近年來台灣地下水質監測資料顯示，濁水溪沖積扇地區有嚴重的硝酸鹽氮污染問題，相關污染主要來自於農業區大規模的肥料，經地表補注水進入地下含水層。本研究藉由DRASTIC模式及GIS建立濁水溪沖積扇的污染潛勢分佈圖，透過判別分析修正DRASTIC模式中參數的原始權重，期藉此達到更佳的預測準確率。本研究結果顯示，飽和含水層及通氣層介質兩項參數因子具顯著正向貢獻，而水位深度的影響則歸類為負貢獻較顯著的因子。修正後的預測模式已將總判釋準確率，由原始模式的68%提高至修正後的86%，相關指標值也由原本的0.1達到0.3以上，ROC的A值更都有0.9以上的優異預測能力，結果皆顯示修正後的權重預測正確性提升顯著，並依據污染門檻值劃定了高風險污染區，同時針對研究區域重新繪製了污染潛勢分佈圖。研究結果能提供政府決策者及民眾更明確的高風險污染分佈範圍，相信在規劃綜合水資源應用計畫上能有更進ㄧ步幫助。 Monitoring data of groundwater quality reveals that many aquifers in natural recharging zones of Taiwan present serious nitrate-N pollution due to agricultural activities. This study uses DRASTIC model and GIS to assess the contamination potential of nitrate-N in the Choushui River alluvial fan. Additionally, this study also develops a predicted model which adopts discriminant analysis to modify DRASTIC model. The model can substantially improve the prediction and revise some drawbacks in the original DRASTIC model used to predict. The study results reveal that the model presents a higher correct ratio of prediction than DRASTIC model, by 68% to 86%. Thus, this model is suggested to map potential pollution areas using GIS and to delineate high-risk contaminated areas. The results of sensitivity analysis show that the aquifer media and impact of vadose zone are significantly positively related to nitrate-N pollution, but the depth to water is negatively associated with nitrate-N pollution. The analyzed findings can provide residents with suggestive strategies against nitrate-N pollution in agricultural regions and government administrators with explicit information of Nitrate-N pollution extents when plans of water resources are considered.
