Probability evaluation method of group landslide based on time and space distribution of soil moisture content

Abstract

The application discloses a kind of group landslide probability evaluation methods based on soil antecedent moisture content space-time distribution, comprising: S1: obtaining static factor data and dynamic soil humidity data related to landslide triggering;S2: select the data of a certain time before group landslide event in dynamic soil humidity data, carry out the depth extraction of time sequence characteristics to obtain dynamic soil humidity data characteristic data;S3: build history landslide directory, combine the static factor data obtained in S1 and the dynamic soil humidity data characteristic data obtained in S2 to build the database required for training;S4: use the database obtained in S3 to train group landslide probability artificial neural network;S5: use the group landslide probability artificial neural network trained in S4 to calculate the landslide occurrence probability of each grid in the study area, generate landslide-prone map, consider the time sequence change correlation of soil humidity in the evaluation method, and the applicability for group landslide prediction is stronger.

Images