- 學位論文
結合岩體評分法與網格不穩定力平衡模式應用於廣域山崩行為之模擬-以溪頭為例
The Simulation of Landslide Behavior in Shitou Area Using GSI Method and GIS Technology
摘要
定量分析邊坡穩定性的方法很多，其中以力學觀點，多採取安全係數評估；分析過程中，若能獲取合理的岩體材料參數，將有助於提升分析成果的精準度。由於現地岩體受風化與水文影響，將導致強度弱化，本研究嘗試建立風化岩體強度參數的評估方法，以GIS網格式的資料結構作為運算平台，應用於廣域邊坡漸進式破壞行為的模擬。 廣域岩體的強度參數複雜多變，前人多仰賴反算分析求得，然而反算過程中，缺少合乎力學邏輯的轉換準則，且反算成果無法反應現地岩體強度，而備受爭議。因此，本研究的第一步，採岩性分區取代地層分區，加入舊山崩、侵蝕溝與斷層等不良地質區，建構基本模型，以滿足現地狀況。接著，結合岩體評分法與現勘成果，蒐集相關試驗資料，訂定風化指標，評估岩體強度受風化與吸水影響，導致強度弱化比例的經驗法則，以求得廣域岩體的強度參數；最後代入無限邊坡理論與前人建立的網格不穩定力平衡模式，期望有效提升模擬邊坡破壞的準確性。 研究區域選取具有詳細災害記錄、5m*5m高精度DTM圖層、以及岩性地質圖之溪頭南部，並採2000年桃芝颱風作為本研究的災害事件。 研究結果顯示：本研究採用的風化岩體強度參數之評估方法，相較於前人的反算方法而言，有更好的分析成果。其最大優勢在於：結合現地勘查與試驗資料，可有效、快速地定量不同風化程度之下，岩體強度參數弱化的比例，提供數值模擬使用。此外，本研究亦針對豪雨事件，配合雨量離散化的方式，探討地下水位與岩體強度隨著累積雨量變動，而導致邊坡開始發生破壞與延展的區位和產狀，期望對於未來進階研究具有參考價值。
並列摘要
Chichi earthquake and Typhoon Toraji has caused severe landslides in Shitou area of Nantou, central Taiwan. Using the great amount of hazard data in Shitou area, we try to create the model reasonably and effectively simulating the landslide behavior in Shitou area. In this study, based on infinite slope model for slope-stability analysis, which assumed the slip surface parallel to the ground surface, a Geographic Information System (GIS) grid-based model is developed. The identification of back-calculation for the equivalent in-situ rock mass parameters of wide area is performed first by making use of geological strength index (GSI) and weakening the strength of rock mass resided in old landslide area. In addition, considering the rainfall-induced factor, we add a new parameter “m”, which is named ratio of groundwater depth and to be the function of accumulated rainfall, infiltration rate, porosity and soil depth. Combining the GIS deterministic grid-based model and GSI method, the safety factor between different grids will interact with each other. The balance process of the unstable forces between grids will keep going on until the safety factor is up to 1, which can decide the situation of grids - failure or non-failure but with potential of driving. The driving grids will move to the direction with maximum slope gradient and decide the source condition of the next flowing process, which keeps moment conversation in the whole process. The simulation process of landslide will end up by means of comparison between simulation outcomes and natural after-hazard sliding distribution. The results show that it can successfully simulate landslide migration behavior in Shitou area. In the future, the integrated GIS model on establishing the database system would offer the helpful reference for further research, hazard prediction and prevention works in Shitou area.