集水區水質監測站址優選模式及線上地理資訊系統

Abstract

建立適當的監測站網可有效回溯集水區污染源，提升污染控制效能。由於過往相關研究所提出的模擬退火演算法不能確保求得最佳解及缺乏穩定性之缺點，本研究因而以均化各監測站點涵蓋範圍內之決策考量因子為目標，建立兩個線性整數規劃優選模式，決策考量因子包括河段數、河段長度、面積、總磷、總氮與沉積物六個因子，成本均化模式考量各河段或子集水區應屬何測站，求取各測站之涵蓋範圍，覆蓋消去均化模式則以測站上游總涵蓋範圍減去其上游已設測站之涵蓋範圍，求出測站之有效涵蓋範圍。並以德基水庫集水區為研究案例，分別以兩均化模式與模擬退火法進行監測站網優選，所得結果顯示本研究新建模式所決策出站網之污染回溯成本函數較模擬退火法所得結果更佳，且可穩定求得全域最佳解。另為了改善站網決策分析的效率，本研究進一步建立一個決策支援系統，結合三個優選模式與網路地理資訊系統 (Web-GIS)，讓決策者可透過Web-GIS之空間分析與適當介面，即時比較及分析不同優選模式之決策結果。

Pollution distribution can be effectively tracked by a well-designed water quality monitoring network. In a previous study, the simulated annealing (SA) algorithm was applied to design a water quality monitoring station network. However, the SA method generally could not obtain the global optimum. Therefore, in this study, two linear optimization models with objectives minimizing the deviation of the distribution of cost factors are developed. Six cost factors are considered, including the number of reaches, the bank length, the size, the phosphorus load, the nitrogen load and the sediment load in the area covered by each monitoring station. For the uniform cost (UC) model, the coverage of each station is individually determined, while the coverage elimination uniform cost (CEUC) model determines the coverage of a station by eliminating the area that has been covered by any upstream station. The study area is the Der-chi reservoir watershed located in central Taiwan. Water quality monitoring station locations determined by the UC and CEUC models are obviously better than those obtained by the SA method. The both models produce similar results, but the CEUC model requires significantly less computational time to solve. A web-based decision support system (DSS) has been also developed to integrate the optimization models and a Web-GIS. The DSS provides friendly interface for using the models and generate graphical charts for presenting various spatial results. The system is intended to facilitate the decision analysis for water quality monitoring station siting.