| Geological disasters in China have the characteristics of large quantity,wide distribution,high frequency,strong destructive power and unpredictable.The special topography and geomorphology of Guizhou Province make its geological disasters more developed.This paper takes Wangmo County,Guizhou Province as the research area.On the basis of summarizing its geological situation and disaster data,this paper analyzes the development characteristics and laws of geological disasters in this area,constructs the evaluation index system to classify and analyze the selected influencing factors one by one,and uses a variety of machine learning models to study the landslide susceptibility of Wangmo County.Finally,through comparison and verification,the evaluation model suitable for Wangmo County landslide geological disasters is selected.The research contents are as follows :(1)Get the general situation and background of geological disasters in Wangmo county.Through collecting data and field investigation,it is concluded that the main types of geological disasters in the study area are landslide,collapse,debris flow and ground collapse.Among them,landslide disasters account for 85.25 %.The small and medium-sized scale of geological disasters,the highest proportion of poor stability.It is mainly distributed in slopes with slope greater than 25 °,soil slopes,clastic rock distribution areas,areas with strong rainfall,and areas with high human engineering intensity.Geological disasters occur more frequently in May,June,July,August and in the deformation stage.(2)The spatial distribution characteristics of landslide disasters in Wangmo County were analyzed.The nearest neighbor analysis method,z-score analysis method and Ripley ’s K function analysis method were used to determine that the more developed landslide disaster points in the study area were clustered and greatly affected by spatial scale.There is no significant fractal feature of landslide disaster points in the study area by box-counting dimension analysis.(3)The relationship between the distribution of landslide disaster points in Wangmo County and the single influencing factor is determined.According to the unique conditions of the study area,11 influencing factors of landslide disasters were selected,and the development of landslide disasters under different classifications of each factor was analyzed by Arc GIS software.The results showed that the landslide was most developed when the elevation was 717-881 m,the slope direction was west,the slope was 18 °-20 °,the curvature was-1.36-0.04,the land use type was grassland,the stratum lithology was soft rock,the water system distance was < 408 m,the road distance was < 300 m,the distance from the fault was < 5750 m,the NDVI value was 0.65-0.75,and the average annual precipitation was 1594-1607 mm.(4)Verify the rationality of the selected landslide susceptibility factors in Wangmo County.Through principal component analysis and collinearity analysis,it is determined that the selected influencing factors meet the requirements of scientificity and independence,and the weight of each evaluation factor in the landslide is obtained,and the sensitivity of each evaluation factor to the landslide is determined.Finally,the annual rainfall index is the most sensitive to the landslide,and the NDVI index is the least sensitive.(5)The best model of landslide hazard susceptibility evaluation in Wangmo County was constructed.The five models of logistic regression model(LR),random forest model(RF),support vector machine model(SVM),LR-SVM and RF-SVM coupling model were used to evaluate and partition the susceptibility of landslide disasters in Wangmo County.By judging the zoning rationality of each evaluation model,comparing the value of several classification accuracy indicators and the receiver operating characteristic curve(ROC curve),the accuracy of the model is verified,and the remote sensing image of the landslide disaster point downloaded by Google earth is used for example verification.Finally,the LR-SVM model is the best choice for the evaluation of landslide susceptibility in Wangmo County. |
PDF Full Text Request