the lower values may better
represent chemical assimilation in the real
environment. This is because chemical assimilation
efficiencies were mostly estimated based on
laboratory studies using manufactured fish foods
spiked with contaminants. The contaminants coated
on the foods are likely to be more susceptible to
digestion and thus more available for absorption by
fish than contaminants accumulated naturally by prey
species in the lake. Therefore, the actual chemical
assimilation efficiencies for species in the real
environment may be lower than what were reported.
Table 5.5.4 gives the calibrated parameters values
for coho salmon which yielded the best overall
agreement between modeled and observed data for
all three supporting forage food webs.
The calibrated value for a particular model parameter
is apparently related to other parameter values. For
example, the estimated value of the food assimilation
efficiency, p, is largely influenced by our selection of
the chemical assimilation efficiency, a. If higher than
those expressed by Equation 5.5.1, the calibrated
values of the food assimilation efficiency listed in
493
-------
Table 5.5.1. Calibrated Parameter Values for Diporeia Submodel
Parameter
Calibrated Value
Water ventilation rate across the respiratory surface, Gw (L/day)
Food ingestion rate, Gd (g-dry/day)
Fraction of food absorbed, p
Organic carbon assimilation efficiency, a
Chemical assimilation efficiency from diet, Ed
Chemical assimilation efficiency from water, Ew
6.0E-03
1.8E-04
5%
46%
0.72
0.60
Table 5.5.2. Calibrated Model Parameters for PCBs in the Sturgeon Bay and Saugatuck Lake Trout
Food Webs
Zooplankton
Mysis
Deepwater Sculpin
Slimy Sculpin
Bloater (Age 1 -3)
Bloater (Age 4-7)
Alewife (Age 1 -2)
Alewife (Age 3-7)
Rainbow Smelt
Lake Trout (Age 1 -4)
Lake Trout (Age 5- 12)
Chemical
Assimilation
Efficiency (a)
0.15
0.50-0.22
0.50-0.22
0.50-0.22
0.50-0.22
0.50-0.22
0.50-0.22
0.50-0.22
0.50-0.22
0.50-0.22
0.50-0.22
Food
Assimilation
Efficiency (P)
0.60
0.80
0.60
0.65
0.25
0.40
0.90
0.40
0.60
0.40
0.20
Chemical Relative
Gill Transfer
Coefficient (E.7EJ
0.7
1.0
0.7
0.5
0.4
0.4
0.7
0.5
0.4
0.6
0.6
Energy Fraction for
Specific Dynamic
Action (SDA)
0.18
0.15
0.15
0.18
0.18
0.00
0.18
0.00
0.15
0.18
Table 5.5.3. Calibrated Model Parameters for PCBs in the Sheboygan Reef Lake Trout Food Web
Zooplankton
Mysis
Deepwater Sculpin
Slimy Sculpin
Bloater (Age 1-3)
Bloater (Age 4-7)
Alewife (Age 1 -2)
Alewife (Age 3-7)
Rainbow Smelt
Lake Trout (Age 1 -4)
Lake Trout (Age 5-1 2)
Chemical
Assimilation
Efficiency (a)
0.15
0.50-0.22
0.50-0.22
0.50-0.22
0.50-0.22
0.50-0.22
0.50-0.22
0.50-0.22
0.50-0.22
0.50-0.22
0.50-0.22
Food
Assimilation
Efficiency (P)
0.35
0.90
0.45
0.50
0.20
0.35
0.90
0.40
0.55
0.45
0.25
Chemical Relative
Gill Transfer
Coefficient (£,/£„)
1.0
1.0
0.8
0.7
0.3
0.4
0.7
0.5
0.3
0.7
0.8
Energy Fraction for
Specific Dynamic
Action (SDA)
0.15
0.18
0.15
0.18
0.18
0.00
0.15
0.00
0.18
0.18
494
-------
Table 5.5.4. Calibrated Model Parameters for PCBs in Lake Michigan Coho Salmon
Coho Salmon (Age 1 )
Coho Salmon (Age 2)
Chemical
Assimilation
Efficiency (a)
0.3
0.6
Food
Assimilation
Efficiency ((3)
0.8
0.6
Chemical Relative
Gill Transfer
Coefficient (£,/£„)
1
0.5
Energy Fraction for
Specific Dynamic
Action (SDA)
0.18
0.18
Tables 5.5.2 and 5.5.3 would have to be adjusted
upward. Therefore, the value of a parameter in the
tables can not be viewed or used independent of
those of other parameters.
Model parameterization is also influenced by the
quality of the available field data. For model
calibrations conducted with limited field data that
include only a few chemicals and an incomplete food
web (species or age classes), model
parameterization can be biased toward certain
species or chemical properties with which it was
calibrated. With the help of the extensive data
collection, which covers a large number of chemicals
with a wide range of hydrophobicities and a more
complete account of species and age classes of a
food web, the calibration results in Tables 5.5.1 -5.5.4
is believed to be less biased and more applicable to
a wide range of chemical contaminants and food
webs in the lake.
5.5.4 Field Data for PCBs in Fish and
Their Comparisons to Calibrated Model
Outputs
Except coho salmon, Lake Michigan fish samples
were collected in three biota zones in 1994 and 1995.
Phytoplankton and zooplankton were collected for
the same time periods in the biota zones. Plankton
samples were collected by pumping and separating
into phytoplankton and zooplankton (< 102 urn and >
102 urn, respectively). Coho salmon samples were
collected from various locations in 1994 and 1995.
Information regarding the sampling stations,
collection procedures, sample preparation, and
methods for PCB analysis are available in detail
(U.S. Environmental Protection Agency, 1997a,b).
For lake trout, samples were further classified into
age classes. The method for age classification is
available from Lake Michigan Mass Balance Study
Methods Compendium (U.S. Environmental
Protection Agency, 1997a) and Madenjian et al.
(1998a,b, 1999).
For lake trout and its forage species, PCB data
exhibited no temporal variation over the two-year
period of 1994-1995. Median values for congener-
based PCB concentrations in each age class or size
class of a species were calculated for each biota
zone. For coho salmon, PCB data showed
considerable temporal variation due to their rapid
growth. Whole lake median values for the
concentrations of individual PCB congeners in coho
salmon for different seasons (size class) were
estimated. The resultant values of the observed PCB
concentrations in Lake Michigan fish, Diporeia,
Mysis, and zooplankton in 1994-1995 are presented
in Appendix 5.5.1. This comprehensive PCB data set
made the Lake Michigan food web model calibration
probably the most complete and systematic in terms
of the completeness of the food web structure and
the range of hydrophobicities of chemical
contaminants, among reported model studies for
chemical bioaccumulation in a food web.
The agreement between simulated model outputs
and observed field data is an important measure of
the quality of the simulated food web model.
Appendix 5.5.2 illustrates overall comparison
between calibrated model results and observed
concentrations of individual PCB congeners for all
species in the three lake trout food webs. To
facilitate the comparison, the measured PCB data for
zooplankton, Mysis, and Diporeia were converted to
wet-weight basis. A dry fraction of 15% was
assumed for zooplankton and Mysis, and 20% for
Diporeia in the lake. Each data point in the plots
denotes the model result for an individual PCB
495
-------
congener and the corresponding field measurement.
For forage fish and Mysis, observed PCB data were
reported for composite samples of several age
classes. The maximum and minimum age classes
included in the composite samples were identified.
An average value of the model results for the
encompassed age classes was used to represent the
model estimate for the PCB concentration in the
composite sample and was compared with the
observed composite data. For example, bloater (>
160 mm) at Sturgeon Bay represents a composite
sample of bloater with age classes ranging from four
to seven years old. Therefore, in Appendix 5.5.1,
each measured PCB congener concentration for
bloater (> 160 mm) was plotted against an average
value of modeled concentration for age four through
age seven bloaters. The solid line in each of the
figures in Appendix 5.5.2 indicates the position of the
"perfect match" between the model simulation and
the observed data.
Among lake trout food webs in the three biota zones,
calibration results for Sturgeon Bay and Saugatuck
agree with the observed data reasonably well for
most species from zooplankton to the top predator,
as demonstrated by the strong positive correlations
between modeled and measured PCB congener
concentrations. The results indicate that the quality
of model simulations increases with the trophic level
of modeled species. This observation is consistent
with the fact that the field measurements for PCBs in
highly contaminated fish species are usually less
variable and better defined than the field PCB data
for less contaminated forage fish and invertebrates in
lower trophic levels. The apparent model biases for
overestimating or underestimating PCBs in
zooplankton, Mysis, and Diporeia may be attributed
to possible errors in the presumed values for water
content in these invertebrates used for converting
dry-weight based PCB data to wet-weight based
values.
Overall the model yielded a satisfactory result for
congener-specific PCBs in the top predator - lake
trout. For forage species which is not specifically
targeted in most previous model studies, the model
results could be improved by adjusting the chemical
assimilation efficiency individually for each species.
However, we decided to limit the parameter
adjustment to minimize the risk of turning the
calibration into a mere curve-fitting exercise.
Considering the large variability in the measured
congener-specific PCB data for water, sediment, and
organisms which were used either as exposure input
or for the comparison to the model output of the
calibration and considering the constraints imposed
on congener-specific model parameters, the
agreement between the calibrated and measured
congener-specific PCB data shown in Appendix 5.5.2
are remarkable.
The calibrated parameter values which result in good
fits for fish PCB data at Saugatuck and Sturgeon Bay
did not yield good model results for Sheboygan Reef
fishes in comparison with the observed data. In
order to improve the agreement between model
results and the observed fish PCB data for the
Sheboygan Reef biota zone, a different set of
calibrated parameter values was required. The
parameter values calibrated specifically for
Sheboygan Reef are given in Table 5.5.3. After the
additional parameter refinement, satisfactory
agreement was obtained between the simulated and
observed PCB concentrations for the food web at
Sheboygan Reef (Appendix 5.5.2).
For the lake-wide coho salmon, there were three
calibrated model results associated with different
forage food web inputs. With a common set of
calibrated model parameters (Table 5.5.4), each of
the calibrated model results agreed reasonably with
the observed data for coho salmon. As an example,
calibrated results associated with the Saugatuck
forage food web are compared in Figure 5.5.1 with
the observed PCB data for coho salmon at different
life stages. The figure shows that except for the
second year coho salmon in spring (April-May), the
calibrated model results agree reasonably well with
the observed PCB data for coho salmon over the
season. The discrepancies for the second year coho
salmon in spring probably result from a mis-
characterization of the coho salmon growth curve
(Table 5.4.9b). Due to large variability in fish weight
at a given age and a gap in weight data collection,
the estimated weight-age relationship may not
properly reflect the fish growth curve in the early days
of two year-old fish. The resulting growth rate may
be smaller than what was actually the case. A small
growth rate indicates a slow dilution process for
chemicals in fish, which results in a build-up of
chemicals in the fish and, consequently, a model
overestimate of chemicals in fish (see Figure 5.7.5).
496
-------
0.018
0.016
o hatchery coho salmon
0.025
0 0.002 0.006 0.010 0.014
measured PCB (ug/g-wet)
0.018
0.005 0.010 0.015 0.020 0.025
measured PCB (ug/g-wet)
0.06
® 0.05
•g 004-
:t
m 0.03
O
£ 0.02
s.
I °01
0
o
o age 2 coho salmon
(April - May)
0.08
0 0.01 0.02 0.03 0.04 0.05
measured PCB (pg/g-wet)
0.06
m 0.04
a.
01
0.02
o age 2 coho salmon (June
o
0.02 0.04 0.06
measured PCB (ug/g-wet)
0.08
0.12
1 0.10-
61
1 0.08-1
§ 0.06-j
a.
| 0.04-1
o 0.02-
o age 2 coho salmon
(July - October)
0.02 0.04 0.06 0.08 0.10
measured PCB (ug/g-wet)
0.12
Figure 5.5.1. Agreement between modeled and observed fish PCB concentrations in coho salmon
using Saugatuck food web (1994 and 1995).
497
-------
Further refinement of the weight-age relationship for
the fish may help reduce the discrepancies between
modeled and observed spring PCB data for two year-
old coho salmon.
Appendix 5.5.2 shows that the calibrated models
overestimated the concentrations of most PCB
congeners in young lake trout, specifically one and
two year-old lake trout at Sturgeon Bay and one and
three year-old lake trout at Saugatuck. It is possible
to improve the agreement for these young age
classes if model parameters were allowed to be
adjusted independently for individual age classes.
However, in this study, model parameters were
defined to be species-specific or assigned to age
groups (young or adult) of a species. Therefore, they
were not individually refined for each age class.
Rather, they were optimized systematically for all age
classes or combined age classes of a species. The
restriction of excessive parameter calibrations is
important to reduce the risk of the calibration process
being a mere curve-fitting exercise. The discrepancy
between the modeled and the observed PCB data for
the young lake trout does not necessarily indicate the
model's limitation. In fact, this discrepancy may be
attributed to the difference in the environmental
condition between model simulated and the real one
occupied by the young lake trout. Lake trout is a
stocked species in Lake Michigan (Holey et al.,
1995). Before it was exposed to the lake
environment and associated food webs, it was reared
in hatchery facilities around the lake (Peck, 1979;
Rybicki, 1990) and was exposed to a controlled
environment and food. It is likely that the
manufactured fish foods used in the hatchery
facilities were less PCB-contaminated than the
natural food items used in the food web models.
Therefore, the stocked young lake trout should have
lower PCB concentrations than that estimated by the
food web model. An incorporation of the exposure
environment in hatchery facilities into the current
model framework may improve the calibrations for
the young lake trout. Until then, higher predicted
PCB levels in the young lake trout are expected.
A similar argument can be made for the calibration
results of young coho salmon, another stocked
species. Figure 5.5.1 shows that, with exception of
few PCB congener data, the model results for young
coho salmon were generally higher than the
observed ones for most PCB congeners.
In order to evaluate the agreement between modeled
and observed data in relation to the hydrophobicity of
individual PCB congeners, an individual comparison
was made for each PCB congener. As an example,
the PCB congener data for all age classes of the lake
trout at Saugatuck are illustrated in Figure 5.5.2. The
results indicate that the calibrated model performed
equally well for PCB congeners over a range of
different hydrophobicities (log Kow ranges from 5.6 to
7.71). For all PCB congeners, the modeled and
observed data agreed well, taking into consideration
of the uncertainty associated with the measured PCB
data for individual congeners.
No comparison could be made of the current
calibration to other modeling studies in terms of
model performance. No similar modeling attempt
has been reported to reproduce congener-specific
PCB data for an entire aquatic food web. Most
previous calibrations were focused on total PCBs
only and were usually performed for adult predators
without consideration of model results for forage
species. While current calibrations yielded good
agreements between the simulated and observed
congener-specific PCB concentrations, it is
interesting to see how well the calibrated models
perform in terms of the total PCB concentrations,
Modeled total PCB data in this study were estimated
by summing model results for individual PCB
congeners and scaling the sum based on the ratio of
total PCBs to the sum of the targeted congeners from
the 1994-1995 observed data. For Saugatuck lake
trout, the ratio was 1.369. Figure 5.5.3 illustrates the
comparison between modeled and observed total
PCB data for all age classes of lake trout at
Saugatuck. The result indicates that the calibrated
food web model reproduces total PCB concentrations
in the lake trout and its bioaccumulation trend for the
age classes reasonably well.
References
Buckman, A.H., S.B. Brown, P.F. Hoekstra, K.R.
Solomon, and AT. Fish. 2004. Toxicokinetics of
Three Polychlorinated Biphenyl Technical
Mixtures in Rainbow Trout (Onchorynchus
mykiss). Environ. Technol. Chem., 23(7):1725-
1736.
498
-------
: :•= -
:x
001
.
its-
0.12-
;:£
004
»9*1
I. lull
r.,,,,
It
I,
• \
llill
ill
[
I
r
r
UUUUfc
llJLf
V Hi"
008
OOb-
0.0-s
002-
0.20
0.15
0.10
CO?-
„
-; =-*
B - — !?
13 ~£3S
J
age 5
- ,
LuJl
•et
e;
«ed
I
L
Ir
I
I
,il
1
t
I
i
r
age2
uJjU
frJ[,,r
0.10
:: =
0.06
0.04
^ c r "
C "i? -
0.25-
aio-
"n
age 3
age 6
t,
f,
1
il
•
1
1
r
r
i
llrlLn
^r.N^C-cc:r-^~- •xOC'.x'— i;.N,J3C'»roc™''^'^'v"^~~^'~
N^-^iT11*- — a:' — ^-r-r-cc^o .•S.T-*-^^ — ^;cr — — ,— ~.vc:— >. TT *r
^.*-*^T-^-^-— ,— r^.Ni »-x-* •*•''"- — — — -- — «-;N,:\. — •*- X
.
.— .-- ^ ^~ v. -^ ~ ^ ^_- •- — r j ,- _•
''
- ~ - c- .*, 3 r - - V
ir- V -- t>- ~- r- ^ • ~ .- c- • -- •-
^ — rj:N~-~~-' £ -- jr o> ;•
- -"'"~ *
. .
? - ^
^ 1C O
. <:
.
:-;
: :-
aos:
ageT
.irrfl
r
I
[
1
, ,
- - , ^y= ' -
ns-
0.4-
-
Unil
f,
1
,
r
~
-,. =c:/r/ee-"s"
:.4- "-sas.ee:
0.3-
0^2-
nil n:~r il
\,
I
J 5C5 * *
J!
i
r
:f
0.4-
.
ii.-lUr
0^-
,t
f,
1
.*«
Jl
1
r
frJL.n
•" ^
; t\
C3-
02-
C 1-
B0»S
•
~
nil
li
i
i
f
r
irJlif
-> e
'
1
f
i
1
1 1,1 I,. I]
06
a4
age 12
-t,M,ll
Ir
•
[
f
f
r
-
llJl-Hl
; - S £ - 5 £ " 2
Figure 5.5.2. Individual comparison between modeled and observed data for PCB congeners in lake
trout at Saugatuck (1994 and 1995).
499
-------
7.0
6r)
.u •
H 5.0
E,
& 4.0
- 3.0
3
~ 2.0-
1.0
•
| • measured
! n modeled
i
.niin
^ 2 3
r
4
5 6
7
r— i
8
9
1
0
i1
1
,
12
lake trout age class
Figure 5.5.3. Comparison between modeled and observed total PCBs for lake trout at Saugatuck (1994
and 1995).
Gobas, F.A.P.C., D.C.G. Muir, and D. Mackay.
1988. Dynamics of Dietary Bioaccumulation and
Faecal Elimination of Hydrophobic Organic
Chemicals in Fish. Chemosphere, 17(5):943-
962.
Holey, M.E., R.W. Rybicki, G.W. Eck, E.H. Brown,
Jr., J.E. Marsden, D.S. Lavis, M.L. Toneys, T.N.
Trudeau, and R.M. Horrall. 1995. Progress
Toward Lake Trout Restoration in Lake Michigan.
J. Great Lakes Res., 21 (Suppl. 1 ):128-151.
Madenjian, C.P., T.J. DeSorcie, and R.M. Stedman.
1998a. Ontogenic and Spatial Patterns in Diet
and Growth of Lake Trout in Lake Michigan.
Trans. Amer. Fish. Soc., 127(2):236-252.
Madenjian, C.P., R.J. Hesselberg, T.J. DeSorcie, L.J.
Schmidt, R.M. Stedman, R.T. Quintal, L.J.
Begnoche, and D.R. Passino-Reader. 1998b.
Estimate of Net Trophic Transfer Efficiency of
PCBs to Lake Michigan Lake Trout From Their
Prey. Environ. Sci. Technol., 32(7):886-891.
Madenjian, C.P., T.J. DeSorcie, R.M. Stedman, E.H.
Brown, Jr., G.W. Eck, L.J. Schmidt, R.J.
Hesselberg, S.M. Chernyak, and D.R. Passino-
Reader. 1999. Spatial Patterns in PCB
Concentrations of Lake Michigan Lake Trout. J.
Great Lakes Res., 25(1 ):149-159.
Muir, D.C.G. and A.L. Yarechewski. 1988. Dietary
Accumulation of Four Chlorinated Dioxin
Congeners by Rainbow Trout and Fathead
Minnows. Environ. Toxicol. Chem., 7(3):227-236.
Niimi, A.J. and B.G. Oliver. 1983. Biological Half-
Lives of Polychlorinated Biphenyl (PCB)
Congeners in Whole fish and Muscle of Rainbow
Trout (Salmo gairdneri). Canadian J. Fish.
Aquat. Sci., 40(9): 1388-1394.
Peck.J.W. 1979. Utilization of Traditional Spawning
Reefs by Hatchery Lake Trout in the Upper Great
Lakes. Michigan Department of Natural
Resources, Lansing, Michigan. Fisheries
Research Report Number 1871, 33 pp.
500
-------
Rybicki, R.W. 1990. Growth, Survival, and Straying
of Three Lake Trout Strains Stocked in the
Refuge of Northern Lake Michigan. Michigan
Department of Natural Resources, Charlevoix.
Michigan. Fisheries Research Report Number
1977.
Stapleton, H.M., R.J. Letcher, J. Li, and J.E. Baker.
2004. Dietary Accumulation and Metabolism of
Potybrominated Diphenyl Ethers by Juvenile Carp
(Cyprinus carpto). Environ. Toxicol. Chem..
23(81:1929-1946.
Thomann, R.V., J.P. Connolly, and T. Parkerton.
1992. An Equilibrium Model of Organic Chemical
Accumulation in Aquatic Food Webs With
Sediment Interaction. Environ. Toxicol. Chem.,
11(51:615-629.
U.S. Environmental Protection Agency. 1997a. Lake
Michigan Mass Balance Study (LMMB) Methods
Compendium, Volume 1: Sample Collection
Techniques. U.S. Environmental Protection
Agency, Great Lakes National Program Office.
Chicago. Illinois. EPA905 R-97,012a. 1.440pp.
U.S. Environmental Protection Agency. 1997b. Lake
Michigan Mass Balance Study (LMMB) Methods
Compendium, Volume 2: Organic and Mercury
Sample Analysis Techniques. U.S.
Environmental Protection Agency, Great Lakes
National Program Office. Chicago. Illinois.
EPA'905.'R-97/012b. 532 pp.
501
-------
PARTS
LM FOOD CHAIN
Appendix 5.5.1. PCB Concentrations
502
-------
PCB Concentrations (pg/gdry) in Lake Michigan Zooplankton (1994 and 1995)
Congener
003
8+5
012
013
15+17
16+32
018
026
018
031
033
037
042
044
049
052
56+60
066
70+76
074
77+110
081
084
085
087
089
092
099
101
105+132
118
123+149
163+138
146
151
153
170+190
172
180
187+182
208+195
196
197
201
203
Sturgeon Bay
942.42
0.00
0.00
0.00
346.18
611.41
428.48
1162.99
0.00
780.54
155.28
627.83
177.23
1551.78
1361.32
3184.89
140.42
2494.88
1819.84
485.56
4734.04
526.94
138.86
1196.67
1172.30
0.00
1717.47
2236.80
3555.53
945.97
1848.85
2055.80
5421.44
604.57
705.24
4141.67
1370.12
0.00
2233.74
1355.95
271 .63
561.42
0.00
859.29
755.51
Sheboygan Reef
4596.10
0.00
0.00
36.50
505.07
14.10
879.95
3091.17
202.80
1008.82
144.44
1055.40
196.84
2167.43
1865.86
6015.69
393.11
4837.90
2449.22
891.89
7091 .83
640.49
209.04
1693.86
1530.19
114.54
4379.51
4110.96
7212.97
575.56
4981 .37
4760.39
10186.65
2325.00
1630.44
10988.05
2069.29
89.13
7918.56
3872.70
768.08
1627.29
0.00
2013.64
2126.44
Saugatuck
2524.99
0.00
20.63
37.72
1701.43
533.16
1532.19
3287.49
275.03
2932.27
557.12
729.66
344.48
3387.58
3048.96
9735.53
430.87
5282.59
4070.87
2982.48
8637.94
929.34
569.02
1576.04
2107.52
1.98
4713.77
5052.15
9913.42
1200.71
5604.17
6475.89
13291.16
1961.19
2229.45
9058.01
2704.17
0.00
6411.65
3372.63
476.79
1231.23
6.23
2227.83
1421.58
Total PCBs (ng/gdry)
79.76
164.96
186.57
503
-------
PCB Concentrations (pg/gdry) in Lake Michigan Mys/s(1994 and 1995)
Congener
003
8+5
012
013
15+17
16+32
018
026
018
031
033
037
042
044
049
052
56+60
066
70+76
074
77+110
081
084
085
087
089
092
099
101
105+132
118
123+149
163+138
146
151
153
170+190
172
180
187+182
208+195
196
197
201
203
Sturgeon Bay
295.99
0.00
0.00
0.00
864.95
247.21
317.42
2999.53
2556.78
2962.47
572.50
642.45
900.57
2114.48
1396.40
2338.15
3878.33
17962.44
7397.18
5483.80
16483.66
1978.49
431.90
5453.35
2871.70
690.72
5534.60
8409.94
10566.23
3981 .38
12965.63
6600.91
23085.36
6185.89
1450.94
18171.41
3114.04
1653.80
9439.54
6134.45
598.11
1273.98
143.04
2629.61
1686.18
Sheboygan Reef
1442.10
0.00
0.00
0.00
764.41
365.10
335.53
2396.32
1844.79
2460.94
437.38
390.94
498.41
1423.97
1001.56
1432.18
3459.49
12566.99
5921.41
3204.08
15121.06
1797.28
462.87
4998.50
2445.71
1253.79
4544.90
7174.27
8639.79
3512.08
12239.49
5732.46
19870.08
4776.05
1198.93
15615.26
2729.24
961.35
7593.51
5843.48
577.48
1164.65
104.25
2469.69
153
Saugatuck
2776.39
0.00
0.00
7.83
727.53
676.49
387.07
3763.90
3874.53
3804.77
768.72
538.32
1219.20
2141.67
1385.91
2443.17
5055.25
15158.91
8394.57
4788.27
18968.39
1554.18
568.36
6336.71
3263.37
1422.84
6474.88
9223.89
10362.89
5012.26
13760.94
7126.46
25270.07
6196.97
1543.19
18526.82
4091.89
1845.96
10890.27
7583.53
950.25
2111.79
168.82
3660.18
2578.83
Total PCBs (ng/gdry)
268.55
231.50
304.20
504
-------
PCB Concentrations (pg/gdry) in Lake Michigan Diporeia (1994 and 1995)
Congener
003
8+5
012
013
15+17
16+32
018
026
018
031
033
037
042
044
049
052
56+60
066
70+76
074
77+110
081
084
085
087
089
092
099
101
105+132
118
123+149
163+138
146
151
153
170+190
172
180
187+182
208+195
196
197
201
203
Sturgeon Bay
3906.66
0.00
189.56
43.98
1770.95
1307.43
1409.46
7416.87
3395.46
4529.39
375.32
1809.82
1294.33
4256.19
3808.59
10389.60
5368.74
16343.52
7109.30
5273.06
22583.42
2503.48
3197.15
6488.84
4386.27
1377.35
7507.20
9411.39
16886.93
5656.26
15147.07
8145.97
23232.90
5766.45
3775.24
18210.67
4153.52
2213.68
12962.69
6036.73
1713.80
2536.71
261.67
5337.86
3508.22
Sheboygan Reef
5658.22
0.00
0.00
22.03
2020.00
1339.92
1624.00
10941.37
3496.20
4809.64
1191.56
1732.81
1711.93
4520.00
4495.64
12587.27
4930.76
18348.06
7900.30
4219.65
27073.80
2486.35
3343.36
7345.93
4808.38
2314.04
7997.86
11629.65
20088.19
8086.64
15025.20
10093.62
31096.81
7260.52
4523.58
23022.24
7085.55
3342.80
16248.07
8097.15
2549.43
3680.19
363.58
7118.87
4486.55
Saugatuck
3962.63
0.00
62.00
0.00
1768.21
718.87
1782.47
10387.19
1071.94
4855.98
1175.22
2324.90
1912.89
5510.35
5355.30
14273.38
1209.98
16440.11
9109.38
4824.45
23745.34
2144.60
3077.91
5647.89
4888.14
884.02
8354.76
10301.72
19176.48
2835.04
13256.51
11363.05
29498.80
6785.31
4541.13
20568.66
7736.04
3414.31
18870.57
8591.93
2502.72
4064.65
305.39
7602.03
5058.51
Total RGBs (ng/gdry)
334.12
438.93
458.97
505
-------
PCB Concentrations (pg/gwet) in Sturgeon Bay Forage Fish (1994 and 1995)
Congener
28+31
33
44
49
52
56+60
66
70+76
74
77
81+87
89+84+92
85
99
101
105
110
118
123
132+153
163+138
146
149
151
170+190
172
180
187+182
195
203+196
197
201
208
Alewife
(< 120 mm)
0.000
10.430
0.000
0.000
0.000
2.030
5.229
2.540
1.466
0.371
14.853
10.112
2.747
3.210
6.199
3.156
7.973
5.786
0.056
8.933
8.511
2.377
4.952
1.313
2.491
0.809
4.646
2.716
0.240
1.752
0.084
1.959
0.142
Alewife
(> 120 mm)
1 1 .309
19.914
4.674
5.844
10.254
1 1 .293
29.891
12.458
8.869
1.595
22.095
37.137
10.889
14.144
20.015
14.297
26.990
25.380
0.176
29.602
38.504
7.384
13.634
3.212
5.596
2.401
13.806
8.151
0.666
4.951
0.223
5.169
0.296
Bloater
(<160mm)
4.610
0.000
6.327
0.000
6.392
9.505
26.145
11.314
-6.745
1.283
18.612
44.757
10.867
15.814
22.667
14.575
28.566
25.432
0.198
37.441
47.722
9.073
11.978
3.860
7.643
3.447
19.987
1 1 .201
1.021
7.438
0.319
6.677
0.531
Bloater
(> 160 mm)
1 1 .290
0.000
5.516
0.000
7.248
12.013
31.444
14.319
8.675
1.772
20.619
54.165
14.158
20.217
26.797
18.511
38.224
33.211
0.308
46.953
52.778
10.518
15.353
4.453
8.045
3.597
21 .033
11.706
1.221
7.788
0.327
6.638
0.545
Deepwater
Sculpin
3.865
0.000
0.000
0.000
3.587
5.964
17.164
2.057
4.268
0.273
12.972
10.183
10.934
12.301
1.791
14.156
1.879
25.661
0.164
44.284
39.733
0.965
1.088
0.870
7.527
0.702
20.044
0.696
1.075
6.710
0.341
1.674
0.417
Slimy
Sculpin
3.750
0.000
0.000
0.000
1.504
6.902
20.223
4.867
5.411
0.589
21 .998
30.950
11.030
13.387
13.398
12.552
14.950
22.083
0.109
35.290
38.979
6.368
3.506
2.932
6.423
2.710
13.790
5.999
0.988
5.650
0.260
4.050
0.406
Rainbow
Smelt
3.897
0.000
0.000
0.000
1.681
5.929
14.597
6.381
3.869
0.561
20.743
23.046
7.338
8.558
9.145
9.082
13.985
15.399
0.117
25.625
32.502
6.106
7.258
1.490
2.648
1.392
8.513
6.541
0.374
2.532
0.159
2.389
0.181
Total PCBs
172.116
560.804
575.407
733.247
314.139
417.064 315.647
506
-------
PCB Concentrations (ng/gwet) in Sheboygan Reef Forage Fish (1994 and 1995)
Congener
28+31
33
44
49
52
56+60
66
70+76
74
77
81+87
89+84+92
85
99
101
105
110
118
123
132+153
163+138
146
149
151
170+190
172
180
187+182
195
203+196
197
201
208
Alewife
(<120mm)
3.256
0.000
0.000
3.794
7.518
4.101
11.592
5.386
2.222
0.258
22.356
18.842
3.998
5.708
10.598
6.140
13.830
8.954
0.114
12.590
18.392
3.655
6.245
1.525
3.283
1.074
5.879
3.330
0.310
2.326
0.089
2.373
0.131
Alewife
(> 120 mm)
5.826
17.442
4.725
0.000
7.477
10.283
30.267
12.757
8.767
1.014
25.641
39.098
10.782
14.739
22.411
14.964
28.398
23.092
0.251
30.020
40.612
8.056
13.816
3.123
6.858
2.643
14.764
8.193
0.756
5.615
0.226
5.539
0.317
Bloater
(<160mm)
5.176
0.000
0.000
0.000
5.680
8.582
22.764
1 1 .334
5.688
1.369
32.846
54.979
14.992
19.385
29.371
19.847
35.807
35.099
0.308
53.335
73.856
15.614
16.067
5.699
13.007
5.204
29.506
14.603
1.623
1 1 .622
0.462
10.145
0.708
Bloater
(> 160 mm)
9.765
0.000
0.000
0.000
4.368
13.409
35.035
17.053
8.894
1.980
31 .378
62.057
19.016
26.322
24.681
26.596
42.061
45.496
0.284
62.731
84.649
17.237
17.228
4.268
10.441
4.516
27.837
18.631
1.249
9.774
0.508
8.146
0.602
Deepwater
Sculpin
0.000
0.000
0.000
0.000
0.000
9.164
25.382
2.514
5.477
0.309
22.663
15.595
14.039
18.989
2.941
25.039
2.737
44.167
0.460
65.938
68.900
0.805
1.661
1.223
14.640
0.836
32.047
0.630
1.909
11.016
0.490
2.093
0.661
Slimy
Sculpin
0.000
0.000
0.000
0.000
1.013
5.943
17.985
4.395
3.853
0.448
26.423
32.187
10.972
16.932
14.064
15.741
13.300
28.068
0.187
39.911
52.639
8.398
3.357
2.748
9.105
3.151
19.841
6.081
1.225
7.622
0.285
4.599
0.501
Rainbow
Smelt
0.000
0.000
0.000
0.000
0.000
5.223
13.927
5.845
3.333
0.543
22.723
24.521
6.759
9.077
9.083
9.552
15.578
16.827
0.224
22.113
30.566
6.086
7.115
1.337
3.100
1.438
8.622
6.368
0.346
2.720
0.137
2.521
0.163
Total PCBs 220.814 546.978 748.376 831.049 501.386 419.502 300.601
507
-------
PCB
Congener
28+31
33
44
49
52
56+60
66
70+76
74
77
81+87
89+84+92
85
99
101
105
110
118
123
132+153
163+138
146
149
151
170+190
172
180
187+182
195
203+196
197
201
208
Concentrations (ng/gwet) in Saugatuck Forage Fish (1994 and 1995)
Alewife
(<120mm)
8.369
12.556
0.000
0.000
5.064
4.031
11.127
5.283
2.762
0.843
21.428
19.299
4.345
6.106
11.825
4.641
13.259
9.053
0.060
12.061
16.115
3.638
7.161
1.928
2.478
1.005
5.795
3.209
0.269
2.153
0.098
2.328
0.130
Alewife
(> 120mm)
9.774
23.427
0.000
12.753
10.687
9.631
25.752
1 1 .572
7.322
1.493
24.871
35.866
9.506
13.533
21 .757
13.427
26.335
23.145
0.169
30.510
40.579
7.868
15.348
3.534
5.940
2.545
16.046
10.416
0.713
5.531
0.229
5.925
0.279
Bloater
(< 160 mm)
3.847
0.000
0.000
0.000
16.495
6.094
16.980
9.069
4.327
1.033
22.140
43.896
9.436
14.445
23.211
13.123
24.681
24.234
0.201
37.195
49.265
10.892
13.537
4.619
9.344
3.632
22.774
12.581
1.255
9.494
0.387
8.416
0.635
Bloater
(> 1 60 mm)
8.098
0.000
0.000
0.000
1 1 .790
11.164
30.477
18.093
9.070
1.633
28.286
63.825
12.142
23.084
28.569
18.116
36.576
31.442
0.255
51 .392
67.268
13.072
18.804
5.717
1 1 .285
4.637
31 .759
16.100
1.514
10.981
0.507
9.230
0.612
Deepwater
Sculpin
0.000
0.000
0.000
0.000
0.000
4.857
13.539
1.205
3.879
0.198
17.683
5.713
8.920
12.242
1.106
14.779
1.076
26.089
0.132
42.229
40.831
0.321
1.044
0.593
13.140
0.365
33.473
0.293
2.331
12.955
0.482
1.224
0.498
Slimy
Sculpin
0.000
0.000
0.000
0.000
8.412
3.654
10.335
3.387
2.999
0.343
21.236
24.274
5.737
8.725
9.976
8.350
8.987
13.556
0.110
24.553
30.389
5.365
2.983
2.913
6.654
2.518
17.227
4.641
1.102
6.902
0.229
4.152
0.400
Rainbow
Smelt
4.240
0.000
0.000
0.000
1.777
4.761
11.979
5.760
3.151
0.593
18.614
19.974
5.141
7.006
7.163
7.826
11.775
14.153
0.084
18.611
24.013
4.155
6.525
1.213
2.594
1.072
9.206
5.999
0.307
2.541
0.122
2.252
0.141
Total PCBs
287.310
573.372
615.045
873.001
365.842
376.329
291.917
508
-------
PCB Concentrations (ng/gwet) in Sturgeon Bay Lake Trout (1994 and 1995)
Congener
28+31
33
44
49
52
56+60
66
70+76
74
77
81+87
89+84+92
85
99
101
105
110
118
123
132+153
163+138
146
149
151
170+190
172
180
187+182
195
203+196
197
201
208
Age1
0.000
0.000
1 8.226
13.725
2.865
4.367
14.793
4.675
4.097
0.266
29.668
18.243
7.875
10.914
8.835
9.559
12.077
17.712
0.094
24.818
30.316
5.660
5.013
1.784
5.711
2.126
12.361
4.102
0.686
4.655
0.150
3.556
0.278
Age 2
0.000
0.000
0.000
0.000
0.000
5.724
17.145
5.936
4.797
0.000
31 .048
27.508
8.563
11.486
13.159
8.991
14.841
16.302
0.208
22.648
29.042
5.411
5.752
1.680
4.696
2.475
13.768
4.351
0.625
4.004
0.069
3.189
0.243
Age3
10.680
0.000
0.000
5.046
13.777
13.488
41 .804
13.636
10.716
1.371
24.925
60.967
19.108
25.777
30.762
24.529
35.922
38.407
0.300
68.139
99.208
17.116
20.771
6.443
12.011
5.097
28.648
17.432
1.579
10.452
0.501
10.328
0.564
Age 4
17.650
0.000
2.880
14.347
22.469
17.320
50.468
22.306
17.430
2.445
36.077
82.217
24.593
30.332
36.979
30.827
45.971
53.027
0.584
82.362
94.453
20.982
26.291
8.044
14.740
6.625
38.273
19.547
1.989
13.647
0.793
12.528
0.771
Age5
22.534
0.000
5.227
13.267
0.000
25.674
72.777
29.391
20.678
2.605
54.880
126.234
34.299
49.448
55.655
48.503
70.921
87.062
0.722
119.892
168.222
30.064
37.733
10.979
23.259
8.950
55.861
32.642
2.608
19.335
0.984
17.812
1.101
Age 6
41.786
32.120
18.855
20.095
27.100
40.163
110.836
45.679
30.696
5.245
49.721
160.033
50.831
66.624
77.678
59.322
106.324
103.727
1.064
165.469
214.542
43.110
54.963
15.430
29.841
12.034
75.209
42.033
3.594
26.749
1.234
24.681
1.423
Age 7
46.792
42.462
24.149
39.700
42.695
39.780
118.203
47.853
34.171
4.702
70.747
168.053
51.816
74.340
79.291
66.525
109.363
123.433
1.306
181.435
220.792
47.982
57.888
16.537
35.128
12.254
80.560
43.520
3.884
29.878
1.252
24.434
1.363
Total PCBs 349.583
364.790
884.714
1287.150
1742.295
2248.412 2478.275
509
-------
PCB Concentrations (ng/gwet) in Sturgeon Bay Lake Trout (1994 and 1995)
(Continued)
Congener
28+31
33
44
49
52
56+60
66
70+76
74
77
81+87
89+84+92
85
99
101
105
110
118
123
132+153
163+138
146
149
151
170+190
172
180
187+182
195
203+196
197
201
208
Age 8
66.954
63.982
33.195
37.845
64.537
55.385
186.923
76.679
50.498
6.685
75.729
245.107
57.672
94.898
113.195
79.250
144.145
142.990
1.401
206.942
260.991
56.778
74.369
21 .626
39.045
16.711
102.704
62.193
4.410
35.589
1.692
33.695
1.824
Age 9
77.598
34.107
49.237
41.842
58.491
82.393
255.298
105.075
70.697
6.907
78.939
277.331
109.100
125.154
151.327
99.385
176.551
168.539
1.715
247.985
281 .228
69.026
81.397
21.294
38.363
20.770
115.630
68.694
4.982
34.609
1.708
32.718
1.815
Age 10
1 1 1 .520
36.697
50.701
40.070
92.921
99.851
264.540
110.659
78.345
12.407
100.210
390.988
102.649
156.480
193.178
140.080
227.895
240.937
1.792
320.117
407.877
84.552
117.008
30.955
64.018
28.197
149.134
91.501
8.105
56.418
2.338
51.387
2.758
Age 11
99.443
75.689
32.234
51.421
82.763
137.322
386.921
152.905
107.210
13.600
140.067
416.209
141.713
209.150
267.999
165.916
309.951
276.779
2.359
320.940
398.335
80.844
127.702
34.259
70.959
33.870
186.764
95.639
7.937
62.819
2.813
58.205
2.785
Age 12
107.946
53.131
28.188
62.538
87.828
73.866
234.730
95.558
66.468
10.319
96.935
284.160
89.222
151.254
160.833
133.617
232.517
226.950
1.702
304.475
304.516
90.964
113.001
32.845
55.435
26.544
156.371
71.585
7.043
51.484
2.245
48.042
2.443
Total PCBs 3395.8 4267.0 5190.4 6416.8 5024.9
510
-------
PCB Concentrations (ng/gwet) in Sheboygan Reef Lake Trout (1994 and 1995)
Congener
28+31
33
44
49
52
56+60
66
70+76
74
77
81+87
89+84+92
85
99
101
105
110
118
123
132+153
163+138
146
149
151
170+190
172
180
187+182
195
203+196
197
201
208
Age3
2.349
0.000
3.051
0.000
0.000
7.959
22.084
8.479
5.341
0.727
29.304
37.496
12.537
14.628
16.199
15.756
20.635
28.838
0.206
45.226
59.090
6.424
8.984
2.774
8.053
3.109
19.163
7.342
1.205
7.677
0.375
4.954
0.453
Age 4
6.491
0.000
0.000
0.000
4.723
1 1 .579
29.180
1 1 .284
7.176
0.869
33.709
49.209
19.777
25.622
20.698
22.053
25.731
42.093
0.270
60.176
75.884
11.309
10.106
3.524
15.352
5.254
36.244
9.645
1.721
12.473
0.540
7.097
0.552
Age 5
8.572
0.000
0.000
0.000
10.966
20.051
52.964
18.908
13.591
1.706
42.660
85.739
28.877
40.493
41.807
35.820
46.707
61.520
0.550
97.252
128.908
20.558
21.734
6.856
21.165
7.616
49.873
20.146
2.812
19.801
0.782
12.938
1.019
Age 6
13.126
0.000
0.000
14.172
12.785
20.911
57.596
26.067
15.090
2.401
49.367
104.224
30.582
40.355
51.197
34.615
64.792
60.930
0.655
87.272
116.502
25.583
27.628
7.979
18.992
7.988
48.246
23.297
2.438
17.086
0.780
14.318
0.924
Age?
21 .200
17.724
0.000
6.689
22.029
27.984
82.542
33.795
21.299
3.278
57.814
133.213
37.123
50.249
61.600
57.052
82.438
100.543
1.069
143.787
180.747
37.545
42.637
13.019
28.705
1 1 .342
74.927
33.271
3.542
26.200
1.251
21 .804
1.378
Age8
23.931
13.163
23.109
9.919
23.444
32.555
90.674
37.338
22.533
3.318
70.299
165.941
42.600
71.108
77.654
75.080
109.961
140.992
1.471
190.680
237.563
49.172
62.837
17.667
37.889
14.512
86.163
45.150
4.914
34.354
1.505
30.457
1.856
Total PCBs 544.1 749.0 1300.1 1329.4 2011.4 2662.7
511
-------
PCB Concentrations (ng/gwet) in Sheboygan Reef Lake Trout (1994 and 1995)
(Continued)
Congener
28+31
33
44
49
52
56+60
66
70+76
74
77
81+87
89+84+92
85
99
101
105
110
118
123
1 32+1 53
163+138
146
149
151
170+190
172
180
187+182
195
203+196
197
201
208
Age 9
27.117
14.768
17.491
14.438
34.453
39.777
118.889
49.819
30.417
4.220
70.967
196.518
38.322
83.588
102.514
85.615
123.866
145.182
1.521
215.050
261.633
59.306
76.345
20.782
39.775
16.805
100.669
56.780
5.210
36.921
1.643
33.076
2.148
Age 10
38.287
1 1 .304
33.339
23.805
49.661
56.678
164.240
70.778
44.126
7.203
87.231
258.625
44.303
109.036
144.378
1 1 1 .966
204.392
182.151
1.666
233.549
334.204
74.663
108.299
28.133
53.027
23.309
128.123
75.421
7.259
49.995
1.948
46.398
2.865
Age 11
59.381
23.388
42.762
31.731
62.762
73.976
222.135
86.727
57.365
8.449
109.337
314.461
67.224
144.470
168.091
170.353
217.835
303.263
2.902
385.133
478.044
109.645
131.518
36.344
82.994
29.654
207.344
85.610
10.952
77.609
3.471
64.124
3.829
Age 12
50.904
7.998
31.584
28.917
51 .654
65.390
186.011
71.353
50.362
7.179
97.456
266.142
47.548
123.786
138.410
145.071
177.645
236.468
2.261
333.006
414.488
82.588
109.975
32.945
83.730
22.048
173.536
74.777
11.943
74.577
2.697
45.504
3.766
Total PCBs
2885.367
3847.910
5291.523
4320.339
512
-------
PCB Concentrations (ng/gwet) In Saugatuck Lake Trout (1994 and 1995)
Congener
28+31
33
44
49
52
56+60
66
70+76
74
77
81+87
89+84+92
85
99
101
105
110
118
123
132+153
163+138
146
149
151
170+190
172
180
187+182
195
203+196
197
201
208
Age1
0.000
0.000
0.000
0.000
0.000
2.407
6.976
2.972
2.101
0.420
15.370
15.212
3.366
4.767
7.864
4.060
8,261
8.237
0.038
15.417
18.143
3.369
4.445
2.005
3.444
1.455
8.902
3.628
0.518
3.647
0.000
2.746
0.282
Age 2
0.000
0.000
0.000
0.000
54.702
10.121
30.022
12.541
8.940
1.597
34.205
65.689
18.363
22.423
32.706
21.851
35.693
41.458
0.192
66.905
87.664
16.388
17.108
7.014
13.279
5.627
35.430
16.926
1.730
1 1 .663
0.727
9.588
0.600
Age3
0.000
0.000
0.000
0.000
15.423
10.335
28.322
13.400
8.828
1.209
28.866
60.687
16.325
20.386
32.377
17.916
35.403
33.447
0.195
56.234
71.280
13.260
1 9.775
6.813
13.325
5,426
35.959
17.716
1.735
12.134
0.632
8.852
0.573
Age 4
12.689
0.000
7.271
0.000
21.443
16.764
50.218
20.900
14.090
1.798
33.756
98.227
27.565
34.339
45.768
34.405
52.105
65.537
0.348
104.088
131.670
22.683
29.044
10.232
18.845
8.320
51 .095
27.286
2.472
17.917
0.945
14.221
0.873
Age5
25.357
60.046
18.186
12.937
28.489
25.861
75.783
32.136
20.651
3.538
40.463
133.026
39.935
49.952
64.437
48.238
74.116
89.691
0.448
146.448
186.316
32.292
41.290
14.458
24.501
10.427
68.493
32.853
3.321
23.482
1.156
18.886
1.162
Age 6
54.836
149.814
26.458
23.683
42.646
41.315
128.176
57.787
35.901
5.481
63.679
228.572
62.355
79.670
117.938
68.026
134.279
127.994
0.733
192.622
255.139
52.359
71.993
22.469
36.805
15.972
95.069
67.401
4.614
34.081
1.664
31.062
1.718
Total PCBs 175.474 1006.024 854.026 1325.389 2048.679 3169.660
513
-------
PCB Concentrations (ng/gwet) in Saugatuck Lake Trout (1994 and 1995)
(Continued)
Congener
28+31
33
44
49
52
56+60
66
70+76
74
77
81+87
89+84+92
85
99
101
105
110
118
123
132+153
163+138
146
149
151
170+190
172
180
187+182
195
203+196
197
201
208
Age?
41.441
0.000
27.384
24.918
53.326
41 .852
134.351
55.184
40.012
5.054
62.337
223.314
59.930
92.361
117.717
90.825
121.589
159.552
0.959
277.072
350.735
52.772
80.649
27.012
55.392
19.716
142.617
81.900
7.168
51.628
2.424
39.138
2.316
Age8
58.278
79.286
38.107
54.032
97.743
65.226
186.428
85.751
50.604
6.480
89.576
315.128
72.836
117.029
175.219
108.533
181.002
182.133
1.659
276.537
371 .566
82.009
107.123
33.280
55.922
25.493
156.647
76.951
7.541
55.952
2.386
49.023
2.662
Age 9
94.278
45.341
60.763
66.320
113.608
86.046
296.478
132.429
85.699
9.185
115.261
440.209
84.631
158.320
241.748
129.450
238.491
222.638
1.817
351.994
461 .226
93.598
125.935
43.959
68.048
30.099
179.478
102.800
9.519
66.377
2.816
57.490
3.202
Age 10
89.884
24.287
57.841
63.576
133.449
86.744
293.366
124.580
84.646
9.300
124.845
483.935
85.427
177.869
275.204
144.524
271.898
263.991
2.146
380.232
532.781
119.568
163.525
48.187
95.684
38.397
249.146
124.901
13.455
99.183
3.763
73.102
4.315
Age 11
71 .240
68.407
0.000
46.492
89.570
106.207
326.176
126.473
91 .269
9.469
132.524
467.124
67.642
226.828
251 .779
273.178
273.766
396.848
1.067
487.895
651.971
116.769
168.020
53.570
133.525
39.965
315.883
162.844
18.328
124.932
5.345
81.521
5.277
Age 12
74.058
32.485
85.738
68.946
110.241
97.381
291.395
127.506
83.357
11.042
119.929
450.639
87.862
191.150
245.095
225.447
261.772
318.105
2.647
438.150
547.906
136.812
176.019
55.721
113.024
39.942
276.035
164.270
15.683
108.448
4.608
80.992
4.428
Total PCBs
3530.741
4430.807
5958.251
6542.673
7476.621
6541.367
514
-------
PCB Concentrations (ng/gwet) in Lake Michigan Coho Salmon (1994 and 1995)
Congener
28+31
33
44
49
52
56+60
66
70+76
074
77
81+87
89+84+92
85
99
101
105
110
118
123
132+153
163+138
146
149
151
170+190
172
180
187+182
195
203+196
197
201
208
Hatchery
0.000
0.000
0.000
0.000
0.000
0.831
4.261
3.380
2.179
0.000
15.588
9.848
1.986
3.236
7.712
2.426
7.763
6.489
0.000
5.912
7.482
1.148
2.887
1.015
0.831
0.278
2.231
2.239
0.098
0.717
0.013
0.670
0.060
Yearling
0.000
0.000
0.000
0.000
0.000
1.955
5.117
2.828
1.311
0.264
13.926
10.935
2.855
3.688
6.708
3.166
6.956
5.805
0.034
11.521
15.237
2.778
4.611
1.305
2.131
1.013
6.447
3.229
0.328
2.383
0.115
2.358
0.133
Age 2
(April-May)
7.456
0.000
0.000
0.000
5.398
3.726
1 1 .520
5.557
3.305
0.632
16.780
18.090
4.998
6.318
10.768
5.716
13.981
10.452
0.105
14.538
20.484
2.877
6.049
1.589
3.522
1.250
7.808
4.324
0.417
2.920
0.123
2.724
0.168
Age 2
(June)
17.375
15.158
10.925
2.036
17.627
9.774
32.110
13.159
9.079
1.436
24.814
50.450
14.463
18.208
28.445
16.229
34.774
32.501
0.331
45.758
66.101
14.276
18.890
5.109
9.169
3.738
21 .549
12.225
1.043
8.102
0.330
7.455
0.431
Age 2
(July-Oct.)
17.334
35.219
11.268
13.136
18.419
17.444
57.477
24.125
16.652
2.594
34.594
95.721
21 .408
34.905
52.062
30.831
59.180
60.284
0.427
86.036
122.707
23.141
33.776
9.346
16.075
7.271
41.654
24.890
2.038
15.307
0.614
15.002
0.667
Age 2
(November)
18.672
15.184
9.721
8.569
15.121
13.042
42.305
18.233
11.621
2.370
28.300
64.203
18.322
24.366
35.947
21.356
41.925
40.341
0.246
60.727
87.200
13.169
21.454
5.672
10.634
4.724
25.801
17.066
1.144
8.816
0.366
8.712
0.403
Total PCBs 112.920 170.560 279.089 690.301 1379.036 926.867
515
-------
PARTS
LM FOOD CHAIN
Appendix 5.5.2. Agreement Between Modeled and Observed
PCB Concentrations
516
-------
Agreement Between Modeled and Observed PCB Concentrations in Zooplankton,
Mysis, and D/pore/aat Sturgeon Bay (1994 and 1995).
0.0010
0.0002 0.0004 0.0006 0.0008 0.0010
measured PCB (pg/g-wet)
•E?
O>
m
O
0.
T3
5?
0)
•o
O
0.004
0.003
0.002
0.001
0.001 0.002 0,003
measured PCB (ug/g-wet)
0.004
•5?
O)
n.
CO
O
DL.
Q)
73
O
0.002 0.004 0.006
measured PCB (pg/g-wet)
0.008
517
-------
Agreement Between Modeled and Observed Fish PCB Concentrations in Forage
Fish at Sturgeon Bay (1994 and 1995)
O Slimy Sculpin
at Sturgeon Bay
0.01 0.02 0.03 0.04 0.05
measured PCB (ug/g-wet)
0.03
O alewife (<120)
at Sturgeon Bay
0.01 0.02
measured PCB (ug/g-wet)
0.03
0.06
0.05
o> 0 04-
O bloater (<160mm)
at Sturgeon Bay
0.01 0.02 0.03 0.04 0.05
measured PCB (ug/g-wet)
0.06
0.04
O rainbow smelt
at Sturgeon Bay
0.01 0.02 0.03
measured PCB (ug/g-wet)
0.04
I
j
co
*
O
0.06
0.05
0.04
0.03
0.02
0.01
O deepwater sculpin
at Sturgeon Bay
0
0.06
0 0.01 0.02 0.03 0.04 0.05 0.06
measured PCB (pg/g-wet)
I
CD
O
a.
I
120mm)
at Sturgeon Bay
0
0
0
n
0
0
0.01 0,02 0.03 0.04 0.05 0.06
measured PCB (ug/g-wet)
I
-S1
en
CO
O bloater (>160mm)
at Sturgeon Bay
0.01 0.02 0.03 0.04 0.05 0.06 0.07
measured PCB (ug/g-wet)
518
-------
Agreement Between Modeled and Observed Fish PCB Concentrations in Lake Trout
at Sturgeon Bay (1994 and 1995).
0.12
0.01 0.02 0.03 0,04 0.05 0.06
measured PCB (ug/g-wet)
0.02 0.04 0.06 0.08 0.10 0.12
measured PCB (|jg/g-wet)
0.12
f 0.10
0.02 0.04 0.06 0.08 0.10 0.12
measured PCB (pg/g-wet)
0.20
0.30
0.04 0.08 0.12
measured PCB (ug/g-wet)
0.04 0.08 0.12 0.16 0.20
measured PCB (ug/g-wet)
0 0.05 0.10 0.15 0.20 0.25 0.30
measured PCB (ug/g-wet)
0.40
f 0.35
1 0.30
lo.25
g 0.20
| 0.15
"5 0.10
1 0.05
0
0.10 0.20 0.30 0.40
measured PCB (ug/g-wet)
0.5
S 04
jji
D>
3 03
O
Q- 0.2
1
§ 0.1
0 0.1 0.2 0.3 0.4 0.5
measured PCB (ug/g-wet)
0.5
Z 0.4
•51
o>
^ 0.3
GO
S- 0.2
0.1 0.2 0.3 0.4 0.5
measured PC B (ug/g-wet)
8
a.
0.6
0.5
0.4
°-3
0-2
0.1
0
0 0.1 0.2 0.3 0.4 0.5
measured PCB (ug/g-wet)
0.6
0.7
06
04
0.3
0.2
0.1
0.1 0.2 0.3 0.4 0.5 0.6
measured PCB (ug/g-wet)
0.7
CO
O
Q_
•o
0.8
0.6
0.4
0.2
0.2 0.4 0.6
measured PCB (pg/g-wet)
0.8
519
-------
Agreement Between Modeled and Observed PCB Concentrations in Zooplankton,
Mysis, and Diporeia at Sheboygan Reef (1994 and 1995)
0.0035
zooplankton at
Sheboygan Reef
0 0.0005 0.0015 0.0025
measured PCB (pg/g-wet)
0.0035
0.005
Mysis at
Sheboygan Reef
0.001 0.002 0.003 0.004 0.005
measured PCB ((jg/g-wet)
diporeia at
Sheboygan Reef
0 0.001 0.003 0.005
measured PCB (ug/g-wet)
0.007
520
-------
Agreement Between Modeled and Observed Fish PCB Concentrations in Forage
Fish at Sheboygan Reef (1994 and 1995)
0.08
slimy sculpin
at Sheboygan Reef
0.10
0.02 0.04 0.06
measured PCB (ug/g-wet)
0,08
0.04
alewife (<120)
at Sheboygan Reef
0.01 O.02 0.03
measured PCB (ug/g-wet)
0.04
bloater (<160mm)
at Sheboygan Reef
0.02 0.04 O.06
measured PCB (ug/g-wet)
0.05
0.01 0.02 O.O3 O.O4
measured PCB (ug/g-wet)
0.08
0.05
i
0.08
0.06
0.04
1
0.02
0.08
0.02 0.04 0.06 0.08
measured PCB (ug/g-wet)
0.10
alewife (>120mm)
at Sheboygan Reef
0.02 0.04 0.06
measured PCB (ug/g-wet)
0.08
bloater (>160mm)
at Sheboygan Reef
0.02 0.04 0.06 0.08
measured PCB (ug/g-wet)
0.1O
521
-------
Agreement Between Modeled and Observed Fish PCB Concentrations in Lake Trout
at Sheboygan Reef (1994 and 1995)
0 0.01 0.03 0.05 0.07 0.09
measured PCB (ug/g-wet)
0.01 0.03 0.05 0.07 0.09
measured PCB (pg/g-wet)
0.25
0.04 0.08 0.12 0.16
measured PCB (ug/g-wet)
0.05 0.10 0.15 0.20 0.25
measured PCB (ug/g-wet)
0.3
0 0.1 0.2 0.3
measured PCB (ug/g-wet)
A)
0.2
CO
O
CL
O)
E
0.1 0.2 0.3 0.4
measured PCB (ug/g-wet)
035
| 0.30
-S»0.25
D)
~0.20
£ 0.15
| 0.10
| 0.05
0
0 0.05 0.15 0.25 0,35
measured PCB (ug/g-wet)
I 0.05 015 0.25 0.35
measured PCB (ug/g-wet)
0.45
0.05 0.15 0.25 0.35
measured PCB (pg/g-wet)
0.35
| 0.30
0.25
0.10
I 0.05
0
005 0.15 025 0.35
measured PCB (ug/g-wet)
522
-------
Agreement Between Modeled and Observed PCB Concentrations in Zooplankton,
Mysis, and D/pore/aat Saugatuck (1994 and 1995)
0.0010 0.0020 0.0030
measured PCB (ug/g-wet)
0.006
0 0.001 0.002 0.003 0.004 0.005 0.006
measured PCB (ug/g-wet)
0.007
ijf 0.006
-g 0.005
3.
m 0.004
O
0.003 -j
o
E
0.002
0.001
0
0 0.001 0.002 0.003 0.004 0.005 0.006 0.007
measured PCB (ug/g-wei)
523
-------
Agreement Between Modeled and Observed PCB Concentrations in Forage Fish at
Saugatuck (1994 and 1995)
0.06
0.01 0.02 0.03 0.04
measured PCB (ug/g-wet)
0.05
0.01 0.02 0.03 0.04 0.05
measured PCB (ug/g-wet)
0.06
0.03
0)
I
m
O
a.
O
0.02-
0.01
0.06
i" 0.05
0.04
alewife (>120mm)
at Saugatuck
O
O
O
_Q
0.01 0,02
measured PCB (pg/g-wet)
0.03
0.01 0.02 0.03 0.04 0.05
measured PCB (ug/g-wet)
0.06
0.10-,
bloater (<160mm)
at Saugatuck
0.01 0.02 0.03 0.04 0.05 0.06
measured PCB (ug/g-wet)
0.07
0.02 0.04 0.06 0.08
measured PCB (pg/g-wet)
0.10
0.05
0.01 0.02 0.03 0.04
measured PCB (ug/g-wet)
0,05
524
-------
Agreement Between Modeled and Observed PCB Concentrations in Lake Trout at
Saugatuck (1994 and 1995)
0.05
0.10
0.01 0.02 0.03 0.04 0.05
measured PCB (ug/g-wet)
0.10
0.02 0.04 0.06 0.08 0.10
measured PCB (ug/g-wet)
0.02 0.04 0.06 0.08 0.10
measured PCB (pg/g-wet)
0.15
2 0.10
CO
o
CL
0.05
0.25
0.05
0.10
0.15
measured PCB (pg/g-wet)
0.05 0.10 0.15 0.20 0.25
measured PCB (ug/g-wet)
0.35
I" 0.30
-0-25
0,20
0.15
0.10
0.05
0
0 0.05 0.10 0.15 0.20 025 0.30 035
measured PCB (ug'g-wet)
0.1 0.2 0,3 0.4
measured PCB (ug/g-wet)
0.5
0.6
0.5
I
0.4
S 0.3
CD
E
0.2
0.1
0.1 0.2 0.3 0.4 0.5
measured PCB (ug/g-wet)
0.6
•5"
O)
^.
CD
O
n_
0.6
0.5
0.4
0.3
0.2
0.1
0.1 0.2 0.3 0,4 0.5 0.6
measured PCB (ug/g-wet)
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0.1 0.2 0.3 0.4 0.5 0.6 0.7
measured PCB (pg/g-wet)
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
measured PCB (ug/g-wet)
0.8
0.6
99 0.4
I
0.2
0.2 0.4 0.6 0.8
measured PCB (pg/g-wet)
525
-------
PARTS
LM FOOD CHAIN
Chapter 6. Model Verification
5.6.1 Introduction
Model validation represents an independent test of a
model's ability to reproduce measured PCB
concentrations in Lake Michigan fish. Ideally, a
calibrated model should be subsequently validated
with another set of observed data that are
independent of those used in the model calibration
process. Additional sample collection and analysis
for PCB concentrations in the Lake Michigan system
is currently underway. Once completed, this work
will provide a new PCB data set necessary for the
validation of the food web models.
In this study, no particular procedures were
performed to validate the calibrated food web models
due to the limited data currently available. In fact,
subsequent validation of a calibrated bioaccumulation
model is rare because the necessary field data are
not readily available.
5.6.2 Model Applicability to Other Sites
An alternative of model evaluation is to test the
applicability of the calibrated models to other sites.
For the lake trout food web models, this test was
possible thanks to the extensive collection of PCB
data at the three biota zones for this project.
Because the lake trout food webs consist of the same
predator and prey species among the three biota
zones, the model parameters required for simulating
bioaccumulation in the food webs should be identical
at these three biota zones. Therefore, model
parameters calibrated for one biota zone could be
applied to model food web bioaccumulation at
another biota zone. The agreement between the
model results and the observed PCB data for the
second biota zone could then be evaluated for the
verification of the calibrated model parameters.
Apparently, a properly calibrated model parameter
set should be able to generate good fits for both biota
zones. This exercise could serve as an independent
test of how well the calibrated model represents fish
bioaccumulation under different environmental
characteristics and food web structures.
The calibration results presented in Chapter 5 show
that the lake trout food webs at Sturgeon Bay and
Saugatuck share a common set of parameter values
(Table 5.5.2). The identical parameter values for
Sturgeon Bay and Saugatuck suggested that the
model parameters calibrated for Sturgeon Bay could
be satisfactorily applied to model the Saugatuck food
web, and vice-versa. In other words, the model
parameters (Table 5.5.2) calibrated with observed
data for Sturgeon Bay have been in effect verified
with the observed data for the Saugatuck biota zone.
Or conversely, the model parameters can be viewed
as having been calibrated with the Saugatuck data
and verified by the observed data for the Sturgeon
Bay.
It should be pointed out that the identical calibrated
parameter values obtained for the Sturgeon Bay and
Saugatuck food webs were not achieved under a
"blind test". In this sense, the applicability to both
Sturgeon Bay and Saugatuck food webs did not
necessarily constitute a strict validation of the
calibrated models. This applicability, however, did
provide us with a certain confidence about the
model's performance at these two biota zones for its
526
-------
intended purpose, which was to establish quantitative
linkage between PCB levels in fish and exposure.
The food web model calibrated for Sheboygan Reef
had a unique set of parameter values (see Table
5.5.3). This indicated that the calibrated model for
this biota zone could not be validated with the
observed data from Sturgeon Bay or Saugatuck. The
reasons for this discrepancy were unclear. The
mismatch in sampling locations for the observed PCB
data for forage species and other food web
components may be one explanation. Due to
difficulty in collecting fish samples at Sheboygan
Reef, the forage fish samples were collected near
Port Washington instead of from the biota zone. It is
possible that the observed PCB data for the forage
fish species did not represent the actual
contamination levels at Sheboygan Reef. Therefore,
the food web model calibrated with the data may not
be optimized properly.
527
-------
PARTS
LM FOOD CHAIN
Chapter 7.
Uncertainty
Model Sensitivity and
5.7.7 Introduction
Sensitivity analysis was conducted to study how
variations in different model parameters affect the
model estimates for polychlorinated biphenyls (PCBs)
in a fish food web. The purpose of this analysis was
to determine which model parameters had the
greatest influence on model simulations for PCB
bioaccumulation in the food web. The results were
very useful for guiding the calibration of the models.
In fact, sensitivity analysis was performed repeatedly
during the model calibration through the trophic
levels and for each of the calibrated parameters in
order to properly direct the calibration effort.
No model uncertainty analysis has been performed
to estimate the uncertainty in the model output due to
the variability in model parameters and other input
variables. There are many uncertain aspects in a
model that affect the model output. For a well-
calibrated model, uncertainties in calibrated model
parameters have a very limited effect on the overall
model uncertainty. This is because the influences of
uncertainty in individual model parameters has been
cancelled out by the model calibration process. The
effect of a change in one parameter can be balanced
by appropriate adjustment to the other (i.e.,
compensating effects). Therefore, for a calibrated
model, the major sources of model uncertainty are
likely from other aspects, such as the quality of the
field data used to guide the model calibration, the
simplified model representations of fish behavior,
food web structure and environmental conditions,
and approximations made during model construction.
It is difficult to adequately quantify these sources of
uncertainty. In addition, the reliability of model
results for chemical bioaccumulation is associated
with individual model applications and, to a large
extent, is predominated by the uncertainty in the
input of exposure chemical concentrations whose
reliability is often not adequately defined. Any
applications of a model beyond the domain for which
the model is calibrated will also likely increase the
uncertainty associated with the model results. It is,
therefore, not an easy task to perform an uncertainty
analysis to define the expected distribution of model
outputs for the fish models.
In this study, model sensitivity to selected parameters
and input variables were quantitatively assessed in
terms of their relative impacts on the output of the
model. The results are presented in this chapter to
illustrate the potential range of uncertainty of the
model in association with different model parameters
and input variables. In addition, by identifying the
most sensitive input variables, the results can also be
useful to guide the effort for effective reduction of
model uncertainty.
5.7.2 Sensitivity Analysis
Model outputs for PCB bioaccumulation in the lake
trout food web at Saugatuck, Lake Michigan, were
used as an example for the model sensitivity
analysis. A total of seven input parameters were
assessed for their influence on model outputs for
PCBs in the top predators of the food web. These
included four calibrated model parameters:
528
-------
• Chemical assimilation efficiency (a);
• Food assimilation efficiency (P);
• Chemical relative gill transfer coefficient (E,/E0);
• The fraction of ingested energy for specific
dynamic action (SDA);
and three other input variables:
• Fish growth rate;
• Octanol-water partition coefficient K,^,;
• Fish diet.
Most of these parameter inputs are species-specific.
For simplicity, sensitivity simulations were performed
for input parameters associated with the top predator
(lake trout) only. This is because the involvement of
input parameters for species in lower trophic levels
complicates the sensitivity analysis. Due to feeding
interactions, an input parameter for a species in
lower trophic levels will impact not only the model
outputs for that particular species but also the model
outputs for species in upper trophic levels. The
quantification of this type of across trophic levels
propagation and aggregation of the impact of a
parameter is strongly dependent on the feeding
relationships between the species and its predators.
In order to properly isolate the impacts of individual
parameters on model output, all parameters in lower
trophic levels were fixed at their nominal values and
the parameters associated with lake trout only were
adjusted for the sensitivity simulations.
The calibrated food web model for lake trout at
Saugatuck was used to conduct model sensitivity
analysis. The analysis was performed by running the
model with the modified values of a tested input
parameter to calculate the steady-state PCB
concentrations in the fish, while the other parameters
were fixed at their calibrated values. PCB
concentration changes in the lake trout (age four)
relative to the calibrated model results are used to
illustrate the sensitivity of modeled PCB data to the
change of a particular input parameter.
5.7.2.1 Chemical Assimilation Efficiency (a)
The chemical assimilation efficiency for fish is
believed to be correlated to the octanol-water
partition coefficient (KoJ of a chemical. In this study,
the correlation between chemical assimilation
efficiency and the K^ value reported by Gobas et al.
(1988) was used in the calibrated model. The values
estimated by this correlation are generally at the
lower end of the reported chemical assimilation data.
For sensitivity analysis, the chemical assimilation
efficiency of the lake trout for all PCB congeners was
increased by 20%. This adjustment of parameter
input essentially increases the chemical intake of a
fish by 20%. The result of a sensitivity simulation
indicates that, except for a few low K^, congeners,
PCBs in lake trout increase by 20% as a result of a
similar increase in chemical assimilation efficiency.
Figure 5.7.1 reflects the proportional increase in
computed PCBs in lake trout responding to a 20%
increase in the values of chemical assimilation
efficiency.
5.7.2.2 Food Assimilation Efficiency (3)
The calibrated values of food assimilation efficiency
for lake trout are related to selection of the values of
chemical assimilation efficiency (a). For sensitivity
simulations, the value of the food assimilation
efficiency for young lake trout (ages 1-4) was
adjusted upward from 0.40 to 0.50. The value of
food assimilation efficiency has a direct impact on the
amount of food or chemical intake by a fish. The
increase in the food assimilation efficiency from 0.40
to 0.50 will result in a proportional reduction in the
amount of food intake necessary to meet the fish's
energy demand. A reduced food intake, 80% of the
original (0.40/0.50 = 0.80), translates into a
correspondingly lower dietary chemical intake by the
fish. The sensitivity simulation indicates that for high
KOW PCB congeners, their concentrations in the lake
trout reduced by 20% as suggested by the expected
reduction in chemical uptake by dietary route. The
concentration reduction for low Kow PCB congeners
in lake trout were not as pronounced as for high K^.,,
PCB congeners (Figure 5.7.2). For low K^ PCBs,
the dietary route of PCB intake by a fish is not as
dominant as for high Kow PCBs.
529
-------
1.4 •
1.2-
o
ro
"- 1 •
c
.0
ro 0.8 •
c
(D
o 0.6 •
c
o
o
m 0.4-
O
0.
0.2 •
n -
_
- -
1
_ i
-
1
-
-
-
;
-
_
_
-
-
-;
—
L
—
'
T -
:
— >,
_
— -
;
; !
=• -
;
Tr
— .'
_
—
-
=j -
-
.
,-
—
_
-
-
^
|
i
-r
-
_
-
€1
-
— <;
,
;
- ;
.
'',
' f
-
• -
'•
•
'
~
-
-
-
•f
-r
~
-
~
_
~
1
5=
j
1
1
1
€
W,
I
1
1
1
-
~
~
-
i
.
:
•
i
t ~
.
-
F
r|
1
'3
K
s
1
1
?
%
n-
-
—
g
1
S
1
f'l
1
?
1
3
-
"
-
1-
-
~
;
T -
-
-|-
-
_
J|
1;
|
i>
^
^
1
|
i
?
B
-
-
T
j
1
;
_
;
'
i
;
]
'
_
'
;
•
:
1
•
f
;
_
!
i
1
f
-j-
r
i
a
t
- \
I
i
|
?
i
s
|
Y
-
~
-i-
-
~~
•n-
" J. "
, i ,
?
5
£
|
|
COUJCMCOh-JDCMCDtDOOncM'SfCnCMOtOtO^-OT- NCMO>lOOl»-OOOifO'^OOmols-ON.aJMT-
/^COC'if't'^/'^flOO*"" f'l^-^'if^O^O^O "*" OO^L O OO^^"^" +OQ + CQ+ +00+ + +CD
a.
OJ
a.
uS
oj" o oo oou00
a° a. 0.0. O.O.Q.
a.
CD
Figure 5.7.1. Sensitivity of PCBs in lake trout (age four) to chemical assimilation efficiency presented
as ratios of model outputs with modified chemical assimilation efficiency to model outputs with the
calibrated chemical assimilation efficiency.
uu , cOCOcDCOCOh^COCQCOcO
OS" O OO OOO
°r ° °r °r °r tVtV£V
coco CD h-co r^cnco
25 O ii co
Figure 5.7.2. Effect of changes in food assimilation efficiency on the computed PCB data for lake trout
in Lake Michigan presented as ratios of model outputs with modified food assimilation efficiency to
model outputs with the calibrated food assimilation efficiency.
530
-------
5.7.2.3 Chemical Relative Gill Transfer
Coefficient
5.7.2.4 The Fraction of Ingested Energy for
Specific Dynamic Action (SDA)
The value of the chemical relative (to oxygen) gill
transfer coefficient (E^EJ used in the model for
young lake trout (ages 1-4) is 0.60. For the
sensitivity analysis, this value was increased to 0.80.
The corresponding changes in model results were
greatest (up to 1 6% decrease) for PCB congeners
with lower Kow values. The model results for highly
chlorinated PCB congeners were less sensitive to the
variation in the chemical relative gill transfer
coefficient (Ec/EJ. For PCB congeners with chlorine
numbers greater than seven, essentially no changes
were observed in response to the increase in this
parameter (see Figure 5.7.3). The increase in this
parameter raises the elimination rate of PCBs
through gill ventilation. The results of this sensitivity
analysis suggests that the chemical elimination via
gill ventilation is an insignificant loss mechanism
relative to growth dilution for highly chlorinated PCB
congeners.
The calibrated value of the SDA for young lake trout
(ages 1-4) is 0.15. For the sensitivity analysis, this
value was reduced to 0.10. The downward
adjustment of this parameter corresponded to
roughly a 5% reduction in chemical dietary uptake
rate based on Equation 5.7.1 (see Part 5, Chapter 4,
Section 5.4.3.2). The results of the sensitivity
simulation show that lower chlorinated PCB
congeners are not sensitive to changes in SDA. For
highly chlorinated PCB congeners, a reduction in
SDA from 0.15 to 0.10 produces about a 5%
reduction in modeled PCB concentration in the fish
(see Figure 5.7.4). This result suggests that for
these highly chlorinated congeners, the dietary
uptake may be the dominant route for chemical
accumulation in fish.
(R Qm+SDA-G-Df)/V-SDA) <5-7-1)
o r*- o h- co co T-
O> CT> CO CO O O O
CN CM CM
CO++CO+ + +DQ
CM U) CD
CO
Figure 5.7.3. Effect of changes in chemical relative gill transfer coefficient (E,JE0) on the computed
PCB data for lake trout in Lake Michigan presented as ratios of model outputs with modified chemical
relative gill transfer coefficient to model outputs with the calibrated chemical relative gill transfer
coefficient.
531
-------
1.0
„, -, CD-'-COaJCOi-COCDOr^-Oh-COCOT-
cococRO1*— •^•inioco^to^cncocoooo
CM CM CM
+ + -t CD
CM in co
CO O) O>
("I CO CO + +CO + O3 .
*£££ssg8og^
LOCO + coCDcOCDCQ
10 o o o
*V ° 'V '•V £V
CO CD CD N- CO
p
CQ CO CD CO
o o o
a. a. a.
en co
CO
Figure 5.7.4. Effect of changes in SDA on the computed PCB data for lake trout in Lake Michigan
presented as ratios of model outputs with modified SDA parameter to model outputs with the
calibrated SDA value.
5.7.2.5 Fish Growth Rate
Fish growth rates in this model were calculated
based on the age-weight relationships presented in
Tables 5.4.9a-d (Part 5, Chapter 4). For young lake
trout at Saugatuck, the average growth rate was
estimated to be about 0.0018 day"1 based on
Equation 5.7.2 (see Part 5, Chapter 4, Section
5.4.2.2). A sensitivity simulation was performed for
this parameter by adjusting the lake trout growth rate
for all age classes from the current growth rate to
zero. Figure 5.7.5 shows that the modeled PCB
concentrations for lake trout (age 4) were sensitive to
the changes in this parameter. A reduction to zero in
the fish growth rate resulted in significant increases
in modeled PCB concentrations for higher chlorinated
PCB congeners and moderate decreases for lower
chlorinated PCB congeners.
(5.7.2)
The reduction in fish growth has two opposite effects
on the modeling of chemical bioaccumulation in fish.
It lowers the estimate for chemical clearance (via
growth dilution) from the fish, which would result in a
higher model estimate for chemical bioaccumulation.
On the other hand, the reduction in fish growth also
cuts back the fish's energy demand in response to a
slow growth, which would result in a lower model
estimate for fish's food intake and associated
chemical uptake. The net effect of a zero fish growth
rate on the overall model output for PCB
bioaccumulation will depend on the individual PCB
congeners.
The elevated PCB levels in the zero-growth fish
suggested that fish growth is an important route of
elimination (via dilution) for highly chlorinated PCB
congeners. For lower chlorinated PCB congeners,
fish growth plays a minor role in the overall clearance
of PCBs from the fish in comparison with other
elimination processes. The net effect of a zero fish
growth is thus dominated by the reduction in energy
demand and consequently the lower chemical
532
-------
O 0
.=! i
ro
c
Q , r
rc Lj
L-
0)
O
O
CO
"0.5
CO I.
£Q
U C
T— I
-
f) r
t- T
3c
*• c
M C
— ^
J C
L £
M C
-
o r
— *
SI 1
-
o
D
J
_
^
!;
;
CC
V
C
a
r
-
3 C
- c
> C
. C
3 C
NJ
•0
J
•O
f
CC
C
0
C*
-'
3 C
- C
) C
. c
3 C
D C
M C
o C
-
0
M
0
O
J
4-
cr
f
C
Q
c^
>
1
)
_
V
CN
•3
t-
P"
CC
C
-
J •<
^
*
•<
-
-t c
* •>
J C
L L
t •<
- :
33 C
f I
J C
L C
i- •>
-•
i
M C
O C.
3- c
'-
3
D
D
D
a
J
cr
u.
t_
a
T
-
) C
) t
' f
- c
c
-
o
*—
3
D
J
j
•3
s
C
Q
T
'
;
• r
!r
- c
1 -
- c
5S
Q T
-
- ^
o c
J C.
L C
3" li
' ' !
- r
0 C
n
n C
'
M
33
t
0
Q
J
n
a.
C
a
ir
3
3
)
j
r
a
C
D
ir
-
) c
3 C
) C
c
) U
.
T>
33
J
L
h
n
c
I
5
-
3
)
-1
\
n
ir-
DC
L
CL
-
> (
)
- J
;-
•n
d-
^
o
M
r-|
^
V
>,
i
-/'
J
si
f
ir
CN
c>-
-
) T
) L
(
•-
n
Q
£
pn
S
cr
c^
c>-
CC
t—
-
)
)
T
cr
^
DC
L
a
-
i <•
- c
U
-
i
D
h
D
V.
~
N
a
4-
N
*—
r
> c
C
C
-',
~> f
O c
Q n
V -
- r
O C
^ -
M U
0 C
— T
-
n c
3 C
i- -
T Cl
13 C
— T
1
3
1
,
-
^
O
DD
O
0.
CO
c\i
a.
CO
a.
CO
in to CD
m
____ o
C£> CO
25 O
CD CD CC3 CD CD
O O O
a. a a.
(O r^. co
P
N- CO CD CO CO
o o o
a. a. a.
s- ci co
CO
Figure 5.7.5. Sensitivity of PCBs in lake trout (age four) to fish growth rate presented as the ratios of
model outputs with zero lake trout growth rate to model outputs with field estimated growth rate.
uptake. As a result, the model generates lower
estimates for PCB concentrations in the zero growth
fish.
5.7.2.6 Octanol-Water Partition Coefficient Kow
Reported Kow values for PCB congeners vary in the
literature (Hawker and Connell, 1988; Li et a/., 2003;
Schenker et a/., 2005; Miller et a/., 1984; Woodburn
etai, 1984). For a given PCB congener, a variation
of up to 0.5 logarithm unit is not uncommon among
reported data of the Kow. Kow values used in the
calibrations of the food web model are those
published by Hawker and Connell (1988). Eight PCB
congeners with different hydrophobicity were
selected for sensitivity analysis. Model sensitivity
simulations were conducted by reducing the log Kow
value for each PCB congener by 5%. Figure 5.7.6
indicates that PCB estimates for lake trout (age four)
are very sensitive to log Kow values. For lower
chlorinated congeners, model estimated PCB
concentrations decrease to about 'one-third in
response to a 5% reduction in log Kow values. The
reduction in model predictions is a result of an
increased gill elimination rate caused by the reduced
l°9 Kow values. The model sensitivity becomes
smaller with higher chlorinated PCB congeners. This
is consistent with the fact that gill elimination is a less
important route for higher chlorinated congeners than
lower chlorinated congeners.
5.7.2.7 Fish Diet
There are considerable variations in dietary data for
fish in Lake Michigan. The dietary compositions
used in the calibrated food web models were the
average values over a sampling time period of two
years (1994 and 1995). For lake trout at Saugatuck,
the main dietary components are alewife, rainbow
smelt, and bloater (Table 5.4.2a). Sensitivity
simulations were run by changing the dietary
composition from their original diet to 50% rainbow
smelt and 50% alewife for all age classes of lake
trout. The results of the sensitivity analyses for four
year-old lake trout are presented as an example in
Figure 5.5.7. Except for a few lower chlorinated PCB
congeners, a 10% to 20% drop in modeled PCB
concentrations occurred in response to the dietary
change.
533
-------
On
,y -
o n R
._ U.o •
"ro
*- n 7
U. / •
C
o
._ /% £•
-*-' u.o •
CT3
i_
•*± n c
c u.o •
0
g 0.4-
o
* °-3'
£ 0.2-
0.1 -
n .
-~
1
-1
*;•
. _
-
-
-
f-
JJ1-
§
1
1
. _
-
-
i
^
1
-
-
-
5
_t
;<,
ir
- _
-
-
-
—
1
- -
;,.
J.
i
1
1
;5
^
s
-
-
-
-
«.
"5
^
£
;r
i
^
I
}
I
;;
1
i-.
-
-
;-
\-
l
+
in r-
1— CO
CD CD
O O
Q_ CL
CO CO
rsi
oo CM
CM in
CO
O
in
oo
ca
O
D.
CO
0.
CD
O>
CM
CO
rj
Figure 5.7.6. Sensitivity of PCBs in lake trout (age four) to octanol-water partition coefficient (logK0J
presented as ratios of model outputs with logKow input reduced by 5% to model outputs with original
logKow values.
CD
U
\.z. -
1.1 -
0.9-
0.8-
0.7-
OR -
— r
. -
-
-
-
!
-
-
-
-
I
'fr-
it
V
c
— ,
-'
~:
*
-'
•
_
-
-
-
-
rn
-
«
I?
<
-
-
-
-
-
-
T_
-
1 ~
-,
~
i.
i,1
fc
'••)
i*
.
-
_
,
• -
_
-
_
-
'r
;
•5'
•^
_ :
_
_
,
-
- '*
- -
-I
_,
_
_
PI
j-
y.
^
^
t
.(
v
_
_
. r
_ ,
•i _
-
*
_•
:
-^
"
.
n
.; ^~
' ' * ~n " -l '
'- \\ ^
hi H
v PI •'
OQ
CQT-T-^-T-T-^^^
QCQCD+ +CQ + CO +
CM CM CM
CLQ.O
^ -- '
O ^
Q. °
CD CD
35 g
CD
0
Q_
-
OO OOO
Q.Q. 0-Q-C1-
Figure 5.7.7. Sensitivity of PCBs in lake trout (age four) to fish diet presented as ratios of model
outputs with modified fish diet to model outputs with field estimated fish diet.
534
-------
Alteration of fish diet had a direct impact on model
estimates for dietary PCB flux into the fish. A shift to
a less contaminated diet will result in a lower PCB
uptake by the fish. Because the diet of lake trout
varies with age classes, a dietary shift to rainbow
smelt and alewife exclusively will likely have different
results for different age classes of the lake trout in
terms of modeled PCB concentrations. For lake trout
at a age class whose original prey is less
contaminated than rainbow smelt or alewife, the
above dietary shift will result in a higher model
estimate of PCBs in the fish. It should be noted that
for the four-year-old lake trout, model results are not
only influenced by its own dietary change but also by
the dietary changes in younger age classes because
the initial PCB level in an older fish is determined by
the PCB levels in the fish while in previous age
classes.
The hypothetical dietary composition for sensitivity
analysis was chosen based on the observation that
Diporeia is disappearing in Lake Michigan (Landrum
ef a/., 2000; Nalepa et a/., 1998). This event will
inevitably affect the population and availability of
deepwater sculpin, slimy sculpin, and bloater that
prey on Diporeia as a primary food source. However,
this sensitivity simulation is by no means a prediction
as to what will happen to the lake trout PCBs in
association with the disappearance of Diporeia in
Lake Michigan because it is not possible to predict
the alternative diet of a fish in response to changes
in prey community composition. We may be able to
estimate the relative abundances of potential
alternative food items with the depletion of a
preferred food source. However, food selectivity of
a fish is not only a function of prey populations but
also determined by other factors, such as food
preference and the ability to actively select a
favorable food.
References
Gobas, F.A.P.C., D.C.G. Muir, and D. Mackay.
1988. Dynamics of Dietary Bioaccumulation and
Faecal Elimination of Hydrophobic Organic
Chemicals in Fish. Chemosphere, 17(5):943-
962.
Hawker, D.W. and D.W. Connell. 1988. Octanol-
Water Partition Coefficients of Polychlorinated
Biphenyl Congeners. Environ. Sci. Technol.,
22(4):382-387.
Landrum, P.F., D.C. Gossiaux, T.F. Nalepa, and D.L.
Fanslow. 2000. Evaluation of Lake Michigan
Sediment for Causes of the Disappearance of
Diporeia spp. in Southern Lake Michigan. J.
Great Lakes Res., 26(4):402-407.
Li, N.Q., F. Wania, Y.D. Lei, and G.L. Daly. 2003. A
Comprehensive and Critical Compilation,
Evaluation, and Selection of Physical-Chemicai
Property Data for Selected Polychlorinated
Biphenyls. J. Phys. Chem. Ref Data, 32(4):1545-
1590.
Miller, M.M., S. Ghodbane, S.P. Wasik, Y.B. Tewari,
and D.E. Martire. 1984. Aqueous Solubilities,
Octanol/Water Partition Coefficients, and
Entropies of Melting of Chlorinated Benzenes and
Biphenyls. J. Chem. Engin. Data, 29(2):184-190.
Nalepa, T.F., D.J. Hartson, D.L. Fanslow, G.A. Lang,
and S.J. Lozano. 1998. Declines in Benthic
Macroinvertebrate Populations in Southern Lake
Michigan, 1980-1993. Canadian J. Fish. Aquat.
Sci., 55(11):2402-2413.
Schenker, U., M. MacLeod, M. Scheringer, and K.
Hungerbuhler. 2005. Improving Data Quality for
Environmental Fate Models: A Least-Squares
Adjustment Procedure for Harmonizing
Physicochemical Properties for Organic
Compounds. Environ. Sci. Technol.,
39(21 ):8434-8441.
Woodburn, K.B., W.J. Doucette, and A.W. Andren.
1984. Generator Column Determination of
Octanol/Water Partition Coefficients for Selected
Polychlorinated Biphenyl Congeners. Environ.
Sci. Technol., 18(6):457-459.
535
-------
PARTS
LM FOOD CHAIN
Chapter 8. Model Application
5.8.1 Introduction
The fish bioaccgmulation models provide a dynamic
linkage between polychlorinated biphenyl (PCB)
levels in fish tissue and PCB concentrations in their
exposure environment. They are valuable tools for
helping us to obtain a quantitative understanding of
the bioaccumulation and trophic transfer of PCBs in
the lake. The bioaccumulation models can be
applied to estimate PCB levels in the fish given the
input of exposure concentrations in the water and
surface sediment. Besides this common application,
the models can also be used to derive estimates of
PCB concentrations in the exposure environment
based on observed fish PCB data. For example, fish
bioaccumulation models have been applied, in
conjunction with fate and transport models, to
reconstruct the time history of contaminant
concentrations in the environment (Gobas et a/.,
1995; DePintoefa/,,2003).
In this chapter, the focus of model application is
primarily on simulating future PCB levels in lake trout
on the basis of projected exposure concentrations in
sediment and overlying water that were provided by
the LM2-Toxic model (see Part 4).
5.8.2 Simulation of Fish PCB Levels
Based on Hypothetical Exposure Inputs
The food chain bioaccumulation model was
developed as a component of an integrated series of
the Lake Michigan Mass Balance models. One of the
main objectives of the models was to evaluate the
impact of PCB load reduction strategies on PCB
concentrations in the Lake Michigan ecosystem.
Several PCB load reduction scenarios were selected
for model analyses. The PCB concentrations in water
and sediment associated with each of the load
reduction scenarios were estimated by the LM2-Toxic
model (Part 4, Chapter 6). These predicted future
environmental concentrations provided a basis for
estimating corresponding fish PCB levels using the
fish bioaccumulation models.
5.8.2.1 Exposure Concentration Inputs Used for
Model Simulations
Environmental concentrations are the most critical
input when models are applied to deduce the
resulting PCB levels in fish. It cannot be over-
emphasized that it is exposure input data which
"drive" the models because, to a first approximation,
fish PCB levels are proportional to the concentrations
in its exposure environment.
Site-specific PCB concentrations in water and
sediment were provided by the LM2-Toxic model for
the Saugatuck and Sturgeon Bay biota zones.
Regional average PCB environmental concentrations
were also provided for two large areas denoted as
segment 2 and segment 3 in the LM2-Toxic model,
respectively. Segment 2 is the southeastern part of
Lake Michigan surrounding the Saugatuck biota
zone, and segment 3 is the northwestern part of the
lake surrounding the Sturgeon Bay biota zone (see
Figure 4.3.1).
For each site, a total of seven hypothetical scenarios
of long-term (1994-2055) PCB environmental
concentrations were provided as exposure input to
536
-------
the food chain models. These were generated by
the LM2-Toxic model as a quantitative prediction of
environmental concentrations under various PCB
load reduction scenarios for the Lake Michigan
ecosystem. As described in Part 4 (Chapter 6) of this
document, the seven PCB load reduction scenarios
were:
A) Constant Conditions - The measured PCB loads
(tributary load plus atmospheric dry and wet
deposition) for the LMMBP period (1994-1995),
but adjusted upward by a factor of 1.98. The
adjusted loads followed the same spatial
distribution and monthly variation patterns
established by the LMMBP measured PCB loads.
The adjusted loadings, the 1994-1995 vapor-
phase concentration, Lake Huron boundary
conditions, and all other forcing functions as
observed in 1994 and 1995 were repeated
throughout the simulation period. Sediment
burial was active as well as all other model
processes.
B) Continued Recovery (Fast) - This was the same
as Scenario "A", but atmospheric components
(vapor phase concentration, wet and dry
deposition) declined with a six-year half-life
(Hillery era/., 1997; Schneider etat., 2001), and
tributary loads declined with a 13-year half-life
(Endicott, 2005; Marti and Armstrong, 1990).
The boundary conditions at the Straits of
Mackinac declined at a rate of 0.17/year (a four-
year half-life) (Schneider et al., 2001). These
rates were applied starting on January 1, 1996.
C) Continued Recovery (Slow) - This was the same
as Scenario "A", but atmospheric components
(vapor phase concentration, wet and dry
deposition) declined with a 20-year half-life
(Buehler era/., 2002) and tributary loads declined
with a 13-year half-life. The boundary conditions
at the Straits of Mackinac declined with a four-
year half-life. These rates applied starting on
January 1, 1996.
D) No Atmospheric Deposition - This was the same
as Scenario "A", but starting on January 1,1996,
the atmospheric loads (dry and wet deposition)
were set to zero. All other forcing functions as
observed in the LMMBP period were repeated
throughout the simulation period.
E) No Tributary Loadings - This was the same as
Scenario "A", but starting on January 1,1996, all
tributary loads were set to zero. All other forcing
functions as observed in the LMMBP period were
repeated throughout the simulation period.
F) Lakewide Sediment Cleanup - This was the
same as Scenario "A", but starting on January 1,
1996, the lake-wide sediment PCB concentration
was instantaneously set to zero. All other
sediment properties remained as existed prior to
sediment clean-up. All other forcing functions as
observed in the LMMBP period and processes
were repeated throughout the simulation period.
G) No Atmospheric Deposition and No Tributary
Loadings - The loading cuts of Scenarios "D" and
"E" were combined. All other forcing functions as
observed in the LMMBP period were repeated
throughout the simulation period.
Each of the seven long-term PCB concentration
scenarios consisted of a set of congener-specific
PCB concentrations in the water column and surface
sediment as functions of time. As an example, some
congener-specific PCB exposure concentration data
for Sturgeon Bay corresponding to Scenario A
(continuation of current PCB loading) are presented
in Figures 5.8.1a and 5.8.1b) as provided by the
water quality model (LM2-Toxic).
5.8.2.2 Responses of Fish Models to Different
Exposure Inputs
The food web models calibrated for Saugatuck and
Sturgeon Bay lake trout food webs were used for the
modeling exercise in this chapter. The Saugatuck
model was applied to simulate fish bioaccumulation
in response to exposure concentrations for
Saugatuck and segment 2. The Sturgeon Bay model
was applied to simulate fish bioaccumulations in
response to exposure concentrations for Sturgeon
Bay and segment 3. We believe that these
bioaccumulation models can reasonably represent
PCB concentrations in fish as a function of exposure
concentrations in water and sediment for the
calibrated biota zones. For the large segments (two
and three) beyond the sites for which the models
were calibrated, the reliability of the models becomes
uncertain.
537
-------
participate water PCB concentration (ng/g Organic Carbon)
Figure 5.8.1 a. PCB congener-specific exposure concentrations at Sturgeon Bay predicted by LM2-
Toxic for Scenario A - PCBs in suspended particles of the water column.
538
-------
sediment PCB concentration (ng/g Organic Carbon)
_^o -- ro co -P». oi o> _».o
-k _». K> ro
O en O en
O O O 0 -* -
3 ro 4*. bi bo b K
TJ
8
CD
to
;
D O O O O O -i O OI O Ul O C
x^"
/ "O
a
CD
OJ
CO
+
CO
CD
g
ro
o
IO
Ol
g
/
/
13
O
O)
CO
to
a
_kOro-t*CT>cooro^
Figure 5.8.1 b. PCB congener-specific exposure concentrations at Sturgeon Bay predicted by LM2-
Toxic for Scenario A - PCBs in the surface sediment.
539
-------
The extrapolative applications of the calibrated
models to segments 2 and 3 were carried out on
assumptions that the average environmental
concentrations in large segments were good
representations of the exposure conditions for fish in
the segments, and that the food web structures used
in the models were still representative beyond the
specific biota zones.
However, caution should be taken while interpreting
these model outputs because assumptions may or
may not be valid. For example, there is evidence
that lake trout are usually congregated in nearshore
areas (Rybicki and Keller, 1978; Schmalz et al.,
2002). This implies that lake trout caught in any area
of Lake Michigan may have actually spent a large
portion of their living history in nearshore areas and
were exposed to PCB concentrations found there.
Therefore, interpretation of the segment-specific fish
model results need to take the actual home range of
fish into consideration.
A fundamental assumption of the fish model
application is that the food web structure, related
biological parameters (such as growth and
consumption rates for each species) and
environmental elements (such as annual
temperature) will remain the same over the entire
time period of the model simulations. Aside from this
assumption, the reliability of the fish model estimates
is primarily dependent on the quality of model inputs
for PCB concentrations in water and sediment (see
Part 4, Chapter 6).
Fish model simulations were initiated with the
assumption that PCBs in fish at day zero of the
model simulations (January 1,1994) were at steady-
state with the exposure concentration. The temporal
trend of PCB concentrations in lake trout food web
species were estimated in response to different
hypothetical long-term exposure concentrations at
Saugatuck and Sturgeon Bay, segment 2 and
segment 3. The simulation results in terms of total
PCBs in an adult lake trout (age 5.5) are presented
in Figures 5.8.2 and 5.8.3.
Lake Trout-age 5.5 at Sturgeon Bay
Lake Trout-age 5.5 at Segment 3
scenario A o historical data
scenario C A LMMBP data
scenario B
2000 2010 2020 2030 2040 2050
year
Lake Trout-age 5.5 at Saugatuck
Lake Trout-age 5.5 at Segment 2
scenario A o historical data
a*= scenario C A LMMBP data
scenario B
0
1990 2000 2010 2020 2030 2040 2050 1990
year
2000
2010
2020
year
2030 2040 2050
Figure 5.8.2. Total PCB concentrations of the lake trout in response to the exposure concentration
inputs associated with various loading scenarios.
540
-------
3-
CD
O
£L
2-
0
3-
CO
u
£L
2
n
Lake Trout-age 5.5 at Sturgeon Bay
Lake Trout-age 5.5 at Segment 3
0
1990 2000 2010 2020 2030 2040 2050 1990 2000 2010 2020 2030 2040 2050
year year
4
Lake Trout-age 5.5 at Saugatuck
Lake Trout-age 5.5 at Segment 2
0
1990 2000 2010 2020 2030 2040 2050 1990 2000 2010 2020 2030 2040 2050
year Vear
Figure 5.8.3. Total PCB concentrations of the lake trout in response to the exposure concentration
inputs associated with various loading scenarios.
The model estimates of total PCB values were
calculated based on the sum of fish model results for
individual PCB congeners. The age 5.5 lake trout
represents the average of five and six year-old lake
trout and was selected for illustration because of the
availability of long-term observed PCB data for the
lake trout at similar ages. The observed total PCB
data were also included in the figures for the
Saugatuck and Sturgeon Bay biota zones.
The results show that the fish model simulations
made for Sturgeon Bay and segment 3 have very
similar outputs. This is because environmental
exposure concentrations for these two sites were
almost identical, and the model for these two sites
used the same food web.
The results also show that the fish model outputs for
Saugatuck and segment 2 were different. Because
the model simulations for these two sites used the
same food web structure, the difference can be
attributed to the different exposure concentrations
used for their simulations.
The results further indicate that, under the same
loading scenario, projected PCB levels in lake trout
declined at a much faster rate at Saugatuck and
segment 2 than at Sturgeon Bay and segment 3.
The difference in rates of decline in fish PCB
concentration was a result of similar declining trends
in the PCB exposure concentrations used as input for
fish model simulations.
Assuming exposure inputs from the LM2-Toxic model
are reasonable depictions of future environmental
PCB concentrations in Lake Michigan under different
PCB load reduction scenarios, the fish models
predicted that total PCB concentrations in age 5.5
lake trout will level off in response to constant
external loading in 2040 (Scenario A) at all four sites.
The estimated steady-state values of total PCB
concentrations in lake trout are expected to be 0.84
ug/g-wet for Sturgeon Bay and segment 3, and 0.77
541
-------
ug/g-wet and 0.56 ug/g-wet for Saugatuck and
segment 2, respectively. The higher final value
estimated for segment 3 (0.84 (jg/g-wet) in
comparison to segment 2 (0.56 ug/g-wet) is
consistent with observations (LMMBP data) that the
lake trout food web at Sturgeon Bay has a
substantially higher bioaccumulation capacity than
the one at Saugatuck.
Based on the exposure input data provided by the
LM2-Toxic model, the fish models further suggested
that for the fast recovery scenario (Scenario B), the
targeted total PCB concentration for fish (0.075 ppm,
see Appendix 3.4.1) would be achieved in about
2036 at Sturgeon Bay and segment 3, and 2033 and
2030 at Saugatuck and segment 2, respectively. All
other PCB reduction scenarios do not achieve the
targeted PCB levels in the lake trout within the time
period of the model simulations (2055).
5.8.2.3 Discussion
It should be noted that the temporal trend of total
PCBs in lake trout for each load scenario (Figures
5.8.2 and 5.8.3) was the result of combined
contributions from the fish model results for individual
PCB congeners. For each of the PCB congeners,
the concentration temporal trend in the fish was
largely a reflection of the time functions of the
exposure concentrations for water and sediment, as
shown in Figures 5.8.1 a and 5.8.1b.
The temporal trends of total PCB concentrations in
water and sediment for various PCB load scenarios
were illustrated in Part 4 of this report (Figures 4.6.3
and 4.6.5). These lake-wide average data are a
good representation of the site-specific
environmental total PCB concentrations used for the
food web bioaccumulation model simulations.
As expected, the model results illustrate that future
PCB levels in fish are closely related to the projected
environmental concentrations. For example, the total
PCB concentrations in age 5.5 lake trout associated
with the Scenario B (Fast Continued Recovery) are
the lowest of all scenarios at the end of the model
simulation (Figure 5.8.2). This result is similar to that
for the modeled total PCB concentrations in water
and sediment (Figures 4.6.3 and 4.6.5).
Similarly, the dramatic decline at the early stage of
the model simulation in fish PCB levels associated
with the Scenario F (Sediment Cleanup) (Figure
5.8.3) is the result of the parallel trends in the total
PCB concentrations for water and sediment (Figures
4.6.3 and 4.6.5). Among the PCB load reduction
scenarios, the sediment cleanup scenario appears to
have the most immediate impact on PCB
concentrations in water and sediment and
consequently on PCB levels in the fish.
For the total PCB concentrations in the fish at the
Sturgeon Bay and Saugatuck biota zones, the model
generated temporal trends for all scenarios appears
to be in line with the expected trend inferred from the
field data. Comparing Scenarios B and C, the
Continued Recovery (Fast) (Scenario B) appears to
fit the field data better at both locations. Additional
field data are needed to confirm this observation.
It should be emphasized that the model results are
the product of the model, its structure, and the
assumptions made. Because the results of fish
bioaccumulation models are highly dependent on
exposure concentrations, the model results should be
interpreted in light of uncertainty in the exposure
predictions (Part 4, Chapter 6).
References
DePinto, J.V., W.M. Larson, J. Kaur, and J. Atkinson.
2003. LOTOX2 Model Documentation In
Support of Development of Load Reduction
Strategies and a TMDL for PCBs in Lake Ontario.
Submitted to New England Interstate Water
Pollution Control Commission, Boott Mills South,
Lowell, Massachusetts. 122 pp.
DeVault, D.S., W.A. Willford, RJ. Hesselberg, D.A.
Nortrupt, E.G.S. Roundberg, A.K. Alwan, and C.
Bautista. 1986. Contaminant Trends in Lake
Trout (Salvelinus namaycush) From the Upper
Great Lakes. Arch. Environ. Contam. Toxicol.,
15(4):349-356.
Gobas, F.A.P.C., M.N. Z'Graggen, and X. Zhang.
1995. Time Response of the Lake Ontario
Ecosystem to Virtual Elimination of PCBs.
Environ. Sci. Technol., 29(8):2038-2046.
542
-------
Golden, K.A., C.S. Wong, J.D. Jeremiason, S.J.
Eisenreich, G. Sanders, J. Hallgren, D.L.
Swackhamer, D.R. Engstrom, and D.T. Long.
1993. Accumulation and Preliminary Inventory of
Organochlorines in Great Lakes Sediments.
Water Sci. Technol., 29(8-9):19-31.
Rybicki, R.W. and M. Keller. 1978. The Lake Trout
Resource in Michigan Waters of Lake Michigan,
1970-1976. Michigan Department of Natural
Resources, Lansing, Michigan. Fisheries
Research Report Number 1863.
Schmalz, P.J., M.J. Hansen, M.E. Holey, PC.
McKee, and M.L. Toneys. 2002. Lake Trout
Movements in Northwestern Lake Michigan.
North Amer. J. Fish. Mgt., 22(3):737-749.
Schneider, A.R., H.M. Stapleton, J. Cornwell, and
J.E. Baker. 2001. Recent Declines in PAH, PCB,
and Toxaphene Levels in the Northern Great
Lakes as Determined from High Resolution
Sediment Cores. Environ. Sci. Technol.,
35(19):3809-3815.
543
-------
PARTS
COMPARISON OF MODEL RESULTS
Timothy J. Feist
Welso Federal Services, LLC
and
Kenneth R. Rygwelski
U.S. Environmental Protection Agency
Office of Research and Development
National Health and Environmental Effects Research Laboratory
Mid-Continent Ecology Division
Large Lakes and Rivers Forecasting Research Branch
Large Lakes Research Station
9311 Groh Road
Grosse lie, Michigan 48138
6.1. Summary
Three levels of models were used or developed
during the Lake Michigan Mass Balance Project
(LMMBP). The toxic fate and transport/-
bioaccumulation portion of the project included a
previously-developed model (MICHTOX) for the
coarse scale Level 1 and newly-developed models
(LM2-Toxic and LM Food Chain) for the middle
resolution Level 2. MICHTOX was used to assist in
the development of the sampling program and to
provide a screening-level assessment of the
polychlorinated biphenyl (PCB) data. The LM2-Toxic
and LM Food Chain models (LM models) were
developed during the LMMBP as part of a suite of
integrated mass balance models. While similar in
function, the Level 1 and Level 2 models had
different development histories and capabilities, and
a comparison of model results was useful to evaluate
suitability for potential future uses. This chapter
provides an examination of the similarities and
differences between the models and compares the
results of scenario predictions from both sets of
models.
The LM models provided a higher resolution
evaluation of PCB dynamics on a spatial, chemical,
and biological scale than MICHTOX. The higher
resolution of the LM models included
hydrodynamically modeled water transport, smaller
water quality segments, congener versus homolog-
level modeling, and a more detailed, data-based food
web structure. While both sets of models used
similar kinetics and forcing functions, the LM models
were more fully calibrated to process data that were
not available during the earlier development of the
MICHTOX model. The higher resolution and more
thorough calibration should allow the LM models to
provide better representation of system processes
and better predictions of the effects of future loading
changes. These features of the LM models should
also allow them to be used with minimal re-calibration
for modeling localized areas of the lake that may
have different congener composition, carbon
production, or lake trout diets.
544
-------
The comparison of model results demonstrated that,
while differences in model parameterization resulted
in different flux rates for important processes in PCB
cycling in Lake Michigan, the annual net changes in
water column concentrations were similar. The
different flux rates resulted in different steady-state
concentration predictions under a hypothetical
constant loading condition scenario. However, under
a scenario with declining loading trends based on
scientific literature, predicted concentrations in water
and fish from both sets of models converged and the
rate of decline was more important to the model
results than differences in model parameterization.
6.2 Comparison of Models
The LMMBP included three levels of models (Figure
6.1). For the PCB contaminant evaluation,
MICHTOX represented the simpler, coarse resolution
models. MICHTOX is comprised of two submodels:
a toxics fate and transport submodel, and a food
chain bioaccumulation submodel. LM2-Toxic and LM
Food Chain (LM models) represented the higher
resolution Level 2 models. This section discusses
the similarities and differences in the models.
6.2.1 Model Similarities
The toxics models were similar in a number of ways.
The MICHTOX fate and transport submodel and the
LM2-Toxic model were both based upon the United
States Environmental Protection Agency (USEPA)
WASP4 toxics model (Ambrose et al., 1988) and
used similar approaches for modeling toxic fate and
transport. While MICHTOX still possessed the
general structure of WASP4, the LM2-Toxic
computer code was completely new. MICHTOX and
LM2-Toxic contained similar kinetics and mass
transport functions. Both models included advection,
dispersion, diffusion, settling and resuspension of
toxics bound to particles, and deep sediment burial.
The air/water exchange functions were also similar in
the models. Both models used the 1994-1995
LMMBP data to develop forcing functions for tributary
loads, atmospheric deposition, and atmospheric
vapor concentrations, although LM2-Toxic simulated
individual PCB congeners while MICHTOX simulated
total PCBs divided into two homologs.
model type
Level 1 models
Level 2 models
Level 3 models
Hydrodynamic and
load models
POM
advective/
dispersive
transport and
bottom
.shear stress
Eutrophication/
sorbent dynamics
internal carbon
aggregated to
Le"vel2
transport
aggregated
to Level 2
Contaminant
transport and fate
Food web
bioaccumulation
Figure 6.1. Supporting models and links for MICHTOX and the LM models.
545
-------
The food chain bioaccumulation models were also
conceptually similar. The MICHTOX food chain
submodel and the LM Food Chain model were both
based upon the concept of mass conservation and
used similar kinetic processes, including uptake,
elimination, and concentration reduction through
growth. Both models used the LMMBP data set for
estimates of fish and invertebrate weight, growth, and
initial concentrations.
6.2.2 Model Differences
The primary differences in the models involved the
level of resolution and the degree of calibration. The
LM models provided a higher resolution evaluation of
PCB dynamics with regard to spatial scales,
hydrodynamic scales, kinetic processes, PCB forms,
and biological components. The LM models were
also rigorously calibrated using extensive field and
process data, while MICHTOX was developed at a
time when few PCB data were available.
LM2-Toxic was developed using a higher resolution
spatial grid than MICHTOX; this provided a more
accurate representation of spatially-dependent
processes and of the effects of spatial variability in
loads and concentrations. The LM2-Toxic water
quality grid was composed of five vertical segments
and 41 total water segments (Figure 4.3.1), while the
MICHTOX grid used two vertical segments and nine
total water segments (Figure 3.3.2). LM2-Toxic was
also much more highly resolved in sediment
segments having 53 sediment segments divided into
non-depositional, transitional, and depositional zones
(Figure 4.3.2). MICHTOX used six sediment
segments (Figure 3.3.2), with the area of the
segments adjusted to represent sediment focusing of
contaminants.
The hydrodynamic processes in the models also
used different resolutions. LM2-Toxic used
hydrodynamic predictions from the fine-scaled
(44,042 cell) Great Lakes version of the Princeton
Ocean Model (GL-POM) aggregated to the 41-
segment LM2-Toxic grid. Horizontal and vertical
flows were obtained from the GL-POM hydrodynamic
model and dispersion coefficients from the
temperature model (see Section 4.5.1 in Part 4,
Chapter 5 for details). MICHTOX used externally-
specified estimates of advective and dispersive
exchanges between its nine segments.
The LM models also contained higher resolution
water quality kinetics and PCB forms. They
simulated over 30 state variables (34 in LM2-Toxic,
40 in LM-Food Chain) representing PCB congeners
or co-eluting congeners. MICHTOX modeled total
PCBs in water, sediment, and the food web as two
homologs. LM2-Toxic contained three carbon
classes for partitioning of toxics: biotic carbon (BIC),
particulate detrital carbon (PDC), and dissolved
organic carbon (DOC). MICHTOX used a single
solids class with a seasonally-specified fraction of
organic carbon along with DOC. LM2-Toxic used
dynamic phytoplankton carbon production loads
estimated using the LM3-Eutro model, while
MICHTOX carbon loads were based upon a steady-
state solids balance and seasonal organic carbon
fraction calculated from historical data.
In addition to containing more detail in PCB state
variables, LM Food Chain was more detailed in its
representation of the food chain than the MICHTOX
food chain submodel. The LM Food Chain model
was based on the LMMBP data and included
phytoplankton, three invertebrates, and six fish
species. The MICHTOX food chain submodel used
an idealized food chain that included phytoplankton,
two invertebrates, and two fish species.
The models also differ in the degree of calibration
applied to them. LM2-Toxic and LM Food Chain
were thoroughly calibrated to the LMMBP data set,
and the LM2-Toxic was confirmed against a sediment
core derived PCB hindcast loading estimate. In
addition, the LM models had process data available
to reduce degrees of freedom during calibration, such
as particulate settling velocities, sediment mixing
zone thicknesses, and complete fish diet data.
MICHTOX was never fully calibrated against a PCB
data set. It was calibrated to solids transport and
plutonium data, validated to the small amount of PCB
data available when the model was originally
developed, and later compared to three hypothetical
hindcast loading scenarios and the LMMBP data set
(Endicott et a/., 2005; Endicott, 2005). At the time of
development of MICHTOX, there were little process
data available against which to constrain the model
parameterization.
546
-------
6.3 Comparison of Model Results
6.3.1 Comparison
Analyses
of Mass Budget
Mass budget analyses were conducted for PCB
simulation results for the 1994-1995 LMMBP
sampling period from both the MICHTOX fate and
transport submodel and LM2-Toxic (Part 3, Chapter
3, Section 3.3.3.3 and Part 4, Chapter 6, Section
4.6.2). For this and the following comparisons,
results from the LM models for individual congeners
were summed and then converted to total PCB
concentrations (Part 4, Chapter 6, Section 4.6.1) for
comparison with results from the MICHTOX model.
There were a number of similarities in the results
from both models. They both demonstrated a net
loss of PCBs from the system. The net loss was
calculated as the sum of fluxes out of the system
minus the sum of fluxes into the system, or (sediment
burial + gross volatilization + export to Lake Huron +
Chicago diversion) (tributary loads + atmospheric
deposition + gas absorption + input from Lake
Huron). Predicted net losses from the system were
2,673 kg/year for MICHTOX and 1,863 kg/year for
the LM2-Toxic. The net loss from the system means
that, under measured 1994-1995 loads, the system
was not at steady-state and observed concentrations
would decline. For the water column only, predicted
net losses of total PCBs were similar for both models:
182 kg/year for MICHTOX and 159 kg/year for LM2-
Toxic.
Both models show that gross volatilization, gas
absorption, resuspension, and settling are significant
mass transfer rate processes in the Lake Michigan
system (Figure 6.2). Net volatilization of PCBs was
the largest flux process. Resuspension of PCBs was
greater than settling for both models, which means
there was a net movement of PCBs from the
sediment to the water column. Export of PCBs to
Lake Huron or through the Chicago diversion was
negligible, and was not shown on Figure 6.2.
The results in Figure 6.2 highlight the differences in
parameterization of the models. PCB resuspension
and settling fluxes were much greater for MICHTOX
than for LM2-Toxic. Volatilization and gas absorption
fluxes were also higher in MICHTOX. MICHTOX was
DUUU
•o 5000
ra
01
5> 4000
X
3
u- 3000
CD
o
- 2000
1000
— I
D MICHTOX
• LM2-Toxic
1
•
1
r - 1
c e
.g o
^o ^1
:•= o
"1 rm
,— 1
L
1
1'
I~L
c: ro c 01 •« c
.0 .£ 0 c .g 0
••t; T3 "co i: ^3 "w
W CD C ^ JD 3
R ^ S S - 1
0)
.c
CL
s
3
J2
Figure 6.2. Comparison of the Lake Michigan
total PCB mass balance analyses results, 1994-
1995.
predicting more PCBs moving from the sediment to
the water column than LM2-Toxic through net
resuspension and diffusion, but was also removing
more PCBs from the water column through net
volatilization. This resulted in the net change in PCB
mass in the water column being similar between the
models. Net resuspension fluxes and diffusion in
MICHTOX were as large of a PCB source to the
water column as the external loads. They were less
than a third of the external loads for the LM2-Toxic.
LM2-Toxic has a greater sediment burial loss from
the system, but the magnitude is small compared to
net volatilization losses.
6.3.2 Comparison of Model Forecast
Scenarios
Results for the model forecast scenarios were
compared from both the fate and transport models
and the food chain/bioaccumulation models. For
comparison purposes, the LM2-Toxic segment
results were volume-weighted and averaged to
match the MICHTOX segmentation. The
bioaccumulation model results provided only an
approximate comparison between models because
the MICHTOX food chain submodel lake trout
concentrations were modeled on a segment-wide
547
-------
basis and the LM Food Chain results were for the
biota boxes used for model calibration.
A comparison of the Constant Conditions Scenario
results showed that MICHTOX predicted lower
concentrations in the epilimnion than LM2-Toxic
(Figure 6.3). This agreed with the model comparison
to 1994-1995 data in which MICHTOX
underpredicted epilimnion concentrations (Part 4,
Chapter 3). The models contained the same general
processes and used the same forcing functions,
therefore, differences in predicted concentrations
were primarily due to model parameterization and
resolution differences. For the southern Lake
Michigan model segments, hypolimnetic
concentrations reached comparable concentrations
at steady-state even though MICHTOX had higher
annual average concentrations initially. While both
models used the same initial conditions, MICHTOX
had higher sediment PCB resuspension rates which
resulted in the higher initial annual average
hypolimnetic water column concentrations shown on
the plot. For the central Lake Michigan hypolimnetic
segments, MICHTOX predicted lower concentrations
than LM2-Toxic.
Sediment concentration predictions were slightly
higher for MICHTOX than for the LM2-Toxic model
(Figure 6.4). The initial sediment concentrations on
a ng/L basis were significantly different between the
models. While both models used initial conditions
based upon measured 1994-1995 PCB
concentrations (ng PCB/g sediment), LM2-Toxicalso
used measured porosity from the 1994-1995 data
while MICHTOX used the same porosity and
sediment density used during the initial model
development.
The water concentration results from the Continued
Recovery - Fast Scenario were also compared
(Figure 6.5). For both sections of Lake Michigan,
0.5
0.4-
.9 0.3 -
ra
d
CO
o 0.2-
CO
O
Q.
CD
O
a.
0.1 -
0
1990
0.5-
0.4-
.° 0.3-
0.2-
0.1-
Southern Lake Michigan
Epilimnion
0.5
0.4-
.9 0.3-
0.2-
O 0.1
2000 2010 2020 2030 2040 2050 2060
Southern Lake Michigan
Hypolimnion
CO
o
o_
1990 2000 2010 2020 2030 2040 2050 2060
0.4-
° 0.3-
0.2-
0.1 •
Central Lake Michigan
Epilimnion
2000 2010 2020 2030 2040 2050 2060
Central Lake Michigan
Hypolimnion
1990 2000 2010 2020 2030 2040 2050 2060
Figure 6.3. Comparison of model output annual average total PCB water concentrations for the
Constant Conditions Scenario.
548
-------
25,000-
n 20,000-
0)
I 15, GOO-
'S
'c
§ 10,000-
o
o
CD
5.000-
Southern Lake Michigan
Sediments
MICHTOX segment 11
—=— LM2 segments 42-55
25,000-
' 20,000-
.9 15,000-
10.000-
CQ
£ 5,000-
0-1
Central Lake Michigan
Sediments
MICHTOX segment 12
-— LM2 segments 56-64
1990 2000 2010 2020 2030 2040 2050 2060
1990 2000 2010 2020 2030 2040 2050 2060
Figure 6.4. Comparison of model output annual average total PCB sediment concentrations for the
Constant Conditions Scenario.
0.5
0.4-
° 0.3-
0.2-
m
O
Q.
0}
O
Q.
0.1-
Southern Lake Michigan
Epilimnion
0.5
3" 0.4-
15>
c
.1 0.3 H
.
o
o
03
o
Q.
1990 2000 2010 2020 2030 2040 2050 2060
Southern Lake Michigan
Hypolimnion
0.2-
0.1-
CQ
O
9=
Central Lake Michigan
Epilimnion
0.4-
0.3-
0.2-
0.1 -
1990 2000 2010 2020 2030 2040 2050 2060
0-
1990
1990 2000 2010 2020 2030 2040 2050 2060
Central Lake Michigan
Hypolimnion
2000
—i 1—
2010 2020
—I 1 1
2030 2040 2050 2060
Figure 6.5. Comparison of model output annual average total PCB water concentrations for the
Continued Recovery - Fast Scenario.
549
-------
the model predictions converged over time to low
concentrations. In the Constant Conditions Scenario
results, differences in model parameterization
resulted in differences in long-term predicted
concentrations. For scenarios in which the external
forcing functions declined over time, the differences
in predicted concentrations were still present, but the
magnitude of the difference was small compared to
the overall decline in concentrations. Thus for long-
term predictions, the rate of decline was more
influential than differences between the model
parameters.
The bioaccumulation models were also compared for
the Constant Conditions Scenario and the Continued
Recovery - Fast Scenario. For the Constant
Conditions Scenario, MICHTOX predicted lower total
PCB concentrations than LM Food Chain in the 5-6
year-old lake trout at both locations (Figure 6.6). The
predicted concentration differences were larger for
the central Lake Michigan/Sturgeon Bay results than
the southern Lake Michigan/Saugatuck results.
Differences in lake trout concentrations predicted by
the models were a function of both the exposure
concentrations predicted by the water quality models
and factors affecting bioaccumulation in the food
chain models.
The results from the bioaccumulation models for the
Continued Recovery - Fast Scenario (Figure 6.7)
were similar to the results from the fate and transport
models. PCB concentrations declined over time to
low concentrations, with the difference in predicted
concentrations between the models becoming
smaller over time. For Saugatuck, the LM Food
Chain model predicted that the 0.075 ug/g lake trout
PCB concentration target would be achieved in 2033,
and the MICHTOX food chain submodel predicted it
would be achieved in 2025. For Sturgeon Bay, the
LM Food Chain predicted a much slower decline than
the MICHTOX food chain submodel, with the target
being reached in 2036 and 2018, respectively. The
difference in time required to achieve the target
concentration was primarily due to the delay in the
start of the concentration decline in the LM Food
Chain predictions.
References
Ambrose, R.B., T.A. Wool, J.P. Connolly, and R.W.
Schanz. 1988. WASP4, a Hydrodynamic and
Water Quality Model - Model Theory, User's
Manual and Programmer's Guide. U.S.
Environmental Protection Agency, Office of
Research and Development, Environmental
Research Laboratory, Athens, Georgia.
EPA/600/3-87/039, 297 pp.
3.0
2.5-
~ 2.0-
o
9: 1.5-1
1.0-
0.5-
• MICHTOX - Southern Lake Michigan
• LM Food Chain
Saugatuck biota box
3.0
2.5-
w 2'°"
CO
O
e= 1.5-j
1.0-
0.5-
0
1990 2000 2010 2020 2030 2040 2050 2060
— MICHTOX - Central Lake Michigan
""LM Food Chain
Sturgeon Bay biota box
1990 2000 2010 2020 2030 2040 2050 2060
Figure 6.6. Comparison of the bioaccumulation model annual average total PCB concentration results
for the Constant Conditions Scenario.
550
-------
MICHTOX - Southern Lake Michigan
LM Food Chain
Saugatuck biota box
MICHTOX - Central Lake Michigan
LM Food Chain
Sturgeon Bay biota box
1990 2000 2010 2020 2030 2040 2050 2060
1990 2000 2010 2020 2030 2040 2050 2060
Figure 6.7. Comparison of the bioaccumulation model annual average total PCB concentration results
for the Continued Recovery - Fast Scenario.
Endicott, D.D. 2005. 2002 Lake Michigan Mass
Balance Project: Modeling Total PCBs Using the
MICHTOX Model. In: R. Rossmann (Ed.),
MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan, Part 2. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse He, Michigan.
EPA/600/R-05/158, 140 pp.
Endicott, D.D., W.L. Richardson, and D.J. Kandt.
2005. 1992 MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan. In: R. Rossmann (Ed.),
MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan, Part 1. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
EPA/600/R-05/158, 140 pp.
551
-------
PART?
APPENDICES
Appendix 7.1 Lake Michigan Mass
Balance Project (LMMBP) PCB Peer
Review Report
Kenneth R. Rygwelski
U.S. Environmental Protection Agency
Office of Research and Development
National Health and Environmental Effects
Research Laboratory
Mid-Continent Ecology Division
Large Lakes and Rivers Forecasting Research
Branch
Large Lakes Research Station
9311 Groh Road
Grosse He, Michigan 48138
7.1.1 Executive Summary
The United States Environmental Protection Agency
(USEPA), National Health and Environmental Effects
Laboratory (NHEERL), Mid-Continent Ecology
Division at Grosse lie, Michigan in cooperation with
the USEPA Great Lakes National Program Office
(GLNPO), conducted a polychlorinated biphenyl
(PCB) transport and fate mass balance modeling
study of PCBs in Lake Michigan to determine
strategies for managing and remediating this toxic
chemical in the lake basin. Some specific programs
that this effort support include the Lake Michigan
Lake-wide Management Plan (LaMP) and the Great
Lakes Water Quality Agreement (GLWQA). Within
the ecosystem approach, the Lake Michigan Mass
Balance Project (LMMBP) models account for the
sources, sinks, transport, fate, and food chain
bioaccumulation of PCBs. The calibrated models
offer an opportunity for running various PCB load
reduction scenarios to get an insight on the effects to
the lake ecosystem. Model forecasting of PCB
concentrations in lake trout is one of the primary end-
points of the investigation as it relates to both
ecosystem and human health. In addition,
demonstration of a whole lake Total Maximum Daily
Load (TMDL) process to yield a desired target PCB
concentrations in lake trout has been achieved. A
significant factor that differentiates this study from
other PCB transport and fate modeling projects is
that PCBs were modeled as single PCB congeners to
predict total PCBs. Also, a high-resolution
hydrodynamic model was applied and a
eutrophication model was used to generate the
primary productivity solids in this system where
autochthonous solids production is significant and
plays an important role in describing PCB transport.
Mass balance estimates indicate that the lake system
is losing approximately 2,000 kg/year of PCBs. Also,
the bioaccumulation model predicts that the target
level for unrestricted consumption in lake trout (0.075
ppm for whole fish) was forecasted to be achieved for
five to six year-old lake trout between the years 2025
and 2035.
The main sampling activity for the project was
conducted in 1994 through 1995; however, a PCB
screening-level model called MICHTOX was
developed and running PCB simulations before the
LMMBP began. This model was developed to gain
an initial insight into the PCB transport and fate in
Lake Michigan including its biota. Later on, the
MICHTOX model was run again using the newer data
collected from the LMMBP. The more advanced
PCB and support models included a hydrodynamic
model called Princeton Ocean Model (POM), a
eutrophication model called LM3-Eutro, and a 41
552
-------
segment PCB model called LM2-Toxic. The output
from the LM2-Toxic was used to define the exposure
concentration for the bioaccumulation model called
LM-Food Chain. Development of a high resolution
PCB model (LM3-Toxic) is proceeding, and is
discussed in the Modelers' Comments section.
On July 27 - 28, 2004, peer reviewers representing
modelers in academia, research, and the USEPA,
convened at the Crowne Plaza Hotel in Romulus,
Michigan to review the LMMBP PCB models (Figure
1). Prior to this review, a June 2004 draft copy of
"Results of the Lake Michigan Mass Balance Project:
PCBs Modeling Report" prepared by the Large Lakes
and Rivers Forecasting Research Branch, Mid-
Continent Ecology Division, NHEERL, Office of
Research and Development (ORD), USEPA at
Grosse lie, Michigan was provided to each of the
peer reviewers. In general, the review panel agrees
that the model construct (spatial, temporal, process
resolution) and application is consistent with the
problem definition for which the model was
developed and for the available resources. In
addition to providing a comprehensive review of the
model, the panel also provided detailed suggestions
for future model improvements. Most of the panel's
comments were captured at the meeting and are
identified as "consensus" comments. James Martin
provided additional post-meeting comments (see
Section 7.1.5), and the modelers' responses to his
questions follow the responses to the consensus
comment section. In addressing the review
comments, we had to carefully consider where to
apply available resources and prioritize actions that
help to ensure model integrity. Hopefully our
responses reflect this balance. Responses identify
actions that have been taken, are on-going, or will be
conducted in the future. The USEPA wishes to thank
the panel for their willingness to participate in this
review and for their constructive comments.
7.1.2 LMMBP Peer Review Panel
Robert B. Ambrose, Jr., P.E.
Environmental Engineer
Ecosystems Research Division
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Office of Research and Development
960 College Station Rd.
Athens, Georgia 30605
Voice: 706-355-8229; Fax:706-355-8104
ambrose.robert@epa.gov
Joel E. Baker, Ph.D.
Professor
Chesapeake Biological Laboratory
P.O. Box 38
2108 Fowler Laboratory
University of Maryland
1 Williams Street
Solomons, Maryland 20688-0038
Voice: 410-326-7205
baker@cbl.umces.edu
Ken Drouillard, Ph.D.
Assistant Professor
Great Lakes Institute for Environmental Research
Biological Sciences Department
401 Sunset Avenue
Windsor, Ontario Canada N9B 3P4
Voice: 519-253-3000 Ext. 4744
kgd@uwindsor.ca
Barry Lesht, Ph.D.
Department of Energy, Acting Director of
Environmental Research Division
Argonne National Laboratory
9700 S. Cass Avenue
Argonne Illinois 60439
Voice: 630-252-4208 Fax: 630-252-2959
bmlesht@anl.gov
James L Martin, Ph.D., P.E.
Professor and Kelly Gene Cook, Sr. Chair in Civil
Engineering
Department of Civil Engineering
Mississippi State University
P.O. Box 9546
222 Walker Engineering Building
Mississippi State, Mississippi 39762-9546
Voice: (662) 325-7194; Fax: (662) 325-7189
jmartin@engr.msstate.edu
7.7.3 LMMBP PCB Charge to Peer
Reviewers
The members of the Peer Review Panel have been
assembled by the USEPA GLNPO because they are
experts in multimedia mass balance modeling and
have expertise in one or more of the multimedia
553
-------
Table 7.1.1. Agenda - Lake Michigan Mass Balance PCB Modeling Peer Review, Crowne Plaza Hotel,
8000 Merriman Road, Romulus, Michigan 48174 at Detroit Metropolitan Airport, Romulus, Michigan,
July 27 and 28, 2004
Time
Tuesday's Agenda
Speaker
8:20 am
8:40 am
8:50 am
9:00 am
9:1 5 am
9:30 am
9:45 am
1 0:00 am
10:15 am
10:40 am
11:1 Oam
11:20 am
11:50 am
12:00pm
1 :00 pm
1 :45 pm
2:00 pm
2:30 pm
2:45 pm
3:00 pm
3:30 pm
3:40 pm
4:10 pm
4:20 pm
4:50 pm
5:00 pm
5:30 pm
Time
8:00 am
8:30 am
8:40 am
9:1 Oam
9:20 am
10:00 am
10:15 am
12:00 pm
Welcome/Introductions and Project Goals/Objectives/Uses
Agenda Overviews/Previous Reviews
Charge to Peer Review Panel
PCB Background and History
PCB QA Report
PCB Data Summary/Representativeness
Questions and Discussion
Break
Modeling Introduction/Overview
Atmospheric Load Modeling
Questions and Discussions
Tributary Load Modeling
Questions and Discussion
Lunch
MICHTOX Level 1 Modeling
Questions and Discussion
Hydrodynamic Modeling and POM to WASP Linkage
Questions and Discussion
Break
Eutrophication Modeling - Autochthonous Carbon Production
Questions and Discussion
PCB Fate and Transport Modeling
Questions and Discussion
Food Chain Bioaccumulation Modeling
Questions and Discussion
Remaining Issues/Wednesday's Agenda
Adjourn for the Day
Wednesday's Agenda
Comparisons of Models
Questions and Discussion
Future Plans and Applications
Questions and Discussion
Summary of Peer Review Panel Recommendations
Break
Wrap-Up Session Final Discussion and Debriefing by
Reviewers
Adjourn
P. Horvatin
J. Keough
G. Warren
R. Kreis
L. Blume
R. Rossmann
R.. Kreis
D. Hornbuckle/J. DePinto
D. Hall
D. Endicott
D. Beletsky
J. Pauer
Xiaomi Zhang
Xin Zhang
R. Kreis
Speaker
K. Rygwelski
R. Kreis
R. Kreis
554
-------
aspects of this modeling approach. Panel Reviewers
are expected to provide an objective, unbiased
review of the Lake Michigan Mass Balance PCB
Modeling: science, best modeling practices, conduct,
and supporting components. Panel Review
comments should be verbally summarized at the end
of the review and then provided in written form,
cognizant of constraints in data availability, staff,
and financial resources associated with the project.
Written comments on the format and content of the
project documentation can be provided, if
appropriate.
7.1.3.1 Overall Multimedia Ecosystem Modeling
Approach
Does the suite of models applied, including
atmospheric and tributary load calculation
models/methodologies, hydrodynamics model,
autochthonous solids (eutrophication) model, water
column and sediment transport and fate models, and
food chain bioaccumulation model, represent an
integrated approach to ecosystem modeling? Are
these models, in combination, state-of-the-art? What
are the strengths and weaknesses of the overall
approach?
7.1.3.2 Overall Model Performance
Overall, how well does the model suite represent
physical, chemical and biological processes? How
consistent are the modeling concepts and
assumptions with current scientific knowledge? Are
the processes being depicted at the spatial and
temporal scales appropriate/adequate for the issues
being addressed and data availability? Overall, how
well are transport, exchange, and partitioning
processes for PCBs accounted for? Are the food
web, trophic structure, and processes which affect
bioaccumulation represented accurately? Overall,
how well is food chain bioaccumulation of PCBs in
Lake Michigan represented? Are model algorithms
used to describe processes appropriate (complexity
versus simplification)? Have the data been
adequately and fully utilized in the modeling? What
are the strengths, weaknesses, and uncertainties of
the overall modeling performance?
7.1.3.3 Suitability for Management
In terms of their predictive capability related to
transport, fate, and bioaccumulation of PCBs in lake
trout, is the suite of models and application sufficient
to evaluate and guide potential PCB load reduction
strategies for Lake Michigan? What are anticipated
modeling strengths and weaknesses for
management uses?
7.7.4 Modelers' Responses to Peer
Review Comments
1. How do you reconcile the difference in peak PCB
production versus peak loads for the hindcast
run?
Modelers' Response - The explanation for this
difference is not readily apparent. A similar
difference has been noted for Lake Ontario (Gobas
et a/., 1995). Peak production occurred in 1970.
Gobas et al. (1995) found the best overall agreement
between observed and predicted totaf PCB
concentrations in water, sediment, and biota
occurred when peak loading was assumed to occur
in 1961. One would not expect peak loading to
necessarily occur in the same year as peak
production. Much of the PCBs produced were used
in transformers and other sealed sources which
would not have an impact on the environment until
product failure occurred. It is believed that most of
the significant loading of PCBs to Lake Michigan
came from PCBs that were used for other purposes.
The use of these PCBs in the basin does not appear
to coincide with production. Within the basin, the first
noted use of PCBs was at Waukegan, Illinois in 1948
when Outboard Marine Corporation purchased
hydraulic fluid with PCBs. From the mid-1950s to
mid-1960s, PCBs from deinking were loaded to the
Kalamazoo River. In the 1950s, PCBs were used in
the Green Bay area for production of PCB-coated
carbonless copy papers. These discharges to the
Fox River peaked in 1969-1970. The use of PCBs
for these papers was phased-out in 1971-1972. In
the 1960s, industrial PCBs were loaded to
Sheboygan Harbor. Thus it appears that PCB use in
the basin began in 1948 and ended in 1972. The
loadings over time from these uses of PCBs is not
currently documented; however, it appears that PCB
loadings to the Lake Michigan basin do not coincide
555
-------
with production or sales figures. This discussion will
be included in Part 1, Chapter 7 of the report.
2. POM: Extend the hydrodynamic record from two
years to ten years.
Modelers' Response - Currently, funding is not
available to extend the POM modeling from two
years to ten years. When preparing to conduct
forecasts, we were concerned about how
representative POM (1994 and 1995) results would
be when repeated along with water temperature,
velocities, and dispersion coefficients for LM2-Toxic
(PCB) and LM3-Eutro (nutrients and carbon) model
runs extending beyond the two-year LMMBP period.
The POM model used lake conditions and forcing
functions present March 31,1982 through November
20, 1983 for calibration purposes. Comparing lake
and atmospheric conditions such as wave height, air
temperature, lake levels, tributary flows, and
precipitation for the 1982-1983 and the LMMBP
period 1994-1995 with the historical record, we found
that neither of the two-year periods were at any
extreme from means. Based on this review, we
believe that the two years of hydrodynamic modeling
fairly represent average lake conditions. More
discussion on the representativeness of the 1994-
1995 period can be found in Part 1-Introduction and
Chapter 4 of the report.
3. LM3-Eutro: Why does the model not predict
dissolved silicon beyond 0.7? Was the code
checked for possible errors? Identify Green Bay
stations on the model versus observed plots.
Modelers' Response - This question is related to
Figure 2.5.3 in the June 2004 draft copy of "Results
of the Lake Michigan Mass Balance Project: PCBs
Modeling Report." Please note that the axes in this
figure were incorrectly labeled. The abscissa axis
should be labeled "model results", and the ordinate
axis should be labeled "field data." The dissolved
silica output from the model was examined carefully.
Although it appears that the model does not predict
values higher than 0.7 mg/L (Figure 2.5.3), closer
inspection of the model output reveals that the
majority of the predicted values are relatively evenly
distributed between 0.4 and 0.76 mg/L, with a few
values as high as 0.78 mg/L. A limitation of the LM3-
Eutro model was the absence of a fully-developed
sediment submodel that reflected seasonal
variations. User-defined soluble reactive
phosphorus, ammonia, dissolved silica, and
dissolved organic carbon sediment fluxes were used
to provide an estimate of the sediment feedback.
However, these fluxes were constant values in space
and time and were selected to provide a reasonable
estimate of annual averages. Due to this limitation,
the model underestimated the silica build-up at the
bottom of the lake during the late summer months
caused by the slow decay of the biogenic silica,
which settled to the bottom during the spring and
early summer diatom blooms, and its potential
resuspension. It is believed that this is the major
reason why model output values are less that 0.8
mg/L whereas several field values are well above 1
mg/L. There is little difference between observed
silica concentrations in Green Bay and the open lake
with large seasonal variations at both locations. In
Lake Michigan, the observed silica range is between
2.1 and 0.04 mg/L, while in Green Bay it ranges
between 1.58 and 0.13 mg/L.
4. LM2-Toxic: Run the model with a conservative
tracer and check that mass balances (set initial
conditions and boundary concentration = 1).
Modelers' Response - This test was completed very
successfully. By setting initial and boundary
concentrations of an assumed conservative tracer
equal to 1 mg/L in both the water column and
sediment segments with no external load, no gas
exchange, and no partitioning process, the model
was run for a short-term simulation (two years) and
a long-term simulation (60 years). The results from
the model runs show no change for the two-year
model run in all media. For the long-term run, an
extremely small change (0.001 %) was found in water
column segments with roughly a 0.5% change in
sediment segments.
5. LM2-Toxic: Consider adding subsurface benthic
layers below the surficial layer. It is likely that
higher PCB concentrations reside in the deeper
layers. Look at historical data.
Modelers' Response - There are subsurface
sediment layers (called ghost layers) defined in the
current LM2-Toxic segmentation. A quasi-
Lagrangian framework is used to allow a moving
sediment-water interface. There is no mass
exchange between the mixed surficial sediment layer
556
-------
and subsurface sediment layer, and between two
adjacent subsurface layers through mixing or
diffusion processes. The cores that demonstrate the
high PCB concentrations in deeper layers are found
in the depositional zones where the potential for
resuspension is minimal. Furthermore, mass transfer
via diffusion between deeper layers is likely to be
minimal for hydrophobic PCBs (see response to
comment 14). A detailed description of the semi-
Lagrangian sediment bed option is detailed in Part 4,
Chapter 3, Section 4.3.4.2.3 of the report and IPX
2.74 documentation (Velleux et at., 2000).
6. LM2-Toxic: Recheck assumptions on scenarios
with attenuation rates for tributary loads and
wet/dry atmospheric deposition. Check Mites'
data against assumed decline in vapor
concentration.
Modelers' Response - The half-life of the PCB
decline in tributary and vapor phase loadings were
assumed to be 12.5 years and six years,
respectively, in our model runs for "natural
attenuation." These rates are consistent with the
PCB tributary and atmospheric loading rates of
decline calculated and used by other researchers
(Velleux and Endicott, 1994; Endicott, 2002; Marti
and Armstrong, 1990; Hillery et a/., 1997; Schneider
etal., 2001). Further examination of additional data
has not revealed any change to these assumptions.
See Part 3, Chapter 3 of the report and Endicott
(2005) for documentation of materials used for
hindcasts and forecasts for MICHTOX and LM2-
Toxic. Eventually, Attachment 4 will become a stand-
alone ORD publication. See Part 4, Chapter 6,
Paragraph 4.6.3 for documentation of sources of
information used for forecasts for the LM2-Toxic. A
detailed description of the uncertainties that would
have an impact on hindcast and forecast choices will
be detailed in a revision of the report in Part 1,
Chapter 7.
7. LM2-Toxic: Compare model projections to water
data post-1998 (southern Lake Michigan).
Modelers' Response - Post-1998 data for southern
Lake Michigan have been located. These data will
be compared with the model's long-term projections
as part of the model verification. The results of the
comparison will be detailed in a revised edition of the
report. Additional verification of the model will occur
after data collected in 2005 are available.
8. LM2-Toxic: The Panel recommended that a
Monte Carlo uncertainty analysis be performed
using a steady-state version of the model.
Modelers' Response - This is certainly a valid
suggestion. However, given the complexity of the
model and the number of solids (three solids) and
PCB congeners (54 congeners) simulated in the
model, it could be very costly and time-consuming to
do the recommended Monte Carlo uncertainty
analysis on even a few selected critical parameters
used in a steady-state version of the model. In
addition to the uncertainties associated with the
parameters defined by chemical and biochemical
processes conceptualized in the model, water
transport, solid cycling rates, numerical algorithms
used in the model, and data input into the model are
all subject to a certain error, and this error
propagates in the model results. The uncertainties
associated with these errors could be much greater
than the ones only related to the chemical-specific
parameters. See Part 4, Chapter 5, Section 4.5.4 for
details on the tasks conducted to reduce the
uncertainties caused by water transport and solid
cycling rates. If computing resources and manpower
become available, this issue will be addressed in the
future.
9. LM2-Toxic: In regards to solids dynamics
(radioisotope calibration), the Panel requested
that the Modelers' examine the decline rate and
add more recent data.
Modelers' Response - We will examine the decline
rate and add more recent data in the future as
available.
10. Peer Review Panel: LM2-Toxic/Eutro: How
sensitive is the PCB model to primary
productivity changes versus sediment net
resuspension changes?
Modelers' Response - The suggested sensitivity
analysis has been thoroughly investigated. For a
50% increase or decrease in primary production
corresponding to the primary production generated
from LM3-Eutro for the 1994-1995 period, the LM2-
Toxic model was tested for both a short-term (the
557
-------
two-year calibration) period and a long-term (62-
year) period. The results from the sensitivity
analyses were compared to the results from the LM2-
Toxic model base runs (i.e. 1994-1995 calibration run
and long-term Constant Condition Scenario). See
Part 4, Chapters 5 and 6 for detailed descriptions of
both of these base runs. The general results from
the tests are: 1) The solids concentrations (DOC -
Dissolved Organic Carbon, BIG - Biotic Carbon, and
PDC - Particulate Detrital Carbon) in the water
column have a substantial deviation from the base
run concentrations for both the short-term calibration
and the long-term scenario simulations; and 2) the
total (particulate plus dissolved) PCB concentrations
in the water column has a noticeable difference from
the base run concentrations for the short-term
calibration simulation, but it has very little difference
for the base run concentrations for the long-term
simulation. The results from the sensitivity analyses
suggest that, under the 1994-1995 PCB
loading/boundary conditions/other forcing functions,
the influence of primary production on the PCB
concentrations in the water column is very small,
especially for long-term forecast scenarios. The
details on the procedures used to conduct the test
and the associated results will be presented in the
revised edition of the report.
11. LM2-Toxic/POM: The model did not consider
ice cover in various processes (volatilization,
resuspension, etc). Perhaps test the affect with
a sensitivity analysis.
Modelers' Response -The POM model was applied
to Lake Michigan by the National Oceanic and
Atmospheric Administration (NOAA)-Great Lakes
Environmental Research Laboratory (GLERL). The
current version does not include ice cover algorithms.
However, in the absence of an ice model, both POM
and LM2 were run with the water temperature steady
at 2°C from the period January 1, 1994 through
March 31, 1994. Lake Michigan ice cover for 1982
and 1994 were greater than the mean and median
whereas 1983 and 1995 were less than the mean
and median. None of the four years (1982-1983
hydrodynamic model calibration years and 1994-
1995 LMMBP years) represented an extreme of
mean daily ice cover. There is an in-depth
discussion on historical ice cover data for Lake
Michigan in Part 1, Chapter 4 of the report.
Both NOAA and Large Lakes Research Station
(LLRS) staff agree that ice cover algorithms in POM
would be worthwhile additions to the model. During
most winters, Lake Michigan ice cover occurs most
often in the nearshore areas only. LM2-Toxic could
utilize ice cover predictions from POM by indicating
the fraction of the surface segment area that is
covered during certain times. However, the coarse
grid structure of LM2-Toxic could not be used to
predict the impact of ice cover in specific small
regions of the lake, such as nearshore zones. At this
time, we do not have the in-house expertise to
develop a revised POM that addresses ice cover;
however, when a revised POM is made available
from GLERL that incorporates these algorithms, we
could incorporate this version into our Level 3 models
where segmentation resolution is fine enough to
better deal with year-to-year and within-year ice
cover variations. GLERL is planning to incorporate
ice cover algorithms in POM for application to Lake
Erie.
The effect of ice cover on PCB mass fluxes across
the air-water interface through gas absorption and
gross volatilization is likely to be small because our
calculations predict that these PCB mass fluxes
decrease substantially with a decrease in
temperature. However, it is recognized that ice cover
could affect both particulate settling rates and
sediment resuspension fluxes of PCBs in certain time
periods in a year in the nearshore regions. It is our
opinion that ice cover most likely will not have a
substantial impact on the long-term results from the
LM2-Toxic model, because ice cover does not affect
the total inventory of PCBs in the lake system.
However, for short-term predictions, ice cover would
be expected to impact the model predictions.
12. LM2-Toxic/Food Chain: Investigate congener
patterns in air, water, fish, and sediment. How
do these compare?
Modelers' Response - The PCB patterns of multiple
media will be compared to determine similarities and
differences within and among media. This technique
is commonly referred to as PCB fingerprinting or PCB
signature recognition and has had mixed success in
the past. This recommendation has minor
implications to the modeling; however, it is a data
analysis tool and has merit for data presentation and
interpretation purposes. The relative percent of total
558
-------
RGBs represented by each congener will be
computed and then expressed as a cumulative
frequency plot for comparative purposes. These will
represent data for an entire study period, will by
tested with both mean and median values, and will
be a composite expression of seasonal and spatial
data. In addition, selected evaluation of pattern
recognition using the LMMBP data set can be found
in Kuehl (2002) and McCarty etal. (2004).
Fingerprints will be calculated for sediment, water
column (dissolved and particulate), vapor phase, wet
and dry atmospheric deposition, and age 5-6 year-old
lake trout signatures from the Saugatuck biota site.
Atmospheric signatures will be based on a subset of
all congeners because vapor phase data were
computed by Keri Hornbuckle for the study, and over-
lake concentrations were only calculated for the
congeners that are being modeled at Grosse He. In
addition, PCB patterns associated with water
discharging from the Kalamazoo River near the
Saugatuck biota site and other selected tributaries
will be compared/contrasted to the lake water. These
results will be presented in Part 1, Chapter 6 of the
report.
13. LM2-Toxic: Consider the missing 120 kg/year
total PCB contribution from Milwaukee (sum
vapor/wet/dry); how sensitive is the PCB model
to atmospheric and tributary loads?
Modelers' Response - The issue, documented in
Wethington and Hornbuckle, 2005, of an additional
PCB load from the Milwaukee area through vapor
exchange, wet deposition, and dry deposition to Lake
Michigan was not included in our model. The
additional PCB load from the atmosphere was
estimated to be at least 120 kg per year. The
sensitivity analysis for the Milwaukee load was
performed by adding a 120 kg/year PCB load into
segment 1 in our model. The results from the
sensitivity analysis were then compared to results
from the LM2-Toxic model long-term (62 years) base
run (Constant Conditions Scenario). The steady-
state concentrations from this simulation show an
increase of less than 5% in the steady-state
concentration compared to the original long-term
base run. The details on the Milwaukee loading
sensitivity analysis and the impact of the external
PCB load changes and vapor phase concentrations
to the projected level of PCB concentrations in Lake
Michigan will be discussed in the revised report. It
should be noted that additional data will always
continue to become available, and this is such a
case.
Another potential missing load to the lake is that load
associated with very large particles greater than 10
um. Although experts disagree on the magnitude of
the PCB load to the lake via large particles, various
scientists indicate that PCB dry deposition associated
with large particles could be a significant PCB source
to the lake (Miller et a/., 2001; Franz et a/., 1998;
Holsen, 1991). Currently, it is not possible to make
reliable estimates of these fluxes to the lake. The
uncertainty in these flux estimates is associated with
the uncertainty of how far these large particles travel
from their sources.
Simulations were run to gain insight into how the
model would respond to increasing the total PCB
load (load from all tributaries + atmospheric load to
the entire lake) by 50% and 100%. The results from
these sensitivity analyses were then compared with
those from the LM2-Toxic model long-term (62 years)
base run (Constant Conditions Scenario). The
steady-state concentrations from this simulation
show an increase of less than 10% and 25%,
respectively, to the steady-state concentration from
the long-term base run.
14. LM2-Toxic: Investigate PCB diffusion from
deeper sediment layers, relative to sediment
resuspension.
Modelers' Response - Because of very large PCB
partition coefficients, most of the PCB mass in the
sediment is associated with the particulate phase.
Therefore, relative to sediment resuspension, the
PCB mass moved by PCB diffusion between deeper
sediment layers through pore water is trivial. Adding
this process into the current semi-Lagrangian
scheme for sediment transport in the LM2-Toxic
model would require considerable effort and time to
modify the code, calibrate the model, and analyze the
output.
15. LM2-Toxic/LM Food Chain: Conduct a
hindcast for PCBs which will be further used in
the LM2 Food Chain hindcast; possibly select
five congeners.
559
-------
Modelers' Response - We agree with the review
panel on the importance of this task. Although it
requires significant resources, time and effort, this is
an effort that is certainly worth doing. This task is
currently underway and all of the selected PCB
congeners (54) will be simulated in the model
hindcast. The LM Food Chain model will be run
using the output of the LM2-Toxic hindcast that was
run to describe the exposure history. The approach,
inputs, model outputs, and interpretation of the
results will be presented in the revised report.
16. LM2-Toxic/POM: Examine the effects of
changing the horizontal grid structure to
evaluate translating POM output to the LM2
grid.
Modelers' Response - This is a valuable and
interesting suggestion. However, it would require
considerable effort and expertise to accomplish the
task. It would require extramural personnel such as
David Schwab of NOAA-GLERL to adapt POM
output to the new grid structure. Considering the
resources and time that would be involved in this
task, it cannot be done in the foreseeable future.
17. LM-Food Chain: Calibrate over declining
exposure concentrations rather than constant
exposure history. To facilitate this effort,
consider hindcasting five PCB congeners.
Modelers' Response - Due to the lack of credible
PCB congener-specific exposure history data, the
measured data for PCBs in water and sediment
(1994-1995) was assumed to be representative of
life-long average exposure concentrations for the
food web and was, therefore, used in model
calibration simulations. Model calibration over
declining exposure concentrations will be attempted
once the temporal profiles of congener-based PCB
concentrations in water and sediment become
available from the LM2-Toxic hindcast.
Theoretically, calibrations over declining exposure
concentrations should yield better results than that
conducted over a constant exposure history because
PCB loads to the Great Lakes have been and will
likely continue to decline toward a steady-state.
However, it is difficult to accurately determine the
rates of decline for exposure concentrations in the
various media. The lake trout, as well as coho
salmon food webs in Lake Michigan, are exposed to
PCBs associated with both the water and sediment.
Therefore, model calibrations for their food webs
require information on temporal variations of PCB
concentrations in both the water and sediment over
the exposure history. However, the temporal trends
of PCB concentrations decline are usually reported
for total PCBs only. Congener-specific PCB data are
rarely available. For total PCBs, the quality of the
estimated exposure decline rates is usually
questionable due to the often considerable variability
and uncertainty of the measured total PCB data in
the water and sediment. Therefore, with limited data
availability and poor data quality, a reconstructed
declining exposure history is not necessarily a better
representation of the actual exposure condition than
the constant exposure assumption. If one can
assume that the PCB concentrations in Lake
Michigan system is currently declining at a very small
rate, then model calibrations using current congener-
specific PCB data as average life-long exposure may
be a more desirable alternative than model
calibrations over declining exposure concentrations
for total PCBs or a limited number of congeners.
18. LM Food Chain: Concern on using specific
dynamic action (SDA) versus activity cost
(respiration). Recommend using activity cost
for calibration. This question will be re-
formulated in the written review.
Modelers' Response - In the LM Food Chain model,
activity cost for each species is estimated as a
function of temperature based on bioenergetic
equations and is not "refined" during calibrations. We
chose to adjust only SDA with the hope to minimize
the risk of turning calibration into a curve-fitting
exercise.
19. LM Food Chain: Explain why 5.5 year-old trout
data are so variable at Saugatuck.
Modelers' Response - It is not uncommon for fish to
have variable PCB levels among individuals of the
same age class and among age classes. The
variability of PCB levels among individual fish can
have a direct impact on the uncertainty interval
associated with the measured PCB data in
composites. The variability of an individual fish's
PCB concentrations may be attributed to, among
other things, the differences in body size, health
560
-------
condition, feeding skill, dietary preference among
individual fish, exposure variances due to their spatial
feeding range, and analytical chemistry variability.
These individual differences are likely to have direct
impact on the growth rate and the amount of dietary
PCB intake of individual fish and subsequently on
fish's PCB bioaccumulation.
The trend for the USEPA historical Saugatuck lake
trout PCB concentration data is clearly downward.
Similarly, the within-year variability represented by
95% confidence intervals for these observations also
demonstrates a decreasing trend in time. Data prior
to 1981 clearly have much more within-year
uncertainty than data collected after that date, and
the concentration means for these earlier data show
greater year-to-year differences than do later
composites. A cursory examination of the historical
PCB lake trout composite concentration data
compared against available mean fish length, fish
weight, % lipids, and % males in the composites
revealed no distinct relationships for the years
examined.
Although a cursory examination of the historical PCB
lake trout composite concentration data collected at
Saugatuck compared against mean fish weight and
length showed no distinct relationship, some of the
within-year variability of composites could be further
attributed to the fact that the monitoring samples
were not collected for a particular age class. Rather,
the lake trout samples were collected and classified
as600-700 mm size class. For many of the historical
600-700 mm lake trout samples, their age classes
are uncertain. Based on accurate age classification
for the 1994-1995 lake trout samples, the 600-700
mm size class can be roughly associated with 5 and
5 year age classes. However, this size-age
correlation may not necessarily be applicable for lake
trout samples collected in other years. In other
words, the monitoring data for adult lake trout at
aaugatuck over the years represent PCB levels in a
range of age classes of lake trout. In our report,
Jese monitoring data were labeled and plotted as
LBs in 5.5 year-old lake trout merely for
onvenience in comparing with age-specific modeled
£B data for lake trout. To demonstrate the range of
, predicted concentrations in lake trout, model
e 4> 5> 6-and 7 year-old lake trout will be
with the Saugatuck historical lake trout data.
This graphic will appear in the revised draft of the
report.
Dietary preference is likely a very important aspect in
evaluating long-term trend data. Food web changes
have and are occurring in Lake Michigan based upon
past and present disparate reports on the topic.
However, the 1994-1995 diet study results suggest a
general consistency with known lake trout diet
preferences in the past. Although these are typically
dominated by alewife, bloater, sculpin, and smelt,
there may be a greater trend toward bottom-dwelling
bloater and sculpin than during the general evidence
in the past two decades.
The year-to-year differences in mean concentrations
and the within-year variability observed in lake trout
could also possibly be related to variable exposure
resulting from meteorological and physical factors.
These factors have the potential to have direct and
indirect impacts on the food web and exposure
gradients within the feeding range of Saugatuck lake
trout. A primary factor is PCB loading events
associated with high flow from the Kalamazoo River
that discharges at Saugatuck. The Kalamazoo River
has a history of PCB contamination (see table that
follows in response to 1.20). Also, periodic low lake
level events have the potential to reduce PCB
exposure to the lake trout food chain in certain zones
which could be reflected in periodic low-level PCB
body burden results. In a period from the mid-1960s
to the late-1980s, the lake levels were at near record
lows and near record highs, respectively.
Much of the analytical chemistry data on fish prior to
1983 was performed using packed column gas
chromatography (GC). Dichlorodiphenyl-
dichloroethylene (DDE) often co-extracted with the
PCBs and was very difficult to analytically separate
from PCBs on the packed-column GC (Michael
Mullin, personal communication). If the concentration
of DDE was significant, and separation was not
complete, then a positive PCB bias results in the
measurement. Some measurements were
performed where sample extracts were analyzed
using joint GC and mass spectrophotometry. This
combined analytical system improved the ability to
separate the DDE from the PCBs. Most of the
analytical work performed post-1983 was done using
capillary GC which significantly improved separation
of DDE from PCBs.
561
-------
20. LM-Food Chain: Develop a time
line/chronology of regulatory and remedial
actions, relative to the fish monitoring
record/trend at Saugatuck.
Modelers' Response - This has been done and will
be found in the revised edition of the report in Part 1,
Chapter 5. Where appropriate, dates have been
added to figures in that chapter. The table used to
summarize the time-line is shown in Table 7.1.2.
21. LM-Food Chain: Provide an estimate of model
error on the fish long-term monitoring trend.
Modelers' Response — There are many sources that
contribute to model errors. They include conceptual
errors and/or omissions, errors in parameterization,
uncharacterized system variability, and errors in data
used for calibrations. In addition, the quality of model
prediction of the long-term fish monitoring trend
Table 7.1.2. Significant Dates in the History of PCBs in the Lake Michigan Basin
Date
Event
1865
1881
1914
1927
1935
1948-1971
1954
Mid-1950s to Mid-1960s
1959-1972
1960s
1969-1970
1970
1971-1972
1973
1975
1977
1984
1985
1989-1990
1991
1991-1992
1998
1997-1999
1997-1998
1998-1999
1994-2000
20002
First PCB-like chemical discovered
First PCBs synthesized
Measurable amounts of PCBs found in bird feathers
PCBs first manufactured at Anniston, Alabama
PCBs manufactured at Anniston, Alabama and Sauget, Illinois
Outboard Marine Corporation at Waukegan, Illinois purchase eight million gallons of
hydraulic fluid with PCBs
Appleton Paper Company began using PCBs as PCB-coated carbonless copy paper
with discharges to the Fox River
PCBs loaded to Kalamazoo River from deinking
Outboard Marine Corporation at Waukegan, Illinois used hydraulic fluid with PCBs for
die-casting
PCBs used by Tecumseh Products Company loaded Sheboygan River
Paper company discharges of PCBs to Fox River peaked
PCB production peaked at 85 million pounds and huge contamination noted at Sauget,
Illinois plant
Appleton Paper Company and NCR Corporation phased-out PCB use. Recycling of
carbonless paper had occurred for several decades
U.S. Food and Drug Administration (USFDA) establish 5 ppm PCB tolerance level in fish
124,000 cans of salmon from Lake Michigan seized because of PCBs
PCB production ends
USFDA lowered PCB tolerance level in fish to 2 ppm
Commercial fishing for carp and other valuable species outlawed on Green Bay
Sheboygan Harbor PCB Remediation
U.S. Department of Health and Human Services label PCBs as possible carcinogen
Waukegan Harbor PCB remediation (1 million pounds PCBs) completed for this action
in 1992. Additional work is planned.
The eight Great Lakes States agreed on a "Great Lakes Protocol for Fish Consumption
Advisories" that lowered the regional standard from the USFDA commercial
standard of 2 ppm down to 0.05 ppm
Kalamazoo River sediment PCB remediation on Bryant Mill Pond (20,000 pounds of
PCBs). Additional work is planned
Milwaukee River PCB remediation
Upper Fox River deposit N (17,000 cubic yards) and sediment management units 56
and 57 dredging partially completed. Additional work is planned on the Fox River
system
Manistique Harbor PCB remediation (141,000 cubic yards)
Possibly begin Grand Calumet River PCB remediation
562
-------
depends on the availability of a realistic projection of
future exposure concentrations in water and
sediment for the food web. Not all errors from these
sources can be quantified. In the final report, we will
provide an estimate of model error associated with
model parameterization and calibration with an
emphasis on error associated with a potential shift in
food web structures.
22. LM Food Chain: Conduct a prey sensitivity
analysis for lake trout.
Modelers' Response - We will be testing the
sensitivity of the model to changes in the food web
structure. In addition to a prey sensitivity analysis for
lake trout at Saugatuck, model sensitivity to fish
growth rate, lipid content, and temperature (among
others) will also provided in the revised edition of the
report.
23. Enhance data presentations of the project data
to provide regional and open lake/nearshore
differences and gradients for multiple media.
Modelers' Response - We agree that data
presentation is an important aspect of this project to
aid in the understanding of modeling results. PCB
data used for the modeling will appear in the revised
edition of the report in Part 1, Chapter 5, "PCBs in
the Lake Michigan Ecosystem."
24. Provide CDs of presentations to the Peer
Review Panel.
Modelers' Response - All presentations given at the
peer review were provided to the panel members in
electronic form after the review.
25. Provide CD copies of draft modeling report,
appendices, and attachments (those available
electronically) to the Peer Review Panel.
Modelers' Response — The Draft Modeling Report,
Appendices, and Attachments were provided to the
panel in bound hard-copy in late June 2004 for
review purposes. CD/electronic copies of the report
and associated materials were provided to the peer
review panel after the review.
7.1.5 Modelers' Responses to Specific
Comments Made by Peer Review Panel
Member - James Martin
Note: Some of Dr. Martin's comments were identical
to those listed in the consensus comments above
and were, therefore, not repeated in this section.
1. MICHTOX: Continue to maintain the Level 1
model, particularly for comparison with Level 2
predictions.
Modelers' Response - While we appreciate the
reviewer's interest, it would be difficult to continue to
maintain the MICHTOX PCB model for purposes of
comparing future Level 2 predictions because of
current resource limitations. Furthermore, it is rather
difficult to make a direct comparison of MICHTOX to
Level 2 because the construct of the two models is
so different. We plan, however, to continue with the
development of a Level 3 PCB transport/fate/-
bioaccumulation model which would offer much more
spatial resolution and would incorporate SEDZL
sediment resuspension velocities along with the
QUICKEST-ULTIMATE sediment algorithms into the
framework. The Level 3 model should; therefore, be
very useful for comparison to the Level 2 model
predictions.
2. MICHTOX, LM2, and LM3: Explore incorporating
specific algorithms, such as the steady-state
algorithm (as exists in MICHTOX), with the Level
2 and potentially Level 3 models.
Modelers' Responses - A steady-state version would
be helpful if Monte Carlo type simulations were
performed to help understand model uncertainty. For
the LM2 and LM3 PCB congener-level models, a
Monte Carlo type uncertainty analysis presents many
challenges as described in our response to comment
8 above. However, for our other two LMMBP toxic
chemicals of interest (mercury and frans-nonachlor),
a Monte Carlo type simulation would be more
feasible due to the fewer number of state variables.
563
-------
3. MICHTOX/LM2: Both models predicted
remarkably similar changes in total PCB
concentrations over time in the long-term
projections. However, there were differences
between the two models such as in the rates of
settling/resuspension used and in the
characterization of the sediment bed. As a result,
the two models predicted similar results for
somewhat dissimilar reasons. It would be of
interest to further investigate factors leading to
the similarity in predictions, which may provide
some additional insights as to the factors
controlling PCBs in Lake Michigan.
Modelers' Responses - In Ken Rygwelski's
PowerPoint presentation at the peer review titled, A
Comparison of Lake Michigan Mass Balance Project
(LMMBP) Polychlorinated Biphenyl Models:
MICHTOX versus LM2-Toxic and LM Food Chain, a
graphic was presented that displayed the whole lake
total PCB concentration projections for the two
models. It is noteworthy that at the beginning of the
runs depicted, January 1,1994, MICHTOX starts out
higherthan LM2-Toxic. The reason is that MICHTOX
concentrations at that time represent the
concentrations predicted from the MICHTOX
hindcast whereas the concentrations form LM2
represent observed lake concentrations at that time.
Also, note that in the same presentation MICHTOX
was losing approximately 2,958 kg/year and LM2-
Toxic was losing 2,043 kg/year. Although MICHTOX
was losing more PCBs per year than LM2-Toxic, it
started out with a higher lake inventory of PCBs in
the water, which can explain, in part, why the two
models predict similar concentrations. The construct
of these two models is rather different in a number of
ways, and this makes comparisons difficult. Some of
these differences, however, were most likely
overcome through the calibration process of the two
models. A discussion on this topic of model
comparability between MICHTOX and LM2-Toxic will
appear in the revised report.
The two food chain models were also very close to
predicting when the 5.5 year-old lake trout at the
Saugatuck biota zone would reach the target
consumption criteria of 0.075 ppm PCBs in whole
fish. MICHTOX predicted year 2025 and LM Food
Chain predicted the year 2026. A major difference
between these two model constructs is that
MICHTOX is composed of four members in the food
web whereas LM Food Chain has 10 members.
Also, MICHTOX is based on two PCB homologs and
LM Food Chain is PCB congener-based. It is likely
that much of the similarities in the predictions of
these two models is due to calibration and the use of
the same rate of decline for PCB loads for natural
attenuation.
4. POM/LM2/LM3: Continue development of the
linked POM and Levels 2/3 models.
Modelers' Response - Currently, neither in-house
expertise nor funding exists to further the
development of POM for Lake Michigan. We do
recognize; however, that simulating ice cover and
incorporating finer spacial resolution for some
nearshore "hot spot" areas could be described better
with a POM upgrade.
In terms of upgrading Level 3 models, we are
currently working on upgrading the coupled LM3-
Eutro and LM3-Toxic (PCB) model. This near-term
goal includes resuspension velocities from SEDZL
into the coupled model. The QUICKEST-ULTIMATE
algorithm will be implemented in the 10 sediment
layers. The model has two particle classes: fine-
grained inorganic fraction and fine-grained organic
fraction. Refractory organic carbon, total
phosphorus, total nitrogen, and total silica will be
associated with the particulate resuspension flux as
well as the PCB modeled congeners. With this new
construct, a LM3 hindcast from 1960 to 1995 will be
run for both the Eutro and Toxic components of the
coupled model. A long-term goal includes adding
sediment diagenesis and dissolved oxygen
algorithms to LM3 Eutro.
LM3-Eco is an enhanced version of LM3-Eutro and
will eventually include Bythothrephes, Mysis,
carnivorous zooplankton, herbivorous zooplankton,
diatoms, and green algae (Phase 1), and additional
state variables including Dreissena, Diporeia,
nitrogen-fixing blue-green algae, and non-nitrogen-
fixing blue-green algae (Phase 2). At this time,
Phase 1 of LM3-Eco is in the model calibration stage.
564
-------
5. Provide documentation of the POM application
and testing, particularly with regard to an
assessment of the applicability of the model to
the transport of PCBs and other water quality
constituents. Perhaps include it as an appendix
to the modeling report.
Modelers' Response - See Schwab and Beletsky
(1998) and Beletsky and Schwab (2001). These
documents are available at http://www.gler!.
noaa.gov/pubs/techrept.html and http://www.
glerl.noaa.gov/pubs/fulltext/2001/20010008.pdf,
respectively. LLRS has electronic and hardcopy
forms of these documents in our library.
6. Investigate potential linkages issues between
POM and with SEDZL.
Modelers' Response - SEDZL has its own
hydrodynamic model and is not at all connected with
POM; therefore, all of the hydrodynamic forcing
functions are input into the SEDZL model. The
Donelan parametric wind wave model applied to Lake
Michigan by David Schwab of NOAA-GLERL is a
stand-alone model which is run before SEDZL is run,
and the output of the wave model becomes input for
the SEDZL model.
7. Investigate assumptions/limitations of using a
sigma grid, particularly in resolving both
nearshore and open lake issues. One potential
limitation to the POM model construct (relative to
this application) is related to the coordinate
system used in the vertical dimension (a sigma
grid). A sigma grid requires a constant number of
vertical layers throughout the model domain
(beneath each of the 5 km horizontal grid cells
(the number of vertical layers was variously cited
as from 15 to 20 in the modeling report, which
should be corrected). This use of the sigma grid
may impact the ability of the model to resolve
vertical gradients, particularly in deeper sections
of the lake while still sufficiently capturing
nearshore circulation patterns. In addition, sigma
grids may produce artificial horizontal transport
patterns. While there are numerical schemes for
compensating for this, I am not aware that they
have been implemented in POM or that any
sigma transport tests have been conducted for an
application such as Lake Michigan.
Modelers' Response - The actual number of sigma
layers throughout the POM construct is 19 water
layers. The citations mentioning 15 or 20 layers will
be corrected in the revised edition of the report.
The potential problem of using the sigma grid
structure for POM is that an extra term is introduced
in the horizontal gradient terms that can lead to
artificial vertical diffusion of heat and momentum,
particularly in areas of large topographic gradients as
was described in Schwab and Beletsky (1998). To
help minimize this affect, the 5 km gridded depths
were slightly smoothed by adjusting the depths to
ensure that the relative depth change between
adjacent grid squares was less than 0.5 while still
preserving the volume of the original grid.
The model did not perform as well in the thermocline
area as it performed near the surface. The simulated
thermocline was too diffuse. To study the effect of
vertical resolution on the vertical temperature
gradients, a model run with 39 sigma levels was
conducted. NOAA also ran the model with zero
horizontal diffusion to test for artificial diffusion along
sigma surfaces. In both cases, very little
improvement was noticed in model results. On the
other hand, experiments with an one-dimensional
version of the model showed that the Mellor-Yamada
scheme can provide a sharp thermocline that is
sensitive to the choice of extinction coefficient which
posses significant spatial and temporal variability in
large lakes but was kept constant in the calculations
because at the time of generating the model runs,
temporal data on the extinction coefficients were not
available. In Schwab and Beletsky, 1998, NOAA
mentions that a 2 km grid structure or higher
resolution would likely improve the results, but would
likely push computer resources beyond current limits
for the hydrodynamic model and associated LM3
water quality models.
8. In addition to spatial averaging, there was
apparently time-averaging of hydrodynamic
predictions as well, allowing a daily time-scale for
the LM2-Toxic model. The procedures used to
average the hydrodynamic predictions, and tests
conducted to determine the impact of that
averaging, should be documented.
Modelers' Response - Schwab and Beletsky (1998)
indicate that aggregated average surface heat flux
565
-------
(on an hourly time scale) and average vertical
temperature profiles (on a six hour time scale) were
computed for each of the 10 LM2 surface segments
and the 41 segments, respectively, for both the 1982-
1983 and 1994-1995 periods. In addition, horizontal
and vertical inter-segment transports averaged over
one-day and six-day intervals were computed for the
10 column LM2 grid with five vertical layers: layer
one, 0-10 m; layer two, 10-20 m; layer three, 20-30
m; layer four, 30-50 m; and layer five, 50 m-bottom.
They do not discuss any tests run to determine the
impact of averaging. A discussion on how these data
were used in LM2 can be found in the report in Part
4, Chapter 4, Sections 4.4.1.2 and 4.4.1.3. All
programs and data sets associated with the LM2
aggregation are on the Final Report CD received
from NOAA-GLERL.
9. LM3-Eutro: Table 2.4.6 lists the two "types" of
data but does not describe how the
transformations were made.
Modelers' Response - The relationship between the
variables measured in the field and state variables
used in the model can be found in Appendix 2.4.1 of
the report.
10. LM3-Eutro: Table 1.1.2 does not indicate that
zooplankton were a measured parameter,
although it is a model state variable and the
text indicate that zooplankton data were
collected (Part 2, Chapter 4, Section 2.4.2.2.4).
Modelers' Response - As part of the introductory
material, Table 1.1.2 was only intended to offer a
general overview of the major types of data collected.
For example, some sub-parameters of PCBs such as
a listing of all of the congeners measured, or the fact
that dissolved and particulate were measured are
missing from the table. However, zooplankton is a
major biological and will be included in the table. A
comment will be added where Table 1.1.2 is
referenced in the text explaining that for a complete
listing of parameters measured, the reader should
see Part 1, Chapter 3 of the report. For modeled
parameters, the reader should see individual
chapters on MICHTOX, MICHTOX Food Chain, LM2-
Toxic, LM Food Chain, or LM3-Eutro modeling in the
report.
11. LM3-Eutro: While I agree that the expansion of
variables to include dissolved organic and labile
and refractory particulate organic forms allows
for more realistic description (which is an
increasingly common practice) there are no
established protocols for measuring these
forms. Therefore, it must have been necessary
to make assumptions regarding, for example,
the partitioning of particulates among labile and
refractory forms. Those assumptions should be
described in the report, and perhaps some
sensitivity analyses performed as to the impact
of differing assumptions on model predictions.
The assumptions regarding the split were
indicated (Part 2, Chapter 4, Section 2.4.1.1)
for atmospheric loads, but not for other loading
sources that this reviewer could find.
Modelers' Response - The LM3-Eutro model has
labile and refractory state variables for particulate
nitrogen and phosphorus whereas particulate silica is
in the refractory form only. Since nitrogen was not a
limited nutrient in the model, the evaluation of the two
particulate nutrient forms focuses on phosphorus
only.
Total phosphorus, dissolved phosphorus, and soluble
reactive phosphorus (SRP) were measured in the
water column of Lake Michigan. The labile and
refractory forms of particulate phosphorus can be
calculated based on equations described in Appendix
2.4.1. For initial lake conditions, particulate
phosphorus was evenly split between the labile and
refractory forms. Somewhat different fractions of
particulate phosphorus were used for the labile and
refractory forms in tributary and atmospheric loads
(e.g. the tributary particulate was 0.55 labile and 0.45
refractory - see Part 2, Chapter 4).
The mineralization rates for the two particulate
phosphorus forms used in the model were very
similar; therefore, no significant differences would be
expected when different fractions of these forms are
used in the model. This was confirmed when several
model sensitivity simulations were performed by
varying the initial lake condition and loading
percentages between 25% and 75% for the two
particulate state variables, and in all cases the results
were virtually the same.
566
-------
12. LM3-Eutro: In Part 2, Chapter 4, Section
2.4.4.2, it is stated that laboratory primary
production rates were used to verify the overall
production rates in the model. These
comparisons should be included in the
modeling report.
Modelers' Response - A comparison of the model-
predicted versus laboratory-measured primary
production rates can be found in Figure 2.5.2.
However, there was no reference to this figure in Part
2, Chapter 4, Section 2.4.4.2. (there was a reference
to the graph in Part 2, Chapter 5, Section 2.5.2). The
text will be updated to include a reference to the
figure in Part 2, Chapter 4, Section 2.4.4.2, and the
figure will be moved.
13. LM3-Eutro: The characterization of non-
diatoms versus diatoms is a useful breakdown.
Since blue-greens were the dominant algae
(see page 99), some additional explanation
would be worthwhile as to how nitrogen
limitation was computed for these algae.
Modelers' Response - Blue-green algae was
present in Lake Michigan in large numbers.
However, because of their very small size, they made
up less than 6% of the total phytoplankton carbon
mass. For this reason, we lumped this group in the
model as part of the "non-diatom" algae group and
we assumed that phosphorus, rather than nitrogen,
was the limiting nutrient. The corresponding section
in the revised edition of the report will be updated to
provide a better and more detailed explanation.
14. LM3-Eutro/LM3-Toxic: Continue to develop the
Eutro Model, for both linkages to the Toxic
Model as well as for use related to addressing
conventional pollution in Lake Michigan and its
tributaries/embayments.
Modelers' Response — We do plan to continue to
develop the coupled LM3-Eutro and LM3-Toxic
(PCB) model. SEDZL provides us with a time-
variable resuspension velocity which we will use in
LM3-Toxic (PCB). Current plans include the addition
of particulate resuspension processes to LM3-Eutro
including particulate forms of nutrients and refractory
organic carbon. Eventually, we will add diagenesis to
the sediment compartment and algorithms to
compute dissolved oxygen. See our response to
Number 4 above for additional details.
15. LM3-Eutro: Explore and document methods to
relate measurable field data to model input
values (e.g., refractory particulate organic
matter).
Modelers' Response - This question has been
answered in Number 11 above.
16. Conduct additional calibration (e.g., to nitrogen
series) as an additional test of the model's
performance and if the model may be used to
address questions in the future with regard to
conventional pollution.
Modelers' Response - Because nitrogen does not
drive this model, relatively little time and effort was
spent on the calibration of the different nitrogen
species and was, therefore, not included in the June
2004 draft copy of the report. However, nitrogen will
be fully calibrated in future modeling efforts
especially when addressing lake nutrient and
phytoplankton (chlorophyll-a) issues.
17. LM3-Eutro: The comparisons of model
predictions and field data were somewhat
limited in Part 2, Chapter 5. Additional
comparisons should be provided, both
graphical and statistical, between model
predictions and observed data. Comparisons
should be provided if possible for all state
variables. For example, no comparisons are
presently provided for nitrogen species.
Modelers' Response - In the revised edition of the
report, Part 2, Chapter 5 (Calibration) was expanded
and updated to include additional graphical and
statistical results of the calibration process.
18. LM3-Eutro/LM3 - Toxic: Presently, the LM2-
and LM3-Eutro codes specify sediment fluxes
as zero order rates, which is a common
practice. However, there are models of
sediment diagenesis that allow prediction,
rather than description, of those rates. While
probably not critical in the context of using the
Eutro predictions for input to the toxic model,
incorporation of a sediment diagenesis model
may be worthwhile should the LM3-Eutro model
567
-------
be used in the future to assess eutrophication
related management questions.
Modelers' Response - Sediment diagenesis will be
added to LM3-Eutro. See "Modelers Response" to
Number 4 above for additional planned
enhancements of LM3-Eutro.
19. LM3-Eutro/POM: The linkage of the POM
model with the LM3-Eutro grid was only briefly
discussed. The incorporation of the
QUICKEST-ULTIMATE routines from the U.S.
Army Corps of Engineers CE-QUAL-ICM model
should provide a suitable numerical framework
for that linkage. However, the linkage of
hydrodynamic and water quality models, even
using a one-to-one spatial grid, is not a trivial
task. For example, because of differing
solution schemes, mass imbalances can occur
which, if not properly treated, can accumulate
and impact long-term model predictions. As
such testing is required to ensure that water
and constituent mass are conserved globally
and locally in the linked water quality model.
This testing needs to be documented and
should be included in the modeling report,
perhaps as an appendix.
Modelers' Response - All of these topics and issues
were covered in four reports written by Ray Chapman
associated with the U.S. Army Corps Of Engineers,
Waterways Experiment Station which will be included
in an updated version of the LM3 User Guide (Settles
et al. 2002). This updated LM3 User Guide will be
included as a new appendix in the revised report.
20. LM3-Eutro: The section of the report (Chapter
1) dealing with the calibration of the LM2-Eutro
and LM3-Eutro was somewhat confusing, with
regard to which model was calibrated against
existing data.
Modelers' Response - The LM3-Eutro model is a
Level 3 model with 44,042 (5x5 km2) segments -
there is no LM2-Eutro model. However, as part of
the post-processing, the model was collapsed to a
Level 2 grid, similar to the LM2-Toxic framework.
This enabled a comparison of field data with model
output on the Level 3 grid and on the Level 2 grid.
Part 2, Chapter 5 of the report (Calibration) was
updated to better explain the calibration of LM3-Eutro
on the Level 2 and Level 3 segmentation schemes.
21. LM2/POM: An overlay grid, such as between
the POM model and LM2-Eutro and LM2-Toxic
is often more problematic than using a one-to-
one spatial grid (between a hydrodynamic and
a water quality model). In this application, it
was suggested that linkage problems did occur,
resulting in the necessity of adding "water
balancing flows" (Part 4, Chapter 3, Section
4.3.3). Adding water-balancing flows is not an
uncommon practice in linking three-dimensional
hydrodynamic and water quality models.
Typically those flows are small but without them
water volume imbalances accumulate over
time. However, it was indicated during
presentations that in this study not including the
"balancing" flows resulted in water volumes
going to zero in some water quality segments
(in Green Bay). This is indicative of a linkage
problem that should be further investigated. In
addition, the approach used to compute vertical
exchanges (Equation 4.4.1) should not have
been applicable if vertical flows (gross not net)
were included with the hydrodynamic linkage.
It is suggested that additional testing of the
linkage be conducted and documented within
the modeling report, perhaps as an appendix.
Modelers' Response - LM2-Eutro, referred to in
sentence one, does not exist. Primary productivity is
estimated by LM3-Eutro in space and time, and this
information is exported to LM2-Toxic (PCB). The
overlay between LM3-Eutro and POM is a one-to-one
spatial grid.
We agree with Dr. Martin's comments and
suggestion on the linkage between POM and LM2.
The linkage problem (mass of water is not balanced
in individual segment basis in LM2-Toxic model) was
identified during the period of testing the linkage, and
the water balancing flow was introduced to correct
the imbalance. This problem was noted in a very
small segment volume in Green Bay and after a run
time of 70 years. See Part 4, Chapter 3, Section
4.3.3 of the report for additional discussion on this
issue. The results of the test will be included as an
appendix in the revised report. NOAA-GLERL
performed the linkage calculations between POM
568
-------
and LM2, so any further investigation would need to
be referred to them.
Dr. Martin is also correct on the applicability of
vertical exchanges. The vertical flows provided by
NOAA based on POM outputs are the net vertical
flows; therefore, the vertical exchange process must
be added in the overall water transport field used in
LM2-Toxic model. Additional discussion and
description of the method of calculating vertical
exchange coefficients and calibration are
documented in Part 4, Chapter 3, Section 4.3.3 and
Part 4, Chapter 4, Section 4.4.1 in the report.
22. LM2-Toxic: As indicated in Part 2, Chapter 6,
Section 4.6.2, the flux contributed by the
diffusive term from the sediment bed was
unexpectedly large relative to the resuspension
flux. This may have been due to the relatively
large specified diffusion coefficient used relative
to the Level 1 model. In addition, it was
indicated in Part 4, Chapter 6, Section 4.6.2
that the total PCB residence time for Lake
Michigan were on the order of 100 days. This
estimate seems low to this reviewer. It would
be interesting to see how this compares to
predictions from a Level 3 model which may
more realistically estimate vertical exchanges in
layers isolated from the water surface. The
Level 3 model could be used to determine if the
rapid removal may be in part an artifact of the
modeling approach used in the Level 2 studies.
As an example, given the rates of settling used,
surface particles would require approximately
one year to reach the bottom, while with a
single vertical-box model it would be assumed
that vertical transport is on average
instantaneous.
Modelers' Response - We agree with Dr. Martin
regarding the high flux contributed by diffusion
between the water column and surficial sediment
layer. We plan to investigate model responses in
both the water column and sediment to various mass
fluxes across the sediment-water interface by
changing the diffusion coefficient and/or mixing
length between the water column and surficial
sediment as identified in Part 4, Chapter 2 of the
report. There is further discussion on the diffusion
coefficient used in LM2-Toxic model in Part 4,
Chapter 6, Section 4.6.2 of the report. This also is
one of the recommendations in Chapter 2 of Part 4.
Like Dr. Martin, we also noticed that the total PCB
residence time for Lake Michigan is relatively low and
will do the comparison with the LM3 model when
these results become available.
23. LM2-Toxic: Apply the model to refine whole-
lake estimates of PCB concentrations.
Modelers' Response-Whole lake, volume-weighted
average concentrations of total PCBs in the lake can
be found in Part 4, Chapter 6, Section 4.6.4 of the
report for various load reduction scenarios, including
the "No-Effects Action."
24. LM2-Toxic: Extend the modeling framework to
include other contaminants of concern (e.g.,
mercury).
Modelers' Response - We will extend the modeling
framework to include other contaminants such as
mercury and frans-nonachlor and believe the LM2-
Toxic model would make an excellent and easy-to-
use screening and diagnostic tool for helping
management personnel and policy makers to
understand the key processes controlling the level of
the contaminant of interest in the water system.
25. LM2-Toxic: The comparisons of measured and
simulated concentrations seem reasonable.
However, since differences occur between
factors controlling PCBs in Lake Michigan and
Green Bay, the results for these two systems
should be reported separately.
Modelers' Response - For LM2-Toxic, most field
measured data were interpolated separately for the
two systems. We also reported the two systems
together and separately for model calibration results
and mass budget diagnosis. See Part 4, Chapters 5
and 6 of the report for details.
26. LM2-Toxic/LM3-Toxic: The sediment bed
model seems reasonable. However, some
additional clarification of the semi-lagrangian
method for simulating the sediment bed (Part 4,
Chapter 3, Section 4.3.4.2.3) would be useful.
In addition, the present construct does not
allow for the tracking of materials buried out of
the layer, or perhaps entrained into the layer
from deeper contaminated sediments. Some
569
-------
additional development of the sediment
algorithms would be useful for the Level 2
model and for incorporation into the Level 3
framework where it may be more important with
regard to near-shore issues).
Modelers' Response - There is a detailed description
on the semi-Lagrangian method for simulating
sediment bed in the IPX model user manual (Velleux
et a/., 2000). An algorithm for tracking masses and
fluxes of both solids and PCBs in the sediment bed,
including deeper sediments, has already been
implemented in the LM2-Toxic model codes. We will
consider additional development of LM2 sediment
algorithms for incorporation into LM3 as another
option. Also, see "Modelers Response" to Number 4
above for additional enhancements planned for the
sediment bed model.
27. LM Food Chain: Continue to develop and
refine the food chain model.
Modelers' Response - Additional calibration
simulations will be run with reconstructed historical
exposure PCB concentrations in water and sediment
as inputs. The new calibration results will be
compared with those from the constant exposure
assumption. The results will be provided and
discussed in the report. Further development and
refinement of the model will be carried out when
additional data for the lake trout food web at
Sheboygan and for the coho salmon population
become available.
28. LM Food Chain: Extend the calibration period
to an evaluation of historical loadings and/or a
period encompassing all available data (not just
the 1994-1995 data set).
Modelers' Response - A new set of calibration
procedures will be performed using estimated
temporal profiles of historical PCB exposure
concentrations in water and sediment as model
inputs. All currently available field PCB monitoring
data for fish in Lake Michigan will be compiled and
used in the calibration. The results will be provided
in the final report. However, the credibility of the
calibration results will be impaired by the lack of
historical information regarding food web structures
and dietary shift, age-specific PCB data for lake trout,
PCB data for forage fish, PCB compositional change,
and congener-specific PCB values.
29. LM Food Chain: Use the model along with any
revisions made to the LM2-Toxic to refine
estimates of future trends in fish PCB
concentrations.
Modelers' Response - As revised PCB exposure
scenarios in the water and sediment provided by
LM2-Toxic becomes available, new model projections
of future trends in lake trout PCB concentrations will
be made and reported in the final report.
30. LM Food Chain: Initiate extending the model
(and data analysis) to other pollutants of
concern (e.g., mercury).
Modelers' Response - It has been our intention to
expand the model to other pollutants of concern,
including application to other organic chemicals such
as frans-nonachlor and mercury.
31. LM Food Chain/LM3-Eutro: Perhaps some
more direct coupling of the eutrophication and
food chain model could be considered in future
applications to aid in addressing questions
regarding impacts of changes in food chain
structure on uptake of PCBs and other
toxicants.
Modelers' Response-This recommendation was not
addressed in the report. The possibility of direct
coupling of the eutrophication and food chain models
could be explored in future applications. However,
the current state of understanding regarding the
mechanism of fish dietary selection/adaption does
not permit prediction of changes in food chain
structures with eutrophication data. Further
investigation is required on this topic before attempts
are made to couple the food web model with the
eutrophication model to address PCB uptake issues.
32. LM3-Toxic/LM2-Toxic: Continue to develop the
LM-3 model in order to test against the LM2-
Toxic predictions to estimate the potential
impact of a more physically realistic model on
lake-wide PCB impacts.
Modelers' Response - We do plan to further develop
the LM3 models to compare to LM2-Toxic. See
570
-------
"Modelers' Response" to comments in Number 4
above. Comparison of the models on a lake-wide
basis, such as volume-weighted averages, will be
performed.
33. LM3-Toxic: Continue to develop the LM3
Model in order to aid in addressing nearshore
impacts which can not be addressed using the
LM2 structure.
Modelers' Response - We will continue to develop
LM3-Toxic and LM3-Eutro/Eco to address nearshore
impacts as best as can be accomplished within the
limitations of the 5 km grid structure. Much finer
space scales would need to be implemented in a
model, however, in order to model specific harbors
associated with an Area of Concern or tributary
mouths. In those cases, a finer scaled model would
need to be constructed, and LM3 could be used to
provide the boundary condition for this finer-scaled
model construct.
34. LM3-Toxic: Continue to develop and test the
linkage between the POM and LM3 models
(both Eutro and Toxic), such as testing to
ensure that mass conservation is maintained.
Modelers' Response - A mass conservation test was
performed several years ago on LM3-Toxic using a
conservative tracer. During the test, all loadings
were shut off. All lake cell concentrations were
equal. The model was run for two years, and no
noticeable change was detected in the
concentrations.
35. LM3-Toxic/SEDZL: Continue to explore
linkages or incorporation of SEDZL routines in
the Level 3 models. This linkage may be of
particular importance in evaluating nearshore
trends and issues.
Modelers' Response - We will continue to pursue
linking SEDZL output to Level 3 models as described
in the "Modelers' Response" to Number 4 above.
If we decide to further develop the linkage beyond
the current construct, we would likely choose SEDZL-
J as it is now being promoted by the experts as the
better model to use versus SEDZL. SEDZL-J utilizes
SED-Flume data and also allows for non-constant
vertical sediment profile data. SED-Flume measures
the total erosion rate on actual sediment cores and
includes both the rate at which sediments are
transferred to the water column (resuspension), but
also the bed-load rate. SEDZL-J includes bed load
and bed armoring whereas SEDZL does not. SED-
Flume data as well as bulk density profiles have been
collected on Lake Michigan cores by the University of
California at Santa Barbara. A project to apply
SEDZL-J was proposed, but funding was not
available. Of course, if bedload, armoring, and non-
constant vertical sediment profile data are not issues,
then there does not seem to be a great advantage of
SEDZL-J over SEDZL. A potential limitation of
applying either SEDZL or SEDZL-J's, two-
dimensional models to deeper portions of Lake
Michigan during stratification exists. At this time, a
three-dimensional, SEDZL-J, is not available.
The extent of this impact on our resuspension rate
estimates is not known.
References
Beletsky, D. and D. J. Schwab. 2001. Modeling
Circulation and Thermal Structure in Lake
Michigan: Annual Cycle and Interannual
Variability. J. Geophy. Res., 106(C9): 19745-
19771.
Endicott, D.D. 2005. 2002 Lake Michigan Mass
Balance Project: Modeling Total PCBs Using the
MICHTOX Model. In: R. Rossmann (Ed.),
MICHTOX: A Mass Balance and
Bioaccumulation Model for Toxic Chemicals in
Lake Michigan, Part 2. U.S. Environmental
Protection Agency, Office of Research and
Development, National Health and Environmental
Effects Research Laboratory, MED-Duluth, Large
Lakes Research Station, Grosse lie, Michigan.
EPA/600/R-05/158, 140 pp.
Franz, T.P., S.J. Eisenreich, and T.M. Holson. 1998.
Dry Deposition of Particulate Polychlorinated
Biphenyls and Polycyclic Aromatic Hydrocarbons
to Lake Michigan. Environ. Sci. Technol.,
32(23):3681-3688.
Gobas, F.A.P.C., M.N. Z'Graggen, and X. Zhang.
1995. Time Response of the Lake Ontario
Ecosystem to Virtual Elimination of PCBs.
Environ. Sci. Technol., 29(8):2038-2046.
571
-------
Hillery, B.R., I. Basu, C.W. Sweet, and R.A. Hites.
1997. Temporal and Spatial Trends in Long-
Term Study of Gas-Phase PCB Concentrations
near the Great Lakes. Environ. Sci. Technol.,
Holsen, T.M., K.E. Noll, S. Liu, and W. Lee. 1991.
Dry Deposition of Polychlorinated Biphenyls in
Urban Areas. Environ. Sci. Technol. 25(6):1075-
1081.
Kuehl, M. 2002. Polychlorinated Biphenyl (PCB)
Congener Patterns in Lake Michigan Mass
Balance Study Biota. Master's Thesis, University
of Wisconsin, Green Bay, Wisconsin. 120 pp.
Marti, E.A. and D.E. Armstrong. 1990.
Polychlorinated Biphenyls in Lake Michigan
Tributaries. J. Great Lakes Res., 16(3):396-405.
McCarty, H.B., J. Schofield, K. Miller, R.N. Brent, P.
Van Hoof, and B. Eadie. 2004. Results of the
Lake Michigan Mass Balance Study:
Polychlorinated Biphyenyls and frans-Nonachlor
Data Report. U.S. Environmental Protection
Agency, Great Lakes Program Office, Chicago,
Illinois. EPA-905/R-01/01 1 , 289 pp.
Miller, S.M., M.L. Green, J.V. DePinto, and K.C.
Hornbuckle. 2001. Results From the Lake
Michigan Mass Balance Study: Concentration
and Fluxes of Atmospheric Polychlorinated
Biphenyis and frans-Nonachlor. Environ. Sci.
Technol., 35(2):278-285.
Schneider, A.R., H.M. Stapleton, J. Cornwell, and
J.E. Baker. 2001. Recent Declines in PAH,
PCB, and Toxaphene Levels in the Northern
Great Lakes as Determined From High
Resolution Sediment Cores. Environ. Sci.
Technol., 35(19):3809-3815.
Schwab, D. and D. Beletsky. 1998. Lake Michigan
Mass Balance Study: Hydrodynamic Modeling
Project. National Oceanic and Atmospheric
Administration, Great Lakes Environmental
Research Laboratory, Ann Arbor, Michigan.
Technical Memorandum ERLGLERL-108,53 pp.
Settles, M., W. Melendez, and J. Pauer. 2002. LM3:
The Lake Michigan Mass Balance Model.
Internal Report. U.S. Environmental Protection
Agency, Office of Research and Development,
National Health and Environmental Effects
Research Laboratory, MED-Duluth, Large Lakes
Research Station, Grosse lie, Michigan. 203 pp.
Velleux, M. and D.D. Endicott. 1994. Development
of a Mass Balance Model for Estimating PCB
Export from the Lower Fox River to Green Bay.
J. Great Lakes Res., 20(2):416-434.
Velleux, M., S. Westenbroek, J. Ruppel, M. Settles,
and D.D. Endicott. 2000. A User's Guide to IPX,
the In-Place Pollutant Export Water Quality
Modeling Framework, Version 2.7.4. U.S.
Environmental Protection Agency, Office of
Research and Development, National Health and
Environmental Effects Research Laboratory,
MED-Duluth, Large Lakes Research Station,
Grosse Me, Michigan. 179 pp.
Wethington, D.M. and K.C. Hornbuckle. 2005.
Milwaukee, Wl, as a Source of Atmospheric
PCBs to Lake Michigan. Environ. Sci. Technol.,
39(1 ):57-63.
572
-------
PART?
APPENDICES
Appendix 7.2 Comments as Received
From Dr. James Martin Peer Review
Summary: Lake Michigan Mass Balance
Project
James L. Martin
Department of Civil Engineering
Mississippi State University
Mississippi State, Mississippi 39762
7.2.1 General Comments
These comments are based upon a review of the
draft (June 24, 2004) documentation "Results of the
Lake Michigan Mass Balance Project: PCBs
Modeling Report" as well as materials presented and
discussed at the July 27-28, 2004 Peer Review
Workshop held in Romulus, Michigan. The charge of
the peer review was to focus on and address three
major categories each with subcomponents,
considering strengths and weaknesses:
1. Overall Multimedia Ecosystem Modeling
Approach
2. Overall Model Performance
3. Suitability for Management
Each of these topics are discussed below followed by
a summary of recommendations for continued and
future development.
7.2.1.1 Overall Multi-Media Ecosystem Modeling
Approach
The overall multi-media and mass balance approach
is a necessity to a system like Lake Michigan where
both the biota concentrations are an end-point for
management decisions and biota impact the PCB
cycling (since such a large component of the organic
solids are of biotic origin). It is also necessary to
include the hydrodynamics of Lake Michigan, since
hydrodynamics impacts contaminant transport. For
the analysis of loadings, consideration of loadings
from all media (tributaries, the atmosphere, etc.) is
also essential. The overall multi-media approach as
implemented in the Lake Michigan Mass Balance
Project included all of these components. The
specific components and their relative importance will
shift as issues progress from whole-lake to nearshore
areas and from PCBs to other contaminants such as
mercury. However, the framework developed for the
Lake Michigan mass balance studies will provide a
suitable base for the extension of the overall
approach into other areas and to other chemicals of
concern.
7.2.1.2 Overall Model Performance
The model(s) is (are) considered by this reviewer to
adequately representthephysical/biological/chemical
processes impacting PCB concentrations in the
water, sediment, and biota of Lake Michigan. The
present construct is considered limited in its
applicability to whole lake issues. However, the
extension of the framework to the LM3 level should
allow for the modeling system to be used to address
nearshore issues as well. The modeling framework
573
-------
is also suitable as a basis for the development of
models of other chemicals of concern, such as
mercury.
7.2.1.3 Suitability for Management
An assessment of the suitability of the modeling
framework for management requires first delineating
the specific management questions that framework
will be asked to address. The strength of the studies
completed to date is that they provide a framework
that can be used in the contributing toward the
"weight-of-evidence" with regard to the relative
importance of loading sources on the average
concentrations of PCBs in the lake as well as in
estimating the time required for natural recovery in
the water column and biota. To the extent that this
weight-of-evidence relates to the management goals
for the Lake, the modeling framework is suitable. It
integrates the present understanding of factors
impacting PCBs in Lake Michigan. While the spatial
segmentation of the LM2-Toxic model is more
detailed than that of the Level 1 models, the structure
is perhaps best suitable for refining whole Lake
estimates of PCB concentrations rather than
predicting local variations.
However, there are a number of issues, such as
nearshore and tributary issues, that the present
model Level 2 models cannot address. For example,
the LM2-Toxic model cannot be used to address
issues related to specific Areas of Concern, other
than as a lake-wide average. It is expected that the
LM3-Toxic model, when completed, would be more
suitable for addressing local variations in PCB
concentrations and exposure.
An additional major strength of the study is serving
as a framework for the evaluation of available data
and in the planning of future data (and modeling)
efforts. For example, analyses designed to
determine model uncertainty provide not only
information concerning predictive uncertainty but can
guide monitoring efforts to reduce that uncertainty.
An iterative program of model development and data
collection may provide for the most efficient use of
limited resources, particularly given the reasonably
long time frame before some of the issues (such as
the Total Maximum Daily Load (TMDL) for Lake
Ontario) need to be addressed. This iterative
approach to the collection and analysis of data,
exemplified by the Levels 1 and 2 studies and leading
to Level 3 is clearly an effective means of organizing
all of the myriad efforts and parties involved in the
collection and analysis of data. The participants are
to be commended on the demonstrated efficacy of
the use of the mass balance approach in the design
of the PCB study.
7.2.2 Specific Recommendations
7.2.2.1 POM and Linkages
1. Continue development of the linked POM and
Level 2 and Level 3 models.
2. Provide documentation of the POM application,
perhaps as an appendix to the modeling report.
3. Assist in developing ice cover algorithms and
linkages with water quality model.
4. Investigate potential linkages issues between
POM and with SEDZL.
5. Investigate assumptions/limitations of using a
sigma grid, particularly in resolving both
nearshore and open-lake issues.
7.2.2.2 LM2-Eutro and LM3-Eutro
1. Continue to develop the Eutro model, for both
linkages to the Toxic model as well as for use
related to addressing conventional pollution in
Lake Michigan and its tributaries/embayments.
2. Explore and document methods to relate
measurable field data to model input values (e.g.,
refractory particulate organic matter).
3. Conduct additional calibration (e.g., to nitrogen
series) as an additional test of the model's
performance and if the model may be used to
address questions in the future with regard to
conventional pollution.
4. Consider including a sediment diagenesis model
if Eutro will be used in the future to address
management questions related to conventional
pollution.
574
-------
5. Consider exploring a more direct linkage with the
food chain model, to address potential changes
in the food chain structure.
7.2.2.3 Level 1 Model
1. Continue to maintain the Level 1 model,
particularly for comparison with Level 2
predictions.
2. Explore incorporating specific algorithms, such as
the steady-state algorithm, with the Level 2 and
potentially Level 3 models.
7.2.2.4 LM2-Toxic
1. Test the linkage with the POM model by running
conservative tracer test to insure that a mass
balance is maintained.
2. Revisit and refine sediment component of the
model (e.g., number of layers).
3. Extend the calibration period to an evaluation of
historical loadings and/or a period encompassing
all available data (not just the 1994-1995 data
set).
4. Investigate why similar predictions were obtained
to those from the Level 1 for what appears to be
dissimilar reasons (differences in settling
velocities, diffusion rates, bed thickness, etc.).
5. Apply the model to refine whole-lake estimates of
PCB concentrations.
6. Extend the modeling framework to include other
contaminants of concern (e.g., mercury).
7.2.2.5 LM Food Chain
1. Continue to develop and refine the food chain
model.
2. Extend the calibration period to an evaluation of
historical loadings and/or a period encompassing
all available data (not just the 1994-1995 data
set).
3. Consider investigating a more direct linkage with
biotic models (such as LM3-Eutro).
4. Use the model along with any revisions made to
LM2-Toxic to refine estimates of future trends in
fish PCB concentrations.
5. Initiate extending the model (and data analysis)
to other pollutants of concern (e.g., mercury).
7.2.2.6 LM3-Toxic
1. Continue to develop the LM3 model in order to
test against the LM2-Toxic predictions to
estimate the potential impact of a more physically
realistic model on lake-wide PCB impacts.
2. Continue to develop the LM3 model in order to
aid in addressing nearshore impacts which can
not be addressed using the LM2 structure.
3. Continue to develop and test the linkage between
the POM and LM3 models (both Eutro and
Toxic), such as testing to ensure that mass
conservation is maintained.
4. Continue to explore linkages or incorporation of
SEDZL routines in the Level 3 models. This
linkage may be of particular importance in
evaluating nearshore trends and issues.
7.2.3 Specific Comments
What follows are some specific comments and
observations regarding each of the modeling
components. Some of these comments request
clarification of specific assumptions and methods
used in the development of the models. While the
documentation provided was extensive, there were
specific areas identified where additional information
and/or clarification would be helpful.
7.2.3.1 Hydrodynamics and POM Linkage
The hydrodynamic model used in this application for
LM3-Eutro, and planned for LM3-Toxic, is the
Princeton Ocean Model (POM). POM is widely used
and accepted and is similar in construct to several
other hydrodynamic models in common usage (e.g.,
the ECOM model, which was based largely on POM,
and EFDC). However, there was no information
provided in the subject modeling report as to the
model application and testing of the model,
particularly with regard to an assessment of the
575
-------
applicability of the model to the transport of PCBs
and other water quality constituents. The information
provided was limited to a brief presentation by
D'mitry Beletsky and David Schwab. As such, there
was not sufficient information presented to assess
the application or performance of the hydrodynamic
model. Since the hydrodynamic model is a critical
component of the Levels 2 and 3 studies, it is
suggested that documentation of the model
application be provided in the PCB modeling report,
perhaps as an appendix.
One potential limitation to the POM model construct
(relative to this application) is related to the
coordinate system used in the vertical dimension (a
sigma grid). A sigma grid requires a constant
number of vertical layers throughout the model
domain (beneath each of the 5 km horizontal grid
cells (the number of vertical layers was variously
cited as from 15 to 20 in the modeling report, which
should be corrected). This use of the sigma grid may
impact the ability of the model to resolve vertical
gradients, particularly in deeper sections of the lake
while still sufficiently capturing nearshore circulation
patterns. In addition, sigma grids may produce
artificial horizontal transport patterns. While there
are numerical schemes for compensating for this, I
am not aware that they have been implemented in
POM or that any sigma transport tests have been
conducted for an application such as Lake Michigan.
A second potential limitation of the POM model is its
present inability to predict ice cover. From the
presentations, it was suggested that ice algorithms
will be added to the model and it is recommended
that the incorporation of ice algorithms be pursued.
The linkage of the POM model with the LM3-Eutro
grid was only briefly discussed. The incorporation of
the QUICKEST-ULTIMATE routines from the U.S.
Army Corps of Engineers CE-QUAL-ICM model
should provide a suitable numerical framework for
that linkage. However, the linkage of hydrodynamic
and water quality models, even using a one-to-one
spatial grid, is not a trivial task. For example,
because of differing solution schemes, mass
imbalances can occur which, if not properly treated,
can accumulate and impact long-term model
predictions. As such, testing is required to ensure
that water and constituent mass are conserved
globally and locally in the linked water quality model.
This testing needs to be documented and should be
included in the modeling report, perhaps as an
appendix.
An overlay grid, such as between the POM model
and LM2-Eutro and LM2-Toxic is often more
problematic than using a one-to-one spatial grid
(between a hydrodynamic and water quality model).
In this application, it was suggested that linkage
problems did occur resulting in the necessity of
adding "water balancing flows" (Part 4, Chapter 3,
Section 4.3.3). Adding water-balancing flows is not
an uncommon practice in linking three-dimensional
hydrodynamic and water quality models. Typically
those flows are small but without them water volume
imbalances accumulate over time. However, it was
indicated during presentations that in this study, not
including the "balancing" flows resulted in water
volumes going to zero in some water quality
segments (in Green Bay). This is indicative of a
linkage problem that should be further investigated.
In addition, the approach used to compute vertical
exchanges (Equation 4.1.1) should not have been
applicable if vertical flows (gross not net) were
included with the hydrodynamic linkage. It is
suggested that additional testing of the linkage be
conducted and documented within the modeling
report, perhaps as an appendix.
In addition to spatial averaging, there was apparently
time-averaging of hydrodynamic predictions as well,
allowing a daily time-scale for the LM2-Toxic model.
The procedures used to average the hydrodynamic
predictions, and tests conducted to determine the
impact of that averaging, should be documented.
In general, the linkage of POM with the LM models
represents an advancement and provides additional
capabilities that should be continued to be
developed. For example, the coupled model should
more accurately predict the transport patterns in the
lake, which are always of questionable accuracy
when based on purely descriptive techniques. In
addition, the coupled model may be more readily
applied to predict conditions in more localized areas,
such as nearshore, and to predict conditions (such as
extreme events) that cannot be adequately
characterized using a descriptive approach.
576
-------
7.2.3.2 LM2-Eutro and LM3-Eutro
The eutrophication model is an important component
of the overall multi-media modeling approach. The
addition of an eutrophication model is of particular
importance for Lake Michigan due to the reported
large fraction of total sorbents that are of biotic origin
(reported to be 90 percent of the total organic carbon
load to the lake, Part 4, Chapter 3, Section 4.3.4).
The overall structure of the eutrophication model
seems reasonable, and seems comparable to other
eutrophication models. The majority of the
comments provided below are related to clarifications
that would be helpful in the modeling report.
Some additional description is needed as to how the
field data (as listed in Table 1.1.2) were converted to
model input or used for model comparisons. Table
2.4.6 lists the two "types" of data but does not
describe how the transformations were made. Table
1.1.2 does not indicate that zooplankton were a
measured parameter, although it is a model state
variable and the text indicate that zooplankton data
were collected (Part 2, Chapter 4, Section 2.4.2.2.4).
Also, while I agree that the expansion of variables to
include dissolved organic and labile and refractory
particulate organic forms allows for more realistic
description (which is an increasingly common
practice) there are no established protocols for
measuring these forms. Therefore, it must have
been necessary to make assumptions regarding, for
example, the partitioning of particulates among labile
and refractory forms. Those assumptions should be
described in the report, and perhaps some sensitivity
analyses performed as to the impact of differing
assumptions on model predictions. The assumptions
regarding the split were indicated (Part 2, Chapter 4,
Section 2.4.1.1) for atmospheric loads, but not for
other loading sources that this reviewer could find.
In Part 2, Chapter 4, Section 2.4.4.2, it is stated that
laboratory primary production rates were used to
verify the overall production rates in the model.
These comparisons should be included in the
modeling report.
The characterization of non-diatoms versus diatoms
is a useful breakdown. Since blue-greens were the
dominant algae, some additional explanation would
be worthwhile as to how nitrogen limitation was
computed for these algae.
The section of the report (Chapter 1) dealing with the
calibration of the LM2-Eutro and LM3-Eutro was
somewhat confusing, with regard to which model was
calibrated against existing data.
The comparisons of model predictions and field data
were somewhat limited in Chapter 5. Additional
comparisons should be provided, both graphical and
statistical, between model predictions and observed
data. Comparisons should be provided if possible for
all state variables. For example, no comparisons are
presently provided for nitrogen species.
Presently, the LM2-Eutro and the LM3-Eutro codes
specify sediment fluxes as zero order rates, which is
a common practice. However, there are models of
sediment diagenesis that allow prediction, rather than
description, of those rates. While probably not critical
in the context of using the Eutro predictions for input
to the Toxic model, incorporation of a sediment
diagenesis model may be worthwhile should the
LM3-Eutro model be used in the future to assess
eutrophication-related management questions.
In general, the linkage of the Lake Michigan
eutrophication and toxicant models represents an
advancement and provides additional capabilities that
should be continued to be developed. This reviewer
considers the existing eutrophication model construct
sufficient for its intended use, to provide biotic solids
for the toxicant model. However, the eutrophication
model is also considered important in its own right,
and should have applicability in addressing questions
regarding conventional pollutants in Lake Michigan.
In addition, perhaps some more direct coupling of the
eutrophication and food chain model could be
considered in future applications, to aid in addressing
questions regarding impacts of changes in food chain
structure on uptake of PCBs and other toxicants.
7.2.3.3 Level 1 Models
The inclusion of the Level 1 model in the modeling
report and presentations, and the contrasting of
model construct and predictions with the Level 2
model, was considered by this reviewer to be very
useful. First, the Level 1 model and its predictions
were useful in providing insights into factors
impacting PCBs in Lake Michigan and addressing
interim management questions. In addition, the
Level 1 modeling studies illustrated what this
577
-------
reviewer considers to be one of the best uses of
models and modeling studies: to first aid in mining
and interpreting available data, to then identify
deficiencies in available data and modeling
approaches, and finally to aid in planning additional
studies and model refinements.
One area that perhaps deserves further investigation
is the similarity in predictions of the Level 1 and Level
2 models. Both models predicted remarkably similar
changes in total PCB concentrations over time in the
long-term projections. However, there were
differences between the two models such as in the
rates of settling/resuspension used and in the
characterization of the sediment bed. As a result, the
two models predicted similar results for somewhat
dissimilar reasons. It would be of interest to further
investigate factors leading to the similarity in
predictions, which may provide some additional
insights as to factors controlling PCBs in Lake
Michigan.
There were some capabilities of the Level 1 model
which should be considered for incorporation into the
Level 2 model. One such capability is the steady-
state solution. A goal in future studies, as expressed
during the presentations, was to assess uncertainty
in the Level 2 model. Uncertainty is most commonly
assessed using steady-state rather than dynamic
predictions. The long simulation time required to
achieve steady-state predictions in the dynamic Level
2 model may preclude conducting uncertainty
analyses. Incorporating steady-state solution
techniques in the Level 2 (and ultimately the Level 3
model) would facilitate the analysis.
7.2.3.4 LM2-Toxic
The Level 2-Toxic model represents an advancement
over its predecessor, the Level 1 model. These
advancements not only include simulation of PCB
congeners, but improvements in transformation
kinetics, such as volatilization. A number of these
improvements resulted from Level 2 investigations,
and the study serves as a very good example of the
benefits achieved through the iterative development
and refinements of models
The coupling of the POM predictions to the Level 2
model seems a reasonable approach. However,
using a 1-1 grid rather than a course-grid overlaying
a fine-grid hydrodynamic model is a preferable
approach, which is the approach planned for the
Level 3 model. Several recommendations regarding
testing of the linkage between the POM
hydrodynamic model and both the LM2-Toxic and
LM3-Toxic model were discussed in a previous
section.
With regards to solids transport, the approach used
for computing sediment resuspension seems
reasonable. However, it is hoped that a more
detailed sediment model (SEDZL, which was part of
the original plan) can be incorporated into the Level
3 framework. The settling velocities used also seem
reasonable but are lower than those used in the
Level 1 study. Since the estimated resuspension
velocities will vary with the settling velocities, the
rates used are also presumably lower than those
used in the Level 1 studies. Since the projections of
the two models were remarkably similar, some
additional investigation as to why similar predictions
were obtained using dissimilar rates would be
worthwhile.
The sediment bed model seems reasonable.
However, some additional clarification of the semi-
Lagrangian method for simulating the sediment bed
(Part 4, Chapter 3, Section 4.3.4.2.3) would be
useful. In addition, the present construct does not
allow for the tracking of materials buried out of the
layer, or perhaps entrained into the layer from deeper
contaminated sediments. Some additional
development of the sediment algorithms would be
useful for the Level 2 model and for incorporation into
the Level 3 framework (where it may be more
important with regard to nearshore issues).
As indicated in Part 4, Chapter 6, Section 4.6.2, the
flux contributed by the diffusive term from the
sediment bed was unexpectedly large relative to the
resuspension flux. This may have been due to the
relatively large specified diffusion coefficient used
relative to the Level 1 model. As suggested above,
some additional testing is recommended to compute
and compare factors causing predicted variations
between the Level 1 and Level 2 studies, and the
diffusive fluxes should be considered in that testing.
In addition, it was indicated in Part 4, Chapter 6,
Section 4.6.2 that the total PCB residence time for
Lake Michigan were on the order of 100 days. This
estimate seems low to this reviewer. It would be
578
-------
interesting to see how this compares to predictions
from a Level 3 model which may more realistically
estimate vertical exchanges in layers isolated from
the water surface. The Level 3 model could be used
to determine if the rapid removal may be in part an
artifact of the modeling approach used in the Level 2
studies. As an example, given the rates of settling
used, surface particles would require approximately
one year to reach the bottom, while with a single
vertical-box model it would be assumed that vertical
transport is on-average instantaneous.
The comparisons of measured and simulated
concentrations seem reasonable. However, since
differences occur between factors controlling PCBs
in Lake Michigan and Green Bay, the results for
these two systems should be reported separately.
One limitation to the Level 2 application, and to many
similar studies, was the limited time-scale to which
the model was applied. The model was applied and
calibrated using data from the 1994 and 1995 field
studies. Given the time scale of changes in Lake
Michigan PCB concentrations, this period is not
sufficient to test the model against long-term trends
in the PCB concentrations for Lake Michigan.
Similarly to the Level 1 studies, it would be
worthwhile as an additional test to run the model with
estimated historical loadings and for comparison to
all existing data, including data from this sampling
period. Such an application would provide additional
testing of the robustness of the model, particularly
since the models intended use is in the projection of
long-term trends in PCB concentrations for Lake
Michigan.
7.2.3.5 LM Food Chain
The food chain model used in this study was based
upon what I consider to be a widely accepted
approach and which I consider adequate for the
purposes of this study. However, my experience in
food chain modeling is limited and dated so I would
defer to others with more recent experience to
evaluate this component of the model. As with the
LM2-Toxic model, the application of the model to a
longer period of record is recommended as an
additional test of the model.
579
-------
EPA600/R-04/167 Dec. 2006
c.3
Results of the Lake Michigan Mass Balance
Project: polychlorinated biphenyls
modeling report
US EPA
MID-CONTINENT
ECOLOGY DIVISION
LABORATORY LIBRARY
DULUTH, MN
DEMCO
-------
United States
Environmental Protection
Agency
Office of Research and Development (8101R)
Washington, DC 20460
Official Business
Penalty for Private Use
$300
EPA/600/R-04/167
December 2006
www.epa.gov
PRESORTED STANDARD
POSTAGE & FEES PAID
EPA
PERMIT No. G-35
Recycled/Recyclable
Printed with vegetable-based ink on
paper that contains a minimum of
50% post-consumer fiber content
processed chlorine free
-------
|