Jump to main content.

Baseline Modeling Report

Executive Summary
May 1999

This report presents results and findings from the application of mathematical models for PCB transport and fate and bioaccumulation in the Upper Hudson River. The modeling effort for the Hudson River PCBs Site Reassessment has been designed to predict future levels of PCBs in Hudson River sediment, water and fish. This report provides predictions under baseline conditions, that is, without remediation (equivalent to a No Action scenario). The outputs from the models, baseline sediment, water and fish PCB concentrations will be used as inputs in the Human Health and Ecological Risk Assessments. Subsequently, the models will also be used in the Feasibility Study (the Phase 3 Report) to evaluate and compare the impacts of various remedial scenarios.

The Baseline Modeling Report (BMR) consists of four books. Books 1 and 2 are on the transport and fate models, with Book 1 containing the report text and Book 2 containing the corresponding tables, figures and plates. Similarly, Books 3 and 4 are on the bioaccumulation models, with Book 3 containing the report text and Book 4 containing the corresponding tables, figures and plates. Predictions from the transport and fate models are used as input values for the bioaccumulation models.

MODELING  OBJECTIVES - The overall goal of the modeling is to develop and field validate scientifically credible models in order to answer the following principal questions:

1. When will PCB levels in fish populations recover to levels meeting human health and ecological risk criteria under continued No Action?

2. Can remedies other than No Action significantly shorten the time required to achieve acceptable risk levels?

3. Are there contaminated sediments now buried that are likely to become "reactivated" following a major flood, possibly resulting in an increase in contamination of the fish population?

The work presented in this Baseline Modeling Report provides information relevant to the first and third questions. Predictions regarding the potential impacts of various remedial scenarios, the second question, will be conducted in the future and be presented in the Feasibility Study (the Phase 3 Report).


A large body of information from site-specific field measurements (as found in Hudson River Database Release 4.1), laboratory experiments and the scientific literature was synthesized within the models to develop the transport and fate and bioaccumulation models. Data from numerous sources were utilized including USEPA, the New York State Department of Environmental Conservation, the National Oceanic and Atmospheric Administration, the US Geological Survey and the General Electric Company.

The proposed modeling approach, a description of the data sets to be used for calibration, and demonstrations of model outputs were made available for public review in the Preliminary Model Calibration Report (PMCR), which was issued in October 1996. In addition, in September 1998, an independent peer review was held on the modeling approach. The modeling framework of the PMCR was revised based on the peer review and public comment. A major revision was the addition of a mechanistic bioaccumulation model, the Gobas Model, as described below. Because of the many uncertainties inherent in modeling bioaccumulation, EPA has used a weight-of-evidence approach employing three different bioaccumulation models at varying levels of complexity, ranging from empirical to mechanistic.

The following models were developed and calibrated for the Baseline Modeling Report:

HUDTOX - The backbone of the modeling effort is the Upper Hudson River Toxic Chemical Model (HUDTOX). The HUDTOX model covers the Hudson River from Fort Edward to Troy, New York. HUDTOX is a transport and fate model, which is based on the principle of conservation of mass. It balances inputs, outputs and internal sources and sinks for the Upper Hudson River. Mass balances are constructed first for water, then sediment and then PCBs. External inputs of water, sediment and PCBs are specified from field observations. Once external inputs are specified, the internal model and system outputs, can be calibrated against field observations. Outputs of PCB concentrations in water and sediment from HUDTOX are used as inputs for the forecasts of the bioaccumulation models.

Depth of Scour Model (DOSM) - The Depth of Scour Model was developed to provide spatially-refined information on sediment erodibility in response to high-flow events such as a 100-year flood. The DOSM model is a two-dimensional, GIS-based sediment erosion model that was applied to the Thompson Island Pool. It is linked with the output from a hydrodynamic model that predicts the velocity and shear stress (force of the water acting on the sediment surface) during a flood. The model was also used to develop relationships between river flow and cohesive sediment resuspension. These relationships were used in the HUDTOX model for evaluating flow-dependent resuspension.

Bivariate BAF Analysis - The Bivariate BAF (Bioaccumulation Factor) Analysis for fish body burdens looks at the data for sediment and summer average water-column PCB concentrations (two variables or "bivariate") and compares them to measured PCB levels in fish tissue. This allows for the interpretation of the relative importance of water and sediment sources to a particular species of fish, in turn reflecting its feeding behavior. As the BAF calculated from this model does not take into account causal relationships, this analysis has limited predictive capabilities compared with the more mechanistic models, described below.

Empirical Probabilistic Food Chain Model - The Empirical Probabilistic Food Chain Model relies upon feeding relationships to link fish body burdens to PCB exposure concentrations in water and sediments. The model combines the information from available PCB exposure measurements with knowledge about the ecology of different fish species and the relationships among larger fish, smaller fish, and invertebrates in the water column and sediments. The Empirical Probabilistic Food Chain Model provides information on the expected range of uncertainty and variability around average body burden estimates (in contrast to the Bivariate BAF Analysis, which just provides the average body burden estimates).

Gobas Mechanistic Time-Varying Model (FISHPATH and FISHRAND) - As a result of the peer review for the Modeling Approach held in September 1998, it was determined that a time-varying, mechanistic model should be included in the suite of models being used to evaluate the potential for PCB uptake into fish tissue. Consequently, two additional mechanistic models were developed describing the uptake, absorption and elimination of PCBs in fish over time. The models are based on the approach of the peer-reviewed uptake model developed by Gobas (1993 and 1995). This is the same form of the model that was used to develop criteria under the Great Lakes Initiative (USEPA, 1995). Two versions of the model were developed for the Reassessment, a deterministic version (average body burdens) referred to as FISHPATH, and a probabilistic version (the average body burdens including estimates of uncertainty and variability, predominantly variability) referred to as FISHRAND. The predictions of future fish tissue concentrations from FISHRAND will be utilized for estimating potential risk in the Human Health and Ecological Risk Assessments.


The HUDTOX model was calibrated for four different forms of PCBs: total PCBs, Tri+, BZ#4, and BZ#52. Total PCBs represents the sum of all measured PCB congeners and represents the entire PCB mass. Tri+ represents the sum of the trichloro- through decachlorobiphenyl homologue groups. This allows for the comparison of data that was analyzed by congener-specific methods with data analyzed by packed column methods that did not separate the various PCBs as well and did not measure many of the mono- and dichlorobiphenyls. Therefore, use of the operationally defined Tri+ term allowed for a consistent basis for comparison over the entire period for which historical data were available. BZ#4 is a dichloro congener that represents a final product of PCB dechlorination in the sediments. In addition, the physical and chemical properties of BZ#4 are different from the other forms of PCBs (e.g., it is more soluble and has a lower partitioning coefficient), which adds to the rigor of the calibration. BZ#52 is a tetrachlorobiphenyl that was selected as a normalizing parameter for congener patterns based on its presence in Aroclor 1242, the main Aroclor used by General Electric at the Hudson River capacitor plants, and due to its resistance to degradation or dechlorination in the environment.

A long-term hindcasting application was conducted for Tri+ for the period of record, from 1977 to 1997. However, the period from 1977 to 1984 had limited PCB data for estimating external Tri+ loadings. The uncertainty introduced by this limited PCB data required that an additional PCB load be added in order for the model to match sediment concentrations in Thompson Island Pool in 1984 and water column observations downstream. Consequently, the long-term hindcast calibration for Tri+ was actually only conducted for the period from 1984 to 1997.

The period from 1991 to 1997 was the principal focus of the calibration effort because this period was relatively data rich in terms of parameters measured, spatial-temporal coverage and data quality. Applications for this period included all four PCB forms: total PCBs, Tri+, BZ#4, and BZ#52.

The HUDTOX model was successful in representing the hydraulics, solids and PCB dynamics of the Upper Hudson River over the historical period of record. This period was characterized by a significant transition from an early phase of high upstream PCB loads, followed by a long declining phase to present-day conditions with upstream PCB loads now very close to detection limits. Results from the HUDTOX calibration applications were consistent with the magnitudes and trends of the best available data for the historical period.


The models were run for a forecast period of 21 years beginning January 1, 1998. The 21-year time frame was selected because it matched the time frame of the 1977 to 1997 hindcast. All flows, solids loadings and other external forcing functions were the same as those used in the hindcast, with the exception of PCB concentrations at Fort Edward. The initial PCB concentrations for the forecast were the same as the final PCB concentrations from the 1991 to 1997 calibration simulation. Forecast simulations were run for two different assumptions for PCB loadings at the upstream boundary at Fort Edward: first, water column PCB concentrations were held constant at a level equal to the annual average PCB concentration that was observed in 1997 (9.9 ng/l); and second, water column PCB concentrations were held constant at zero. Note that these simulations assume that there will be no future load increases from any upstream sources. In particular, it was assumed that during the forecasts PCB migration from the GE Hudson Falls Plant site would not increase and that there would not be any type of event similar to the releases that occurred with the alleged partial failure of the Allen Mill gate structure in 1991. Based on the expectation that the PCB load from the GE Hudson Falls Plant site would decrease in the future due to the implementation of remedial measures there, these forecasting simulations were designed to bound the estimates of system responses.

Appropriate target levels for fish body burdens have not yet been established. In the Feasibility Study, site-specific target levels to be protective of human health and the environment will be developed from the risk assessments. However, it is beneficial at this time to compare forecasted fish body burden levels against certain available criteria as a matter of perspective. These include: the 2 ppm wet weight Action Level used by the Food and Drug Administration (FDA) for regulating fish in commerce, and the Great Lakes Sport Fish Advisory Task Force values of 1.1 ppm wet weight for consumption of six fish meals per year, and 0.2 ppm wet weight for consumption of one fish meal per month. Again, these are not endorsements of these values for decision making, and appropriate values will be developed in the Feasibility Study for the site.

Forecasts using the mechanistic Gobas model, FISHRAND, were run only under the constant upstream boundary condition because predicted sediment and water exposure concentrations from HUDTOX were virtually the same for the constant and zero upstream boundary conditions, which would result in virtually the same body burden predictions. Species modeled were largemouth bass, brown bullhead, yellow perch, white perch and pumpkinseed. The reported time period for achieving target values for fish body burdens may extend beyond the 21-year forecast period after consideration of the uncertainty around the best estimated values.


The primary objective in the modeling effort is to construct a scientifically credible tool to help in the understanding of PCB transport and fate and bioaccumulation in the Upper Hudson River, and to use that tool for making forecasts of what will happen in the future. As such, one of the major findings was that it was possible to construct a suite of models that generate output that matches the observed data reasonably well. Subsequent to this report, the model predictions can be used to evaluate ecological and human health risks and to assess the time it takes for the river to recover under various remedial scenarios.

There are numerous general observations about the river that are apparent from the mass balance exercises. Some important observations that impact USEPA's understanding of the system include: the tributaries along the length of the river contribute the vast bulk of the solids load carried by the system; the river is net depositional in the Thompson Island Pool and apparently also in the downstream reaches; and, the models indicate a gradual decline in the mass transport of PCBs down river over time.

Beyond the general observations above, the development of the models and the analysis of model outputs have provided USEPA with the following findings regarding PCBs in the Upper Hudson River:

1.  The future projection for PCB concentrations in the water column is controlled by inputs from the sediment. Although the constant upstream PCB load in the forecast simulations contributes to the PCB concentration in the water column, the shape of the response curve is set by the sediment-to-water PCB fluxes.

2.  A 100-year peak flow event would not be expected to have substantial impacts on the recovery rate of the Upper Hudson River.

3.  Although there has been net deposition of sediment in the Thompson Island Pool (as well as the entire upper Hudson), there have been losses of PCBs from the sediment. In other words, net deposition does not mean that PCBs will be unavailable to the water column. For example, from 1984 to 1994 (the same time frame analyzed in the Low Resolution Sediment Coring Report) the model estimated that 2000 kg of Tri+ were lost from the Thompson Island Pool sediment inventory, while at the same time 1.6 cm of net sediment deposition occurred on a poolwide basis.

4.  There is a contribution of PCBs from the sediment that is not dependent on the flow of the river. Some of the processes that may cause non-flow dependent resuspension are: wind driven dispersion, bioturbation by benthic organisms, bioturbation by demersal fish, mechanical scour by propwash, boats and floating debris, and uprooting of macrophytes by flow, wind or biological action. Such a non-flow dependent load is important because the model calibration suggests that approximately 80 percent of the total PCB transport down the river from 1991 to 1997 took place during low-flow periods.

5.  Forecasts for the FISHRAND model suggest that largemouth bass will achieve 2.0 ppm on an average wet weight basis between 2008 and 2014, with the best estimate of 2011 for river mile 189 (within the Thompson Island Pool), and between 2011 and 2019 (best estimate 2015) for river mile 168 (Stillwater) under constant upstream boundary conditions. Largemouth bass average values will not achieve target levels of 1.1 ppm or 0.2 ppm within the 21-year forecast period at these locations. In addition, the 95th percentile value (a statistically important value that is frequently used in evaluating a high-end risk and/or as part of the evaluation of uncertainty around the range of predicted values) will not achieve any of the target levels in the forecast period. Note that the target levels are for comparison purposes only, and that appropriate levels will be determined in the Feasibility Study.

6.  Forecasts suggest that for river mile 189, average values for yellow perch will achieve 2.0 ppm between 2007 and 2014 (best estimate 2010), and 1.1 ppm between 2015 and 2021. 95th percentile values would not reach any of the targets within the forecast period. Average yellow perch values will achieve 2.0 ppm between 2008 and 2014 (best estimate 2011) for river mile168, but the lower target values and the 95th percentile values will be not reached within the forecast period.

7.  For brown bullhead, the average fish body burden is forecasted to reach 2.0 ppm between 2014 and 2020 (best estimate 2017) at river mile 168. Within the 21-year forecast period, no other target levels will be achieved for average brown bullhead at river mile 168, and none of the target levels are achieved at river mile 189.

8.  At river miles 157 and 154, forecasts for all species modeled achieved the FDA action level of 2 ppm by 2021, even at the 95th percentile value.

9.  For all locations and species modeled, predicted average body burdens did not fall below 0.5 ppm within the 21-year forecast period.


The principal processes that control contemporary PCB dynamics in the Upper Hudson River are hydraulics, external solids load, sediment-to-water fluxes, water-to-air fluxes and PCB fate in the bedded sediments. It appears that the river is currently on the tail of a long PCB washout curve controlled largely by the rate at which PCBs are being reduced in the upper mixed sediment layer. Consequently, forecasts of system responses depend on an accurate representation of processes controlling solids dynamics and PCB interactions across the sediment-water interface.

The forecasting results suggest that the water column and sediments of the Upper Hudson River will not have reached steady-state by 2018 (the end of the forecast period). At that time, even with constant upstream PCB loads, water concentrations still show a declining trend, suggesting that the sediments continue to be a source of PCBs to the system.

In their present forms, the models are useful tools for providing information on PCB exposure concentrations for the Human Health and Ecological Risk Assessments. Additional modeling efforts will be conducted to fine tune the model for predicting the time it takes for the system to recover. The results of these additional modeling efforts will be made available as part of the Responsiveness Summary for this report.


Jump to main content.