Based on the geological hazard risk census in Liwan District of Guangzhou City,the susceptibility of land subsidence and ground collapse geological hazards in Liwan District has been studied in this thesis.Through on-site investigation,the geological environment conditions and disaster distribution characteristics of Liwan District have been mastered.Different factor classification and multi-model mixing are used to improve the accuracy of geological disaster susceptibility evaluation in Liwan District.The main research contents and results are as follows:(1)Based on the selection principle of evaluation index and multi-source data,taking Liwan District of Guangzhou City as the research area,combined with multi-source data such as geology,remote sensing and hydrology,a total of 7 evaluation factors such as fault buffer zone,water system buffer zone,filling thickness,silt thickness,clay thickness,sand thickness and ground object classification were selected.(2)Statistical analysis of geological disaster information in Liwan District determines that the main types of disasters in the study area are land subsidence and ground collapse.The land subsidence is mainly distributed in Qiaozhong Street and Dongjiao Street,and the ground collapse is mainly distributed in the area of Datansha in Qiaozhong Street.(3)Different grading treatment of evaluation factors.The evaluation factors are divided into different subsets based on the natural discontinuity method and the frequency ratio method,and the accuracy of the same model under different grading conditions was compared.At the same time,the appropriate evaluation factors were screened by Pearson correlation coefficient and collinearity diagnostic cross analysis.(4)The analytic hierarchy process(AHP)and information volume(IV)mixed model,certainty factor(CF)and logistic regression(LR)mixed model,and random forest model(based on En Map-Box software package)were used to evaluate the susceptibility of geological disasters in the study area.According to the equal interval method,the probability value of susceptibility is divided into four grades.(5)Through the model rationality test and ROC curve,the comprehensive analysis shows that the mixing model of certainty factor(CF)and logistic regression(LR)based on frequency ratio method has the highest accuracy and is more suitable for the evaluation of geological disaster susceptibility in Liwan District. |