Study on the Factors of Landslide Susceptibility Analysis in Chen- You-Lan Watershed

各方學者經常以歷史事件資料，進行崩塌潛感分析來做為預測未來災害發生的可能性，期望減少崩塌所造成的傷害，而崩塌潛感模式經常針對可能造成崩塌之促崩因子進行分析，經資料蒐集回顧後發現，降雨強度常被用來做為因子或計算因子參數使用，前人所使用之降雨因子如最大降雨強度、三小時、六小時、九小時累積之降雨強度等。 莫拉克侵台期間對陳有蘭集水區內之邊坡造成破壞，該邊坡破壞情況與降雨強度具顯著相關性，因此本文分析各斜坡單元與不同連續時間降雨強度之關係；另張紘揚(2018)指出因為降雨分佈的時空不均勻性，各雨量站幾乎不可能同時達到最大降雨強度，若以最大降雨強度進行崩塌潛勢分析，勢必與實際情況有所差異。故本研究選用降雨強度包括1.視為同時版本最大降雨強度因子(1小時最大降雨強度)；2.逐時版本最大降雨強度因子，張紘揚(2018)；3.總累積雨量因子(144小時)，經分析後結果可做為未來預警系統降雨強度參考值，以提供山崩潛感分析降雨因子之選用。而促崩因子處理係以克利金法估算斜坡單元之最大降雨強度、累積降雨強度及張紘揚(2018)之逐時版本最大降雨強度，再以羅吉斯迴歸進行降雨型崩塌潛勢分析並比較其差異。 分析結果顯示，崩塌最小門檻值面積625平方公尺所判釋的崩塌面積，分別計算操作者特征曲線下面積(AUC)，結果AUC介於0.622至0.701；再以崩壞比15%之崩塌判釋結果分別計算操作者特征曲線下面積(AUC)，結果接收者操作特征曲線下面積介於0.723至0.751，較為良好。由研究結果可知，以崩壞比15%做為崩塌分析樣本，並取最大降雨強度因子進行降雨型崩塌潛勢分析有較好的判釋成效。Scholars from all areas often use the historical data as a possibility to predict the occurrence of future disasters to analyze the landslide susceptibility, and hope to reduce the damage caused by the collapse. The Landslide Susceptibility model is used to analyze the collapse-causing factors that may cause collapse. After reviewing the data, scholars found that the rainfall intensity is considered as a factor or calculation factor parameter; for example, predecessors often use the parameters such as the maximum rainfall intensity, the three-hour, six-hour, and nine-hour cumulative rainfall intensity. Typhoon Morakot caused damage to the slope in the catchment area of Chen-You-Lan Watershed.There is a significant correlation between the slope damage and the rainfall intensity. This paper analyzes the relationship between the slope units and the rainfall intensity at different continuous times; Zhang Yuyang (2018) pointed out that because of the spatial and temporal heterogeneity of rainfall distribution, each rainfall station could hardly to reach the maximum rainfall intensity at the same time. If the collapse potential analysis is carried out with the maximum rainfall intensity, it must be different from the real situation. Therefore, the rainfall intensity selected in this study includes 1. the maximum rainfall intensity factor (1 hour maximum rainfall intensity) as the same version; 2. the Estimate Rainfall Intensity Hourly, Zhang Yuyang (2018); 3. the total cumulative rainfall factor (144 hours). After analysis, the result can be used as a reference of the rainfall intensity, and thus it can be a future warning system to be as one of the selections for the landslide potential analysis. On the other hand, the accretion factor treatment is to estimate the maximum rainfall intensity by Kriging method and the cumulative rainfall intensity and the maximum rainfall intensity of Zhang Yuyang (2018) in a Estimate Rainfall Intensity Hourly version; then, analyze and compare the rainfall-type collapse potential with Logit Regression Analysis. The analysis results show that the collapse of 625 square meters, the collapse area of the minimum threshold, is calculated with the area under the operator characteristic curve (AUC). And the result AUC is between 0.622 and 0.701; then, the area under the operator characteristic curve (AUC) was calculated by the collapse test result of the collapse ratio of 15%, and the area under the receiver operating characteristic curve was between 0.723 and 0.751 which it's more better. According to the research results, it is known that the landslide ratio is 15% as the collapse analysis sample, and the maximum rainfall intensity factor is used to analyze the rainfall-type collapse potential.摘要 i Abstract ii 表目錄 vii 圖目錄 viii 第一章 前言 1 1.1 研究動機 1 1.2 研究目的 2 1.3 研究流程 2 第二章 文獻回顧 4 2.1 以降雨為促崩因子相關研究 4 2.2 分析單元比較 5 2.2.1 網格單元法 5 2.2.2 斜坡單元法 5 2.2.3 特別條件單元法 6 2.2.4 行政單元法 6 2.2.5 地貌單元法 6 2.3 崩塌地判釋方法 8 2.4 山崩潛勢研究方法 8 2.4.1 定性法 8 2.4.2 定量法 9 第三章 研究材料與方法 11 3.1 研究區域概述 11 3.1.1 地形 11 3.1.2 地質 16 3.2 基本資料蒐集 19 3.3 斜坡單元 20 3.3.1 陳有蘭集水區斜坡單元繪製 20 3.3.2 分析單元擷取 21 3.4 地形因子選用 22 3.5 雨量資料處理 23 3.5.1 視為同時版本最大降雨強度因子資料 26 3.5.2 逐時版本最大降雨強度因子資料 26 3.5.3 總累積雨量因子(144小時)資料 27 3.6 羅吉斯迴歸模式 28 3.6.1 理論概述 28 3.6.2 羅吉斯迴歸評估方法 33 第四章 結果與討論 37 4.1 崩塌分析樣本選取 37 4.1.1 崩塌最小門檻值 37 4.1.2 崩壞比 40 4.2 模型結果 43 4.2.1 接收者操作特征曲線下面積（AUC）結果 43 4.2.2 準確率計算 48 4.2.3 崩塌模型方程式 51 4.2.4 羅吉斯迴歸係數顯著性比較 53 第五章 結論與建議 55 5.1 結論 55 5.2 建議 56 參考文獻 57 附錄一 崩壞比15%判釋崩塌地之判識結果 附59 附錄二 ROC坡向分析崩壞比結果 附66 附錄三 羅吉斯迴歸係數結果 附6