A Bayesian Framework for Cost Effective Management of Sediment Reduction in the Minnesota River Basin

Water quality impairments remain a pressing concern in the United States. Selecting appropriate management actions (i.e., best management practices) to improve water quality involves tradeoffs between cost and effectiveness, both of which are prone to uncertainty. In addition, significant uncertainties exist in the scientific understanding of the natural system. To address these concerns, we have developed a framework to identify the optimal set of actions to reduce turbidity and sedimentation in the Minnesota River Basin, while explicitly incorporating uncertainty. The framework combines Bayesian inference with multiobjective programming models to select the optimal combination of research actions, which improve our understanding of the natural system, and management actions, which reduce sediment contributions to the river basin. We also evaluate the value of information of each research action, to identify where further study is warranted. Our results communicate the general nature of the tradeoffs and complex interactions among sediment loadings, best management practices, and the possible research projects. More important, our results indicate that explicit consideration of uncertainty can lead to improved selection of management actions that most accurately address the complexity of the problem.