中文摘要 台灣土石流災害嚴重,目前預警系統無法達到好的效果,致使土石流的發生對人民的生命與安全威脅甚劇。然而土石流發生並非如同地震般難以察覺和預測,故有關土石流預測值得繼續研究。 以往學者有關土石流發生臨界降雨線僅考慮降雨量和降雨強度以統計方法作線性迴歸分析,本文主旨在建立土石流發生臨界線之模糊迴歸曲線與有效降雨強度及有效累積雨量的模糊隸屬關係,由土石流發生臨界線之模糊迴歸曲線,找出土石流發生和降雨之模糊隸屬函數,進而建立土石流危險程度之評判分析模式。另以模糊理論做分析的基礎,建立土石流成因之統計資料,作為預警決策分析之參考;選取各集水區形狀因子、溪床坡度、地質特性作為土石流發生之潛在因子,並以土石流災區之降雨強度及降雨量等水文資料為激發因子和以臨界線之模糊迴歸分析、模糊隸屬度統計各因子與土石流發生之相關性。 最後以模糊綜合評判及邏輯推論作為評判方法,分析各因子與土石流發生的綜合程度,判斷其危害等級。實例演算各因子對土石流發生的綜合程度,發現本文評判分析結果之危險等級可作為預警決策的參考。
Abstract Disasters caused by debris flow are very serious in Taiwan. Advanced warning system can,t reach good result so far. The occurrence of debris flow is not so difficult to predict as earthquake, so research in advanced warning system of debris flow has to be reinforce. The scholar relate when debris flow will occur critical rainfall line only thought about rainfall and rainfall intensity by statistic method analysis in linear regression before. The purpose of this paper is based on three main ideas. First, how to establish the curve in fuzzy regression of critical line when debris flow occurs and its relationship with effective rainfall intensity and accumulated rainfall. Second, according to the curve mentioned above try to find out the fuzzy statistics function, and then build a decision-analysis model for the risky grade of debris flow. Third, use fuzzy statistics for the weighed analysis and set up the database of the factors controlling the debris in order to establish the advanced warning system. To achieve this system, there are two types of factors need to be probed. One is a latency factor that contain form factor of watershed, valley slope geology. The other is a stimulated factor that contains rainfall intensity and accumulated rainfall. Analyze the stimulated factors and the critical line with fuzzy regression to relate each factors in fuzzy statistics with the occurrence of debris flow. Finally, the fuzzy inference and fuzzy decision making method were used to analyze the relationship among factors and estimate the risky grade of debris flow. The risky grade of debris flow takes in the example. It is found that this new decision analysis method is a useful reference material for the advanced warning system. Keywords: Debris Flow, Membership Function, Fuzzy Regressions, Fuzzy Decision Making Method, Fuzzy Inference.