A Decision Analysis Approach to TMDL Implementation Decisions: Mercury TMDLS in the San Francisco Bay Area

Abstract

This paper describes a decision analysis approach to TMDL implementation decisions for mercury using a hypothetical mine-impacted tributary in the San Francisco Bay as an example. Decision analysis is a theoretically sound approach for making significant decisions under uncertainty (see, e.g., Howard, 1968;Keeney and Raiffa, 1976;Clemen, 1996;Merkhofer, 1999). The Bayesian probabilistic nature of decision analysis makes it ideal for integrating diverse information, including the results from scientific and engineering models, cost and benefit models, empirical data, and expert judgment. One significant advantage of a decision analysis approach is its explicit separation of a decision problem into alternatives, information, and preferences. This, in theory, allows decision makers and stakeholders to separate "what we know" from "what we want". It is hypothesized that a more explicit separation of information and values/preferences will focus the debate. While traditional decision analysis assumes a single rational decision maker (where "single" may also denote a group that agrees on information and preferences), it can be extended to multiple decision maker situations in a variety of ways. Evaluating various extensions of decision analysis in a TMDL implementation stakeholder context is one of the primary goals of this on-going study. It is hypothesized that, in general, decision analysis provides a helpful decision framework for a TMDL implementation planning/stakeholder process in many circumstances.

Incorporating Uncertainty in Watershed Management Decision-Making: A Mercury TMDL Case Study

“…In the context of public environmental decision making, these could be cast as: 1) decision framing/strategy generation; 2) information modeling/synthesis/forecasting; and 3) multiattribute utility analysis, negotiation among interest groups, or other methods of eliciting and representing preferences. Each of these aspects of decision are described in detail elsewhere (e.g., Merkhofer, 1999;Labiosa et al, 2003, and references cited within). The goal of decision analysis is to create decision clarity in a complex decision problem.…”

“…Since budgets are limited, the ability to contract outside expertise is also limited. Collectively, these issues point to a need for a decision framework that takes into consideration the very large uncertainties involved and the resource constraints of the State agencies tasked with TMDL development and implementation planning (Labiosa et al, 2003;NRC, 2001;Ruffolo, 1999).…”

“…While this example is monetized, this is not necessary to fully apply decision analysis. For examples of using a decision analytical approach with an incomplete or missing value model (i.e., no consensus on values), see Labiosa et al (2003). As can be seen from the figure, the "Mine Mitigation" strategy has lower value than "Status Quo", which means that doing nothing would be better than implementing the "Mine Mitigation" strategy, given current information and understanding.…”

“…As an example of the value of collecting new information, it should be noted that collecting more information and reducing material uncertainties may result in the "Mine Mitigation" be more valuable than "Status Quo". Formal value of information analyses based on sensitivity analysis can also be performed in a decision analytical framework to prioritize information collection activities (see, e.g., (Howard et al, 1972); Labiosa et al, 2003).…”

Offset-Based Decision Support Models for Mitigating Mercury Sources in the Cache Creek Watershed, North Central California

“…Because the uncertainties involved in estimating HgT loading and in predicting the impacts of load-reduction projects are significant, an approach that explicitly treats uncertainty is warranted for decision support (Labiosa, 2003).…”