You are here:
EVALUATING THE REGIONAL PREDICTIVE CAPACITY OF A PROCESS-BASED MERCURY EXPOSURE MODEL (R-MCM) FOR LAKES ACROSS VERMONT AND NEW HAMPSHIRE, USA
Citation:
KNIGHTES, C. D. EVALUATING THE REGIONAL PREDICTIVE CAPACITY OF A PROCESS-BASED MERCURY EXPOSURE MODEL (R-MCM) FOR LAKES ACROSS VERMONT AND NEW HAMPSHIRE, USA. Presented at Society of Environmental Toxicology and Chemistry Annual Meeting, Montreal, QC, CANADA, November 05 - 09, 2006.
Impact/Purpose:
The objective of this task is to develop, support and transfer a wide variety of tools and mathematical models that can be used to support watershed and water quality protection programs in support of OW, OSWER, and the Regions.
Description:
Regulatory agencies are confronted with a daunting task of developing fish consumption advisories for a large number of lakes and rivers with little resources. A feasible mechanism to develop region-wide fish advisories is by using a process-based mathematical model. One model of this type is the Regional Mercury Cycling Model (R-MCM), which has been specifically designed to model a series of lakes for a given region. Using the Regional Environmental Monitoring Assessment Program (REMAP) data set, R-MCM was applied to 91 lakes across Vermont and New Hampshire using a series of parameter refinement tiers. Using the simplest application through site-specific parameterization, visual analyses and statistical metrics were used to evaluate the trends in simulated results versus observed mercury concentrations. R-MCM was found to generally under-predict HgT and MeHg water column concentrations, while over-predicting sediment MeHg concentrations. Through a series of progressive parameter refinements, potential improvements in model prediction were evaluated as a function of the different model parameters. Separation of observed and predicted data by lake characteristics identified some patterns of bias and fit, but it was difficult to make any solid conclusions regarding model performance versus lake type. Default level input parameterization produced the largest amount of scatter in the predicted data points. By using site-specific values for the default level characteristics, scatter was reduced but model performance was not generally improved. This analysis demonstrates that R-MCM and the general state of mercury science and modeling are not at a point where fish advisories can be determined using a regional predictive model a priori. However, this work suggests that need for additional research on the transport and transformation of mercury to account for higher total mercury and methylmercury concentrations in these water bodies.