考量降雨及逕流影響之淺層坡地崩塌模擬分析

Abstract

台灣因地質脆弱且颱風豪雨頻繁，每逢豪雨常發生坡地崩塌，造成民眾生命財產損失極大之傷害，為有效避免或降低降雨引致之邊坡崩塌災害，坡地崩塌分析為必要之工作。國內外評估降雨所引發淺層崩塌已能由物理力學模式分析，但是過去研究多利用地下水入滲模式結合邊坡穩定分析，分析單一坡面降雨入滲引致崩塌行為；然而，颱風豪雨期間，相較於坡地內水平側向邊界之地下水位，坡地之地表逕流具有較快的反應時間，所以對於降雨引發之坡地淺崩塌分析，有其必要在固定坡地內水平側向的地下水流邊界條件下，考量地表逕流之影響。因此，本研究為能分析降雨與地表逕流對淺層坡地崩塌之影響，耦合計算一維運動波漫地流與二維理查氏方程式，再根據極限平衡法坡地崩塌分析，建構降雨及逕流引致之淺層坡地崩塌模式。 模式完成後，首先對參數進行敏感度分析，分析結果顯示土壤強度參數影響較大，尤其以非飽和摩擦角度，即基質吸力增加所造成摩擦角度影響最大。進一步探討分析不同水文條件對崩塌之影響，結果顯示，降雨之雨量、延時與雨型明顯影響崩塌是否發生，對崩塌發生深度與崩塌時間有重大之影響。逕流入流大小與延時亦明顯影響崩塌是否發生與崩塌發生時間。在上游逕流入流延遲發生狀況下，不同逕流延遲時間，可能造成降雨過程未發生崩塌，降雨停止後可能因地表逕流持續入流而發生崩塌。此外，逕流入流型態亦影響崩塌是否發生，崩塌發生時間可能因而不同。 應用模式建構降雨門檻值曲線發現，二維模式因考慮側向水流影響，所製作門檻值曲線，具有上下邊界，過去一維模式由於無法考量側向水流影響，因此所製作之門檻值曲線上邊界為一固定雨量值。另本研究於台灣阿里山道路37k+500~40k+100進行現地應用實例模擬與分析，對39k+450處上邊坡進行細部模擬建立降雨逕流門檻值曲線，由所建立門檻值曲線可做為此處初步預警之條件。

Heavy rainfalls brought by Typhoons frequently plague Taiwan. Due to vulnerable geology, rainfall-induced shallow landslides occur very often and sometimes induce tremendous loss in human lives and properties. In order to reduce the landslide disasters, landslide modeling has become an important tool for landslide potential analysis. Generally, rainfall infiltration is taken into account for the boundary condition in landslide models. Many studies investigated the rainfall-induced landslide through physical modeling by integrating the rainfall infiltration simulation and slope stability analysis. Pore water pressure is affected both by groundwater and infiltration caused by rainfall and/or surface runoff. The pore water pressure is investigated to accumulate faster from infiltration of slope surface runoff than groundwater level rise; and thus, slope surface runoff plays a key role in pore water pressure accumulation leading to slope failure. To tackle with the surface runoff infiltration problem, this study proposes a rainfall-induced shallow landslide model based on the slope stability analysis with the limit equilibrium method coupled with the 1D kinematic wave runoff and 2D Richard equations. Sensitivity analysis is later employed to prioritize to the most sensitive parameters (in sensitivity ranking order) including the: (1) friction angle of unsaturated soils, which slightly increases with increasing matric suction; (2) effective friction angle; (3) effective cohesion; and (4) soil specific gravity. This study further set up various scenarios of hydrologic conditions for testing the proposed landslide model. Results show that the rainfall amounts, durations and patterns greatly influence on the landslide occurrence in terms of the landslide depth and occurrence time. So do the surface runoff rates with different durations and patterns. Some cases indicate that slopes do not slide during a rainfall event but slide after the rainfall with the continuous surface runoff inflow into soil bodies. Thus, surface runoff may also influence the occurrence of landslide and its timing. This study further develops several pre-warning threshold curves for a design slope based on the proposed 2D landslide model with various design rainfalls. Furthermore, the slopes located at 37K+500-40K+100 of the Ali-Shan Highway System are taken as a case study. Results show that unlike the threshold curves given by the conventional 1D landslide model, the proposed 2D landslide model portrays the threshold curves with upper and lowers bounds for issuing delicate pre-warnings.