Debris-flow susceptibility assessment using the characteristic factors of a catchment

In traditional application of GIS, the susceptibility mapping based on distinct grids usually neglect the organic unity of the natural element. For this reason, this article investigates the susceptibility model based on catchments, using the valleys of the Jinsha River between Jinjiang and Lindangke as a working example. Firstly, the study area is divided into 1 570 catchments, in which 70% (1 090 catchments) are used to build the susceptibility model and 30% (480 catchments) are used as validation samples. Secondly, 9 feature factors, including the melton ratio, average stream power index of catchment, average topographic wetness index of catchment, catchment elongation, average height of catchment, drainage density of catchment, average fractional vegetation cover of catchment, average slope of catchment and catchment relief ratio are normalized and then selected as independent variables. Whether debris-flows occur is the dependent variable. Susceptibility assessment models of debris-flow are then built using the logistic regression and are classified into five degrees. Finally, the model can predict well (AUC=0.821) through the test of independent validation samples. Moreover, both the absolute and relative incidence of debris-flows increase with the susceptibility degree.