MODELING AND RISK ANALYSIS OF NONPOINT-SOURCE POLLUTION CAUSED BY ATRAZINE USING SWAT

摘要

The SWAT (Soil Water Assessment Tool) model was calibrated and validated to evaluate its performance to predict atrazine loads in streams for the period 1996-2004 at eleven sampling sites in the St. Joseph River watershed in northeast Indiana. This watershed encompasses 280,000 ha, and 60% the area is in agricultural crops of corn and soybeans. Daily streamflow calibration and validation were completed before starting pesticide calibration. During the validation period, Nash-Sutcliffe values varied from 0.33 to 0.60 for daily streamflow and between 0.64 and 0.74 for monthly streamflow. The estimation of the timing of atrazine application was very important in the calibration-validation process, and it proved to be a key input for predicting the amount and timing of pesticide released to streams. Monthly atrazine concentrations were predicted with average R 2 values of 0.60 and 0.49 and average Nash-Sutcliffe coefficients of 0.38 and -0.91 for the calibration and validation periods, respectively. The total mass of atrazine released by the whole basin between 2000 and 2003, for the period April to September, was closely predicted by the model. The observed average amount of atrazine released during the four seasons was 1002.1 kg/season, and SWAT predicted 950.1 kg/season. Risk analysis was performed based on the outputs generated by the model by computation of exceedance probability curves and thematic and probability maps. The model was suitable to estimate levels of atrazine released to streams in rural watersheds and to conduct NPS pollution risk analysis at a basin scale to evaluate long-term effects of management practices and environmental changes