Fuzzy Logic Approach For Regional Landslide Susceptibility Analysis Constrained By Spatial Characteristics Of Environmental Factors

The geological conditions of the mountainous areas in western China are complex,internal and external dynamics are strong,and catastrophic landslide events occur frequently.Landslide susceptibility analysis has become a necessary means for landslide risk management,scientific early warning and active prevention,for rapid and effective investigation and monitoring of large-scale regional landslide distribution.It's also provides a reference for the rapid and effective investigation and monitoring of landslide distribution and activity status in a large area,and for the government's disaster reduction and prevention work.Faced with the objective status quo of high concealment of landslides,incomplete landslide inventories data,large amount of observation data but small amount of temporal and spatial information,the single knowledge-driven or data-driven approach in traditional landslide susceptibility analysis have obvious limitations and can't give full play to its advantages.How to systematically analyze the disaster mechanism of landslides,effectively solve the heterogeneity of the landslide environment,and obtain reliable and accurate landslide susceptibility results,has become a bottleneck problem that need to be broken.This thesis takes the landslide-prone belt in Fengjie County of Chongqing city in the Yangtze River Basin as the research area,analyzes the spatial characteristics of landslide hazards of environmental factors such as topography,geomorphology,traffic water system,etc.,and constructs regional landslide susceptibility coupled with information entropy weight and fuzzy logic model.The paper also tests and explores the influence of different restraint modes and different evaluation units on the accuracy of the fuzzy logic model susceptibility analysis results,which showed the feasibility of the combination of information entropy weight model and fuzzy logic theory in landslide susceptibility analysis.The main research results of this thesis are as follows:(1)Collecting and processing the basic geographic data of the study area,analyzing the natural geographic environment characteristics of the study area,the disaster mechanism of landslide disasters and the disaster environment,selecting six landslide hazard environmental factors,such as slope,rock,distance from the fault,and distance from river,distance from roads,and tree coverage,which constitute a landslide susceptibility evaluation factor set,and the Pearson coefficient method is used to test the correlation between the factors.(2)From the perspective of spatial characteristics,using statistical indicators such as the frequency ratio and information entropy of disaster environment factors to explicitly describe the spatial distribution characteristics and relative contribution of landslide disaster environment factors,and restrict the fuzzy logic process in regional landslide susceptibility analysis.(3)Constructing disaster environmental factor membership function based on the spatial distribution characteristics of environmental factors,and establishing a nonlinear mapping relationship between disaster factors and landslide susceptibility in geographic space;and selecting disaster environmental factors with the relative information entropy to participate in the hierarchical superposition process of the fuzzy membership set of landslide susceptibility,using two nonlinear fuzzy superposition operators,Fuzzy Gamma and Fuzzy Product to obtain a comprehensive landslide susceptibility membership layer.(4)comparing the landslide susceptibility results of the fuzzy logic model with multiconstraint mode,and discussing the suitability of the model to the slope unit.It also compare the landslide susceptibility analysis results of improved fuzzy logic model with information volume model and information volume-logistic regression model.the accuracy of result is verified by quantitative accuracy evaluation indicators such as confusion matrix and ROC curve,which proves the better feasibility and superiority of the method in this paper.