Optimization and Decision Heuristics for Chesapeake Bay Nutrient Reduction Strategies

Abstract

Regional policies to achieve water quality goals assign a unique pollution control technology to every pollution source in a watershed, thereby defining a watershed strategy. For watersheds with even a modest number of pollution sources and control alternatives, the decision problem has combinatorial complexity. The perception of complexity—manifested in innumerable feasible watershed strategies—commonly induces the use of simplifying decision heuristics and ad hoc decision rules that reduce decision complexity by limiting the choice set to a “manageable” number of alternatives. In problems with large complex choice sets, these decision heuristics simplify decision making by excluding the vast majority of feasible alternatives a priori. In contrast, watershed-scale optimization enables decision makers to consider all feasible alternatives implicitly, exploiting rather than restricting the complexity of the feasible choice set. This contrast is illustrated using mixed-integer linear programming to identify interstate watershed strategies that achieve Chesapeake Bay nutrient reduction goals for the Potomac River Basin. The use of optimization in collaborative decision making helped refine and capture decision makers’ underlying values and preferences in policy-relevant constraints reflecting equity and political feasibility. Optimization formulations incorporating these constraints identified more effective and desirable management alternatives that would not otherwise have been considered using familiar decision heuristics and traditional comparisons among a limited number of ad hoc scenarios. Incorporating optimization in collaborative decision making generated superior watershed strategies and eased the cognitive limitations on decision making by substituting the computational burden of solving mixed-integer linear programs for decision makers’ cognitive burden of enumerating alternatives and scenarios for environmental systems with combinatorial complexity.