Geological Hazard Susceptibility Assessment In Baqiao District Based On RF And SVM Model

In recent years,the frequency of geological disasters has been speeding up,leading to many disaster accidents and destroying infrastructure such as houses and roads,which has become a major hidden danger for the safe development of people in disaster areas and restricted the sustainable development of society and economy.More and more attention has been paid to geological hazard susceptibility assessment.On the basis of analyzing the general situation of geological environment in Baqiao district,this paper evaluates and studies the vulnerability of geological disasters in Baqiao district,and obtains the following main research results:(1)Based on the geological disaster investigation data of the bridge area,the geological environment conditions,the types,the development characteristics and the distribution characteristics of the geological hazards in the study area are mastered.(2)the influence factors of geomorphology,elevation,slope,slope direction,curvature,rainfall,water system,stratigraphic lithology,rock and soil mass,normalized vegetation index,road and fault are selected as the main evaluation factors of geological hazards.The spatial distribution relationship between each factor and geological hazard is analyzed by disaster point density.The correlation of each factor was analyzed by Spearman correlation coefficient,and the evaluation index system of geological hazard susceptibility assessment in Baqiao district was established.(3)the random forest(RF)and support vector machine model based on linear kernel function(LN),polynomial kernel function(PL),radial basis function(RBF)was used to evaluate the susceptibility of the study area.The index of susceptibility evaluation under different models is obtained.According to the susceptibility index,the whole study area is divided into four levels:very high susceptibility area,middle susceptibility area and low susceptibility area.Finally,the map of regional susceptibility is drawn.(4)the accuracy of the model is judged by ROC curve and Kappa coefficient respectively.The results show that the prediction precision of random forest model is the highest,and the prediction precision of radial basis function and polynomial function under support vector machine is the second(5)the superiority of the stochastic forest model is verified by the density of disaster points in the extremely high prone area and the high prone area,and the final distribution result diagram is obtained.The weight of the evaluation factors under each model is obtained and the main disaster factors in the study area are determined.