An evaluation of landslide susceptibility using probability statistic modeling and GIS's spatial clustering analysis

As landslides caused multiple casualties and destructions on a global scale, various models were applied to landslide susceptibility evaluation. In this study, a probability statistic model called the certain factor was adopted to assess the landslide susceptibility of Danba, a county in southwestern China, where the landslide events occur frequently but were poorly understood in regional landslide susceptibility. With the validation of area under the prediction rate curve, the resulting susceptibility map has the accuracies of 0.8211 and 0.8288 in experiment area and verification area, respectively. The validated assessment result was further processed to identify landslide-prone areas with the aid of the spatial clustering analysis of geographic information system. Two clustering indexes including Moran's I statistic and local indicator of spatial association (LISA) were involved. The Moran's I index of 0.959 and the LISA identification result accordant with previous investigations proved that the proposed method was rational and efficient to find the landslide-prone regions and make relevant decisions.