Formation sensitivity of landslide dam based on geomorphic characteristics

Abstract

Landslide dams formed by unstable slope usually blocks the river, and later fail to trigger huge floods threatening the downstream residents' lives and properties.In order to improve the ability of active disaster reduction and prevention, it is urgent to construct a method to quickly predict and judge the ability of landslide to build dams.This paper firstly extracts the geomorphic characteristic parameters of 70 typical landslides based on literature reading and remote sensing, including 50 damming landslides.The five parameters (landslide volume, area, relief, length and river width) are determined by K-S test and M-W U test, and the Boruta algorithm is used to determine the factor importance.Then, the Bayes discriminance method and logistic regression method are used to build a prediction model of the landslide dam, in which the accuracy exceeds 90%.The threshold equation combined with H and L/Wr is determined by using factors with high importance and significant differences and the ratio method in order to carry out a rapid judgment of landslide dams formation.The statistic analysis of the landslide geomorphic characteristics of different triggering factors and the prediction result of V-Wr empirical formula determine the relationship between landslide dams formation and triggering factors, providing technical support for the further construction of prediction models for the formation of landslide dams with different inducements.

Keywords

• landslide dam
• geomorphic characteristic
• bayes discriminant method
• logistic regression
• threshold equation