Non-point source water quality management under input information uncertainty

Abstract

Non-point source pollution in the form of erosion can be controlled with watershed management schemes that identify the optimum set of farming practices for given conditions of topography, climate, crop yield, crop prices and allowable sediment load. However, since input conditions may be difficult to estimate or may vary naturally, it is important to develop approaches for determining good farming practices while accounting for the uncertainty of the input conditions. This work identifies: (1) the sensitivity of different watershed management policies to uncertain input information; (2) the input information that is significant for modelling and managing erosion and sedimentation; and (3) the farming practices that are least sensitive to uncertainty in the input information. Monte Carlo Simulation, and Modified Generalized Sensitivity and Regret Analyses are applied in combination to address these issues and this approach is demonstrated for the control of erosion and sedimentation in the Highland Silver Lake Watershed in Illinois. Two different watershed management policies, the Erosion Standard and Erosion Tax Policies, and a benchmark representing a least cost bound, the Least Opportunity Cost, are examined. Results show that for the Highland Silver Lake Watershed the Least Opportunity Cost approach yields the least variable solution with respect to opportunity cost and that the Erosion Standard Policy produces the most variable «optimal» opportunity cost values under different watershed conditions, crop yields and crop prices. The rainfall erosivity factor and the prices of corn and soybeans are the most important parameters in the linked sediment delivery and economic model used in this work and the importance of the crop yield parameters depends on the management policy chosen. The Regret Analysis shows that the most often selected «optimal» sets of farming practices under many scenarios of uncertain input data and parameters vary depending on how regret is expressed, but in general, only a small selection of farming practices are identified to be relatively less sensitive to uncertain inputs.