由於氣候變遷導致極端氣候，過多降雨導致地下水位上升，進而誘發崩塌災害，若能及早預知地下水是否達到崩塌臨界值，應能發揮早期預警機制，減少生命財產損失。 本研究發展深層崩塌之地下水位預測模式，首先建立無限邊坡理論與試驗驗證，推得深層崩塌臨界地下水位及位置，再採用線性水庫，以降雨量及地下水位觀測資料，預估下一小時地下水位。 並以嘉義縣中心崙崩塌地，收集地文、降雨及地下水位，先以無限邊坡理論，計算理論臨界地下水位，再以線性水庫模式，找出降雨與地下水位關係較佳之監測站，測試後於選擇模擬較佳之二處測站，進行線性水庫及類神經網路模式比較，將可提供主管機關做為早期大規模崩塌預警之依據。 Due to extreme weather caused by climate change, excessive rainfall rises the groundwater level, and induces landslide. If the prediction of groundwater level can be realized, an early warning mechanism may reduce the loss of life and property. This research develops a groundwater level prediction model for deep-seated landslide. First, the infinite slope theory and experimental verification are established, and the critical groundwater level and location of deep-seated landslide are deduced. Then, a linear reservoir model is used to predict the groundwater level in the next hour based on the observation data of rainfall and groundwater level. The geology, rainfall and groundwater level are collected from the landslide in the Chiayi County. First, the infinite slope theory is used to calculate the theoretical critical groundwater level, and then the linear reservoir model is used to find out the monitoring station with a better relationship between rainfall and groundwater level. After selecting the two stations with the best simulation results, a comparison of the linear reservoir and the neural network model will be carried out, which will provide the competent authority as a basis for early warning of large-scale collapse.
