淺層崩塌潛勢及影響範圍分析-以南投縣為例

淺層崩塌為最常見的邊坡破壞現象，經常因為降雨而發生，且淺層崩塌會提供豐富的土砂材料來源，可能會觸發土石流的發生，南投縣內地形主要以山地為主，包含了不少地質破碎及地勢陡峭的區域，位於南投縣內，省道為居民重要的聯外道路，其突發性的邊坡破壞發生於主要道路上，會造成道路阻斷，對居民的生計也會造成影響。本研究對南投縣內的淺層崩塌進行分析，收集可能導致淺層崩塌的影響因子，將各因子經統計及相關性分析後，選取出坡度、坡向、坡型、全坡高、地層、距稜線距離、距河道距離及距斷層距離等8個淺層崩塌潛勢因子，對南投縣內淺層崩塌與各因子進行區別分析，並利用正判率及ROC曲線驗證區別分析之結果，由驗證結果得出區別分析之正判率為74.15%，且ROC曲線法之曲線下面積AUC=0.818，可確認本研究使用區別分析所得之淺層崩塌潛勢值F，再判別是否為淺層崩塌上具有良好的判別能力，將其結果淺層崩塌潛勢值F分成低、中、高與極高四個等級，並套疊於現地調查之區域，繪製出各淺層崩塌區域之淺層崩塌潛勢圖。於淺層崩塌影響範圍分析，比較流向演算法(Flow-R)與顆粒流模擬法(Rockyfor3D)兩種不同的演算法所模擬之影響範圍，根據現地調查之淺層崩塌模擬結果，可得知流向演算法易受地形之影響，如蝕溝及道路，顆粒流模擬法之料源擴散不會因為道路而停止。本研究使用陰影角(Fahrböschung angle)對實際崩塌情況、流向演算法及顆粒流模擬法進行比對，所計算出的陰影角分別為36.04°、37.58°及35.99°，其顆粒流模擬法與實際崩塌之陰影角較符合。

關鍵字

淺層崩塌；定量統計法；影響範圍分析

並列摘要

Shallow landslide is the most common slope failure phenomenon, which usually occurs due to rainfall. Shallow landslide will provide a rich source of soil and sand materials, which may trigger debris flow. The terrain in Nantou County is mainly mountainous, including many areas with broken geology and steep terrain. Provincial highways are important roads connected to the outside world. The main road's sudden slope failure can cause roadblocks, the livelihood of the residents will be affected. This study analyzes the shallow landslide in Nantou County and collects the influencing factors that may lead to the shallow landslide. After statistics and correlation analysis of each factor, select eight potential factors for shallow collapse, including slope, aspect, slope type, total slope height, strata, distance from ridge, distance from the river, and distance from the fault. Distinguish analysis of the shallow layer collapse in Nantou County and various factors, and verify the difference analysis result using the positive judgment rate and ROC curve. According to the verification results, the differential analysis's positive judgment rate is 74.15%, and the area under the curve of the ROC curve method AUC=0.818, which can confirm that this study uses the shallow collapse potential value F obtained from the differential analysis. It has an excellent ability to discriminate whether it is a shallow collapse. The resulting shallow collapse potential value F is divided into four levels: low, medium, high, and too high, and overlay it on the area of on-site investigation, draw the shallow collapse potential map of each shallow collapse area. In analyzing the impact range of shallow avalanche, compare the impact range simulated by two different algorithms of flow direction algorithm (Flow-R) and particle flow simulation method (Rockyfor3D). According to the shallow collapse simulation results of the field survey, it can be known that the flow direction algorithm is easily affected by the terrain, such as erosion ditches and roads, and the material diffusion of the particle flow simulation method will not stop because of the road. This study uses the shadow angle (Fahrböschung angle) to compare the actual collapse, flow direction algorithm, and particle flow simulation method. The calculated shadow angles are 36.04°, 37.58°, and 35.99°, respectively. The particle flow simulation method is more consistent with the shadow angle of the actual collapse.