Nonpoint-Source Pollution Reduction for an Iowa Watershed: An Application of Evolutionary Algorithms

We apply an integrated simulation-optimization framework to search for cost-efficient mix and location of agricultural conservation practices in a typical agricultural watershed for two types of nitrogen reduction targets: control of mean annual nitrogen loadings, and a "safety-first" type constraint, insisting that nitrogen targets be met in every weather realization (weather-resilient solutions). Evolutionary algorithms are developed for each of the appropriate water quality targets. Our approach allows for the derivation of a watershed-level total and marginal nitrogen abatement cost curve. Controlling for the probability of meeting water quality targets (looking for weather-resilient solutions) is found to be significantly more costly than controlling the average nitrogen loadings. Both types of solutions are assessed for robustness with respect to weather uncertainty: solutions selected to reduce average loadings do well under weather uncertainty, while the robustness of solutions selected to be resilient decreases with the stringency of the water quality goal. © 2010 Canadian Agricultural Economics Society.