GIS-Based Susceptibility Mapping with Comparisons of Results from Data-Driven Bivariate versus Logistic Regression in the Three Gorges Area, China
The purpose of this study is to evaluate and to compare the results of data-driven bivariate and logistic regression model for landslide susceptibility mapping based on GIS. In order to achieve this goal the Zhongxian segment in the Three Gorges area, China was selected as a test area because the Three Gorges Region is an area of high landslide hazard in the history. And this region has received a large amount of attention and research because of the Three Gorges Dam and Reservoir Project and the potentially strong impact on the environment, geo-hazards, and socioeconomy. The site covers an area of 260.9 km2 with a landslide area of 5.3 km2. A detailed landslide inventory map of the study area was identified by interpreting of 1:20,000-scale color aerial photographs and extensive field surveys. Four data domains are used in this study: remote sensing products, thematic maps, geological maps, and topographical maps. The size of pixels for all of the data layers was 25×25 m2. The new bivariate and logistic regression methods were applied to the lest zone to overcome some disadvantages of previous methods, and two separate susceptibility maps were produced. Both methods were validated and compared between Che susceptibility maps and the 6793 landslide seed cells. This indicates that the logistic regression map was better to capture the reality on the ground than the bivariate method equivalent.
Shibiao Bai Pinggen Zhou Jian Wang Shengshan Hou Guonian Lu Fanyu Zhang
National Education Administration Key Laboratory of Virtual Geographic Environments, Nanjing Normal Chinese Institute of Geological Environment Monitoring, Beijing 100081 ,China School of Civil Engineering and Mechanics, Lanzhou University, Lanzhou 730000, China
国际会议
The 12th Conference of the International Association for Mathematical Geology(第12届国际数学地质大会)
北京
英文
163-165
2007-08-26（万方平台首次上网日期，不代表论文的发表时间）