• 城市与乡村 •

### GIS支持下三峡库区秭归县滑坡灾害空间预测

1. 中国地质大学地球物理与空间信息学院,武汉430074
• 收稿日期:2010-02-05 修回日期:2010-05-13 出版日期:2010-10-20 发布日期:2010-10-20
• 通讯作者: 牛瑞卿(1969-),男,博士,副教授。主要研究方向为天空地一体化地球观测信息融合与可视化、人类工程活动与岩石体变化遥感监测技术、遥感信息定量化和反演。
• 作者简介:彭令(1984-),男,重庆人,博士研究生。主要研究方向为地学信息分析、空间知识发现。E-mail:wuhanpl@gmail.com。
牛瑞卿(1969-),男,博士,副教授。主要研究方向为天空地一体化地球观测信息融合与可视化、人类工程活动与岩石体变化遥感监测技术、遥感信息定量化和反演。
• 基金资助:

国家自然科学基金项目(40672205);国家高技术研究发展计划(2007AA12Z100);国土资源部三峡库区三期地质灾害防治重大科学研究项目(SXKY3-2-2)

### Landslide hazard spatial prediction in Zigui County ofthe Three Gorges Reservoir Area based on GIS

PENG Ling, NIU Rui-qing, CHEN Li-xia

1. 1. Institute of Geophysics and Geomatics, China University of GeoSciences, Wuhan 430074, China
• Received:2010-02-05 Revised:2010-05-13 Online:2010-10-20 Published:2010-10-20

Abstract:

Landslide prediction is very important in disaster prevention and reduction procedures, and it is one of practical research fields to evaluate and predict landslide hazards using statistic analysis model and spatial analysis of GIS. The aim of this study is to analyze landslide susceptibility using Logistic regression model in Zigui County of the Three Gorges Reservoir Area. In this paper, seven evaluation factors are selected, i.e. topographic slope, topographic aspect, bed rock-slope relationship, lithology, land use and distance from road and drainage. In susceptibility mapping, the use of logistic regression is to find the best fitting function to describe the relationship between the presence or absence of landslides (dependent variable) and a set of evaluation factors such as topographic slope and lithology. Here, an inventory map concerning 37 landslides was used to produce a variable, which takes a value of 1 for the presence and 0 for the absence of slope failures. In order to improve the accuracy and credibility of the model prediction, methods to reduce spatial autocorrelation in a logistic regression framework are also discussed. An optimal sampling scheme that can eliminate spatial autocorrelation whilst maintaining enough samples to achieve the accuracy based on the model is developed. The model was tested by the overall model statistics, and the results indicate that the model fits the dataset. The effect of each parameter on landslide occurrence was assessed from the corresponding coefficient that appears in the logistic regression function. The interpretation of the coefficients showed that land use plays a major role in determining landslide occurrence and distribution, although field observations showed that engineering construction exerts great influence on slope stability. With the help of a predicted probability map, the study area was classified into four categories of landslide susceptibility: high, moderate, low and none. The moderate and high susceptibility zones make up 38.9% of the total study area. In comparison to the occurrence of historical landslide hazards, the precision using logistic regression model can be up to 77.57%.