非點源污染模式參數不確定性分析

Abstract

非點源污染控制是水庫環境污染防治工作中甚為重要的一環，當使用非點 源污染模式來協助分析時，其輸入參數之不確定性往往對模擬結果有很大 的影響，也因此可能造成非點源決策分析的錯誤，所以本研究針對非點源 污染模式的參數進行不確定性分析。由於逕流排向、降雨量這兩項參數具 有顯著的不確定性，因此本研究在逕流排向方面除了建立一套修正逕流排 向的方法之外，並就逕流排向的演算法分析其影響 AGNPS非點源污染流佈 模擬結果之程度；降雨量方面，乃結合 Auto-MOUSE 與 AGNPS進行不確定 性分析。本研究以離槽水庫－寶山水庫集水區進行案例研討，發現不同的 逕流排向演算法造成不同的污染物空間分佈情況，影響了分區式總量管制 對象之研判；降雨之不確定性分析，提供污染物總量季節性變動之預估， 可以降低模擬分析錯誤的風險。 本研究並以地理資訊支援系統 (GRASS) 配合模式提供圖形化的分析結果，以利決策者作適當的決策，更可提高決 策分析的效率。 Non-point source pollution control(NPSPC) for reservoir has become a major environmental protection mission in Taiwan, ROC. Mathematical models are applied for evaluation of NPSPC related tasks. Model parameter uncertainty, however, introduces significant effect on modeling results and thus the decision made on the basis of these results may not be appropriate. This research was therefore initiated for analyzing the parameter uncertainty for exploring an improved modeling procedure. Drainage pattern generated from DEM data and rainfall intensity were the two parameters studied for their uncertainties in this research. A program was developed to resolve conflicting directions in a DEM generated drainage pattern for use by a grid based nonpoint source pollution model, AGNPS. Auto-MOUSE, a Monte-Carlo analysis package, was integrated with AGNPS for assessing the rainfall uncertainty. A case study for the watershed of the Po-San off-channel reservoir in Hsinchu County was implemented. Significant spatial variation of pollution distribution simulated by using different drainage pattern generating methods was observed. The effect of rainfall randomness on seasonal and spatial loading distribution was assessed and computed based on a Monte- Carlo simulation. Graphical presentations of research results and spatial data analysis were implemented by a geographical information system, Grass. It is expected that the quality of decision making can be effectively improved with the proposed parameter uncertainty analysis procedure.