A Discussion on the Influence Factors of the Landslide Density of the Small Catchment in the Tseng-Wen Reservoir Basin of Taiwan

• Main Authors: Hsiu-Chen Wei, 魏秀珍
• Other Authors: Shih-Hsu Chen
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/71792849514961888784

碩士 === 國立成功大學 === 資源工程學系碩博士班 === 94 === 　　The landslides in the reservoir basin affect the effective life of the reservoir and the quality of water supply. Many factors may affect the stability of the slopes in the reservoir basin, such as topography, geology, climate, humanity activity and land utilization. This study chose the Tseng-Wen reservoir basin, which is 481 square km in size, to study the affect of the influence factors on the landslide. Since the study area is too big for cell by cell operation method such as GIS method, the landslide density of each small catchment was then calculated, and the relationship between the influence factors and the landslides densities was then postulated by multiple regression and neural network methods. The landslide density of each catchment area is defined as the ratio of the landslide area divided by the whole area of the catchment. 　　The first step of this study is to divide the reservoir basin into 157 small cactchment by ArcGIS. Then the landslide inventory and land use map of the study area were made by aerial photo interpretation by using 20 cm ground resolution digital aerial photos and Stereo Graphics Monitor Screen. All the topographic data such as elevation, roughness of elevation, slope angle, and the roughness of the slope angles, are derived from the 40m DEM of the study area. The annual precipitation data was obtained from the weather stations among and near the study area, and the average annual precipitation of each small catchment was obtained by Kriging method. 　　In the first step of the data analysis, the coefficients of correlation between each pair of data (4 topographic factors, 2 land use factors, 1 climate factor, 1 geology factor, and the landslide density data) were examined. It is found that many influence factors are closely related, in other words, many influence factors are not independent from each other. It is also found that two lad use factors, farm land and construction land density, are negatively related to the landslide density. If all the eight influence factors are put into analysis by the multiple regression and neural network methods, some unreasonable results may be obtained such as negative correlation between slope angle (or annual precipitation) with the landslide density. However, if the real independent factors are used in the analysis work, all the unreasonable result will disappear, and the land use densities will also show a positive correlation with the landslide density. It is concluded that the topographic factors are the most affective factors on the stability of the slopes in the study area, and the negative correlation between land use factors and landslide density is caused by the stronger relationship between the topographic factors and landslide density.