臺灣地形陡峭，高強度降雨時期所觸發之土壤沖蝕、崩塌及土石流等自然災害頻繁，故以水文模式結合無限邊坡模式進行區域性斜坡穩定模擬已成為近十多年來的趨勢，但前人研究多基於使用靜態或準動態指標代表土壤含水量進行模擬，或僅考慮非連續性短時距之土壤含水量變化，故連續性動態模擬仍未十分普及。且有鑒於一般若非透過現場或實驗室之水工試驗，將難以獲得詳細土壤凝聚力與相關參數。為彌補此不足，本研究以蓮華池四號及五號集水區為例，透過土壤水分演算模式 (Soil Moisture Routing Model, SMR) 模擬土壤含水量隨降雨之動態變化，後將分析結果代入可計算邊坡穩定程度之無限邊坡模式，搭配蒙地卡羅模擬進行小尺度動態崩塌潛勢運算，並利用集水區內均無崩塌之特性，對照集水區實際崩塌情況反演土壤與植生根部凝聚力參數。研究結果顯示，集水區內具較低安全係數之網格，會隨不同季節降雨與土壤含水量而於面積上呈現動態增減，坡度、土壤凝聚力與植物根部凝聚力則影響安全係數值與其隨降雨變化之幅度。另透過參數反演法得知，在假設土體單位重為18 kN/m³之情況下，若欲維持蓮華池集水區深度1m 範圍內之邊坡穩定，則土壤與植生根部凝聚力之總和至少需10.8 kN/m³。而透過此崩塌潛勢動態模擬流程將有助於分析降雨變化對坡地穩定之影響程度，並可協助決策管理者選擇最適合之土地利用型態。

Landslides and debris flow frequently triggered by heavy rainfall in Taiwan are due to steep terrain, fragile geology, and soil erosions. Under this condition, the analysis composed of hydrological model and the infinite slope model has become a trend in the recent decades.
Consequently, previous studies often used static or quasi-dynamic indices as soil moisture in simulation process, or only concerned about un
continuous short-term change of soil moisture variety. Therefore,
continuous dynamic simulation has not been commonly implemented. As it is very difficult to obtain detailed and relevant soil cohesion and relative coefficient without on-site measurement or laboratory test, this study took Lien-Hua-Chih Watershed for example, simulated soil moisture varied with rainfall by using Soil Moisture Routing Model, and then calculated dynamic landslides susceptibility with infinite slope model and Monte Carlo simulation. Additionally, we also deduced the minimum root
cohesion and soil cohesion of maintaining slope stability by utilizing the character of no landslide within the watershed. Results showed that the proportion of susceptible and non-susceptible cells changed spatially and temporally with effective precipitation and soil moisture. In addition, the factor of safety is strongly influenced by slope, root
cohesion and soil cohesion. For example, if the sum of soil cohesion and root cohesion are higher than 10.8 kN/m³(asuumed soil unit weight= 18 kN/m³), the 1m depth slope of Lien-Hua-Chih watershed will maintain stable. Such dynamic model-based processes may be used to present the great effects upon landslide susceptibility of rainfall intensity, and help decision-makers choose the most suitable land use.