This interim report describes an alternative approach for evaluating the efficacy of using mercury (Hg) offsets to improve water quality. Hg-offset programs may allow dischargers facing higher-pollution control costs to meet their regulatory obligations by making more cost effective pollutant-reduction decisions. Efficient Hg management requires methods to translate that science and economics into a regulatory decision framework.

This report documents the work in progress by the U.S. Geological Survey’s Western Geographic Science Center in collaboration with Stanford University toward developing this decision framework to help managers, regulators, and other stakeholders decide whether offsets can cost effectively meet the Hg total maximum daily load (TMDL) requirements in the Sacramento River watershed. Two key approaches being considered are: (1) a probabilistic approach that explicitly incorporates scientific uncertainty, cost information, and value judgments; and (2) a quantitative approach that captures uncertainty in testing the feasibility of Hg offsets.

Current fate and transport-process models commonly attempt to predict chemical transformations and transport pathways deterministically. However, the physical, chemical, and biologic processes controlling the fate and transport of Hg in aquatic environments are complex and poorly understood. Deterministic models of Hg environmental behavior contain large uncertainties, reflecting this lack of understanding. The uncertainty in these underlying physical processes may produce similarly large uncertainties in the decisionmaking process. However, decisions about control strategies are still being made despite the large uncertainties in current Hg loadings, the relations between total Hg (HgT) loading and methylmercury (MeHg) formation, and the relations between control efforts and Hg content in fish.

The research presented here focuses on an alternative analytical approach to the current use of safety factors and deterministic methods for Hg TMDL decision support, one that is fully compatible with an adaptive management approach. This alternative approach uses empirical data and informed judgment to provide a scientific and technical basis for helping National Pollutant Discharge Elimination System (NPDES) permit holders make management decisions. An Hg-offset system would be an option if a wastewater-treatment plant could not achieve NPDES permit requirements for HgT reduction.

We develop a probabilistic decision-analytical model consisting of three submodels for HgT loading, MeHg, and cost mitigation within a Bayesian network that integrates information of varying rigor and detail into a simple model of a complex system. Hg processes are identified and quantified by using a combination of historical data, statistical models, and expert judgment. Such an integrated approach to uncertainty analysis allows easy updating of prediction and inference when observations of model variables are made. We demonstrate our approach with data from the Cache Creek watershed (a subbasin of the Sacramento River watershed).

The empirical models used to generate the needed probability distributions are based on the same empirical models currently being used by the Central Valley Regional Water Quality Control Cache Creek Hg TMDL working group. The significant difference is that input uncertainty and error are explicitly included in the model and propagated throughout its algorithms. This work demonstrates how to integrate uncertainty into the complex and highly uncertain Hg TMDL decisionmaking process. The various sources of uncertainty are propagated as decision risk that allows decisionmakers to simultaneously consider uncertainties in remediation/implementation costs while attempting to meet environmental/ecologic targets.

We must note that this research is on going. As more data are collected, the HgT and cost-mitigation submodels are updated and the uncertainties may be reduced. Subsequently, the value of using a probabilistic framework for estimating and explicitly stating these uncertainties within a decisionmaking process can be estimated when new data are collected.

Future work includes the design and implementation of a Bayesian network decision support system (BN-DSS) to produce mitigation scenarios for offset-project evaluation in the Cache Creek watershed. The decisionmaker, a wastewater-treatment plant, is expected to evaluate potential Hg-offset programs in terms of changes in HgT load changes, MeHg-production potential, project cost, and other suitability criteria. Subsequently, scenarios can be analyzed by performing sensitivity analyses and ranking environmental and economic uncertainties in terms of the decisionmaker’s preferences and risk choices. Such an analysis allows decisionmakers and stakeholders to explore various scenarios and predict the consequences of different stated preferences over outcomes.