基于要素分析和二元统计模型的区域滑坡危险等级制图——以国道212线陇南段为例

A Study of Landslide Susceptibility Mapping Based on Factor Analysis and Bivariate Statistics——With a Case Study in Longnan Area of National Highway 212

国道212线陇南段是我国地质灾害最发育的地区之一,绘制该区的滑坡危险等级地图对灾害管理和发展规划是极其必要的。基于滑坡的野外调查、机理研究和室内试验等工作,分析了滑坡与各种要素的相关性,选择控制滑坡的9个重要要素作为评价要素,利用GIS和二元统计的信息值模型和滑坡先验风险要素模型绘制了研究区的滑坡危险等级地图。最后,选用区内11个具有明显滑动位移的活动滑坡与滑坡危险等级地图比较,检验其可靠度。结果表明,活动的滑坡绝大部分都位于危险等级很高和高的范围内,说明两种模型的评价结果与研究区实际情况相吻合,同时也反映出信息值模型与实际情况更加相符。

Longnan of National Highway 212 is one of the most serious areas of geo-disasters in China. Therefore, landslide susceptibility mapping is vitally necessary for disaster management and planning development activities in the area. Based on the field, laboratory and office studies in the study area, and with a study of the relationship between landslides and their causative factors, thematics date layers of petrofabrics types, slope, aspect, altitude, the distance to river, the distance to road, precipitation, land-use types and vegetation cover thematies data layers were used to create the landslide susceptibility map. In the study, a landslide susceptibility map of Longnan of National Highway 212 was produced using the Information Value （InforV） and Landslide Nominal Risk Factor （LNRF） of bivariate statistics with the help of Geographical Information Systems （GIS）. The study area was classified into five classes of relative landslide susceptibility, namely, very low, low, moderate, high, and very high. To prove the practicality, the landslide susceptibility maps were compared with a landslide activity map containing 11 active landslides with sliding displacement. The outcome was that the overwhelming areas of the active landslide zones fit into the high and very high susceptibility classes. The results showed that both the InforV method and the LNRF method corresponded to the actual distribution of landslide susceptibility. At the same time, the InforV method gave a more realistic picture of the actual distribution of landslide susceptibility than LNRF method.