Evaluating the vulnerability of slope-related geological hazards in Southern China based on support vector machine:A case study of Huaiji County,Zhaoqing City
South China,particularly represented by Guangdong Province,is prone to frequent geological disasters that are characterized by small scale and wide distribution.These disasters are predominantly skewed towards certain regions.According to relevant statistics,Zhaoqing City in Guangdong Province has over 400 hidden dangers related to geological disasters,directly threatening approximately 40,000 people.The prevention and control of these geological disasters have become increasingly challenging due to several factors.These include an increase in extreme weather conditions such as local heavy rainfall and strong tropical storms,as well as the impact of human-made engineering activities like urban construction,transportation development,water conservancy projects,and major industrial initiatives driven by economic growth.Huaiji County within Zhaoqing City possesses an extremely fragile geological environment.The region experiences frequent and intense precipitation during its annual rainy season.It exhibits typical geological and geomorphological characteristics of South China,making it a high incidence area for geological hazards.This makes Huaiji County both an important and difficult area for evaluation and research.In this study,Huaiji County was selected as the research pilot area.Utilizing multi-source remote sensing satellite data and airborne LiDAR point cloud data,we evaluated slope height,slope angle,rainfall,surface lithology,and vegetation index as key factors.Both information models and support vector machine(SVM)models were employed to comprehensively interpret and compare the spatial distribution rules and influencing factors of geological hazards in the area.The evaluation results for slope geological hazards in the study area were obtained and verified through superposition analysis using existing geological disaster datasets.The findings indicate that the geological hazard susceptibility evaluation results align well with the distribution of existing geological hazard sites.Compared to the area under the curve(AUC)value of 0.6036 from the ROC curve accuracy of the information model,the SVM model demonstrated superior performance with an AUC value of 0.8043.In conclusion,the support vector machine model method proposed in this study proves to be feasible and reliable for evaluating the vulnerability of slope geological hazards.This research constructed a landslide susceptibility map of the study area,confirming that the region is highly susceptible to landslide dangers,especially near fault zones.Future steps involve deepening the interpretation of micro-scale geological hazards from a geological structure perspective,thereby laying a foundation for effective identification of disaster hazards and scientific assessment and grading of risks.
NETL
NSTL
万方数据
