Application of Multicriteria Decision Analysis with A Priori Knowledge to Identify Optimal Nonpoint Source Pollution Control Plans

Abstract

Control of agricultural nonpoint sources of pollution is achievable through implementation of conservation practices at the farm or field level. There are several approaches to achieve a targeted implementation of conservation practices at the watershed scale. Recent studies have shown that optimization methods hold great promise for optimal allocation of nonpoint source pollution control measures. However, the use of optimization is a computationally intensive task and ultimately depends upon availability of automated optimization tools and expertise to analyze the results. In this study, a novel multicriteria decision analysis (MCDA) framework is proposed to identify a near-optimal suite of conservation practices at the watershed scale using a priori knowledge about the system. The proposed framework requires: (1) selecting a set of criteria, depending upon the objectives of the study, that should be considered in ranking the alternatives; (2) constructing an evaluation matrix; and (3) using a computational MCDA method to aggregate the scores based upon the various criteria and rank the alternatives. The framework was used to identify optimal placement of four types of conservation practices for nutrient and pesticide load control at minimum cost in the Eagle Creek Watershed, Indiana, United States. Results were compared with optimal solutions obtained from an optimization framework coupled with the soil and water assessment tool (SWAT). The results of this study showed that the proposed framework can be an effective and efficient approach in identifying near-optimal solutions for nonpoint source pollution control. The MCDA framework outperformed the optimization method by identifying similar solutions with more diversity without any need for iterative search algorithms. For highly complex problems or for a poorly established evaluation matrix, the MCDA framework may fail to identify near-optimal solutions; however, the results can effectively serve as a good initial population in a hybrid MCDA and optimization framework. A hybrid framework substantially improved efficiency of the search algorithm, optimality of the Pareto-front, and diversity of the solutions. This study also highlighted importance of the defining proper decision variables and accurate scoring of the conservation practices for successful implementation of conservation plans at the watershed scale.