Using Hong Kong slope risk management to study landslide susceptility in Taiwan - Keelung and Taipei as examples

• Main Authors: Yeuh-Chun Tu, 杜岳峻
• Other Authors: Hung-Jiun Liao
• Format: Others
• Language: zh-TW
• Online Access: http://ndltd.ncl.edu.tw/handle/n8z5rx

碩士 === 國立臺灣科技大學 === 營建工程系 === 95 === This research tries to apply the slope risk management system developed by the Geotechnical Engineering Office of Hong Kong Government to evaluate the landslide susceptibility in Taiwan. Two methods of the slope risk management system are studied: the New Priority Classification Systems (NPCSs) and the Logistic Regression Analysis. Taipei and Keelung are chosen as the study cases. During this study, the NPCSs was found difficult to follow due to the difficulties in obtaining the parameters needed for the evaluation process. In contrast, the Logistic Regression Analysis showed a better result. By inputting the same influencing factors as those used in Hong Kong, the overall accuracies in predicting the landslide potentials of Taipei and Keelung are equal to 73.69% and 77.74% respectively. These numbers are close to that of Hong Kong (=74.4%). Considering the extensive database has been established in Hong Kong, the results obtained from Taipei and Keelung are satisfactory. In other words,, the applicability of the Logistic Regression Analysis in evaluating the landslide potential in Taiwan is acceptable. Finally, a recent landslide case occurred in Keelung area (08/06/2007) is selected to test the suitability of this method. The calculated failure potential of this case is 68%. But the failure potential can be increased to 82% after adding two additional features in the analysis: the distances to river and fault. It indicates that the accuracy of the Logistic Regression Analysis can be improved if it is trained with local factors.