BMP Evaluation for Nutrient Control in a Subtropical Reservoir Watershed Using SWAT Model

Abstract

The eutrophication problem that results from the inappropriate management in upstream watershed is prevailing among most reservoirs in Taiwan. As a result, implementation of appropriate Best Management Practices (BMPs) in upland watershed has proved to be the most effective measure to reduce or alleviate the nutrient loadings and the risk of eutrophication. This study aims to employ the uncertainty analysis and risk probability on SWAT (Soil and Water Assessment Tool) simulation for assessing the effectiveness of various BMP scenarios implemented in Feitsui reservoir watershed. The SWAT model was applied to assess the water quality impacts due to the implementation of various BMP scenarios. Five BMPs were selected and implemented on each agricultural land for overall evaluation of pollution reduction efficiency both at field and watershed scales. Finally, the effects of uncertainty in model input parameters on pollution reduction efficiency of BMPs are evaluated using constrained Monte Carlo simulation based on Latin hypercube sampling (LHS) and Orthogonal sampling (OS) schemes. Moreover, the Cumulative Distribution Functions (CDF) of average annual TP and chlorophyll a concentrations were created for further assessment on eutrophication risk. The simulation results show that the reduction of eutrophication risk approaches 13% (from 15 to 2%) for Caelson Trophic Satus Index (CTSI) methods based on implementation of 5 BMPs (a combined BMP scenario). Overall, this study concludes that incooperating uncertainty analysis in modeling will provide model output with specific probability, which not only assists the decision-maker in developing BMP and TMDL plans but also assesses the reliability of NPS pollution control strategies and risk of water quality degradation. Index Terms—Best Management Practices (BMPs), nonpoint source pollution, nutrient, SWAT model, uncertainty analysis.