Landslide Susceptibility Assessment Based on GIS: A Case Study in Wanzhou County, Three Gorges Reservoir

Three gorges reservoir was operated from 2003. Influenced by water level fluctuation, more than 20 counties and towns are threatened by landslide hazards, which are widely distributed, frequently happened; fast moved and would result in serious consequents. In order to predict the landslide susceptibility in county area impacted by reservoir, Wanzhou district in Three Gorges Reservoir is taken as the sample research area. Based on the index factor state grading and correlation analysis results, seven influence factors are chosen to be the evaluation index: the slope gradient, slope direction, slope structure, stratum lithology, geological structure, the role of water, and land use. According to more than 700 landslides sample data are drawn. The abrupt change points on landslide frequency curve and information magnitude curve are taking as the critical value to ascertain the factor states, based on which the susceptibility evaluation index system can be established. The susceptibility evaluation is developed by GIS grid data model and weighted information value method. The results show: (1) The high and relatively high susceptibility areas are mainly distributed in the construction land, where human activities strongly changed the natural geology environment; (2) Landslides are largely developed in J2s2 and J2s3 stratum, where exist interbedding of thick mudstones and sandstones. Montmorillonites, contained in the interbed, would swell and soften when meeting across water, which would decrease the shear strength of the interbed and result in landslide; (3) Reservoir water fluctuation area also led high susceptibility. Then, the urban area of Wanzhou County is largely developed and rehabilitated by the government and it shows high susceptibility and low risk when no powerful extra factors are impacted on. The statistic results shows the high and relatively high area is 1210km2, which takes up 9.71% and 25.9% of the whole area, The assessment accuracy is high up to 86.2%, and it can provide effective data for large scale prediction of landslides primary source areas.