應用羅吉斯迴歸法進行阿里山地區山崩潛勢評估

The Application of Logistic Regression for Landslide Susceptibility Analysis-A Case Study in Alishan Area

李嶸泰(Jung-Tai Lee) ； 張嘉琪(Chia-Chi Chang) ； 詹勳全(Hsun-Chuam Chan[2) ； 廖珮妤(Pei-Yu Laio) ； 洪雨柔(Yu-Jou Hung)

《中華水土保持學報》 43卷2期 (2012/06) Pp. 167-176

https://doi.org/10.29417/JCSWC.201206_43(2).0006

摘要

山崩潛感分析通常需要結合數種能辨別山崩的地形、地質及水文等因子，共同找出一組能解釋山崩的因子組合，建立山崩潛感模型和計算潛感值，繪製成山崩潛感圖。本研究以阿里山地區為例，應用多變量統計中的羅吉斯迴歸方法進行山崩潛勢評估研究，為了有效挑選山崩潛感因子，針對各因子進行檢定、篩選，山崩潛感分析的結果並加以驗證；並利用分類誤差矩陣表檢驗羅吉斯迴歸預測之準確性，了解分析成果之品質；最後分別以10、25、50及100年不同重現期距之降雨量計算降雨誘發山崩潛感值，並將崩塌潛勢分為高潛勢、中高潛勢、中潛勢及低潛勢四個等級以繪製山崩潛感圖。

關鍵字

山崩潛感 ； 羅吉斯迴歸 ； 分類誤差矩陣 ； 山崩潛感圖

並列摘要

Landslide susceptibility analysis usually combines several factors, including the terrain, geology, and hydrology. The analysis tries to find a suitable combination of these factors in order to establish a landslide susceptibility model and calculate the susceptibility value. A potential landslide map can be established by using the calculated the susceptibility value of landslide. This study took Alishan area as an example and aimed to assess landslide susceptibility analysis by Logistic regression, a multivariate analysis method. In order to select the factors efficiently, the calibration and selection procedure were performed. The results were verified by a previous typhoon event. The classification error matrix was used to evaluate the accuracy of landslide predicted by the present model. Finally, this study applied 10-, 25-, 50-, and 100-year return periods precipitation to estimate the susceptibility values for the study area. The landslide susceptibilities were separated into four levels, including high, medium-high, medium, and low, to delineate the map of potential landslide.

並列關鍵字

Landslide susceptibility ； logistic regression ； map of potential landslide