Study Of Landslide Susceptibility Prediction And Failure Mechanism In Xunwu County,Ganzhou City

Landslides have been as the most widespread natural disaster with the characteristics of wide distribution,short occurrence time,high intensity and serious damages to the safety of human life and property.With the rapid development of our country's economic construction,a series of human activities represented by large-scale engineering projects are also rapidly developing,producing a large number of engineering excavation slopes.Because the mountain slope itself has inherent disastercausing characteristics,human activities have destroyed the original slope shape,and under the action of extreme weather such as rainfall,and it is prone to landslide instability.Rainfall and engineering slope excavation have become two important disaster-causing factors for landslide instability of red clay accumulation in southern hilly areas of China.Therefore,it is necessary to research the degree of influence of rainfall and human engineering activities on landslide occurrence and its disastercausing mechanism.Based on this,we can obtain information on landslide cataloguing in Ganzhou City,Jiangxi Province,through field surveys,and the date on the topography,distribution of human activities,stratigraphic lithology,hydrometeorological characteristics,ground cover and other relevant date of the study area.Focusing on the inappropriateness of some environmental factors for landslide susceptibility modeling in existing studies and the effects of human engineering activities on landslide occurrence,the environmental factors in this study are divided into three groups,namely,1)original factors: including elevation,slope,aspect,plane curvature,section curvature,terrain relief,NDBI,NDVI,geotechnical type,distance from highway,distance from river;2)original add rainfall factors: combining the annual rainfall factor with the original factors;3)improved factors: obtaining by using road and gully density environmental factors to replace the environmental factors of distance to roads and rivers based on the original add rainfall factors.Then analyze the relationship between landslide occurrence and each factor through the classification of frequency ratio,and combine three machine learning models such as Support vector machines(SVM)?Logistic regression(LR)?Random forest(RF)to obtain the prediction probability of landslide susceptibility based on above three groups,then use four landslides susceptibility accuracy evaluation methods including factor importance,landslide susceptibility map,accuracy evaluation,and landslide susceptibility index distribution to compare the accuracy evaluation results of each group landslide susceptibility under each model,and by analyzing the prediction results of landslide susceptibility to obtain the best group of environmental factors and model,and to explore the suitability of the above combinations of environmental factors for landslide susceptibility modelling,and the environmental factors that influence landslide development significantly as well as the extent to which rainfall,human engineering activities have an impact on landslide susceptibility.In addition,in order to deeply explore the influence of human engineering activities and rainfall factors on the instability mechanism of red clay landslides,based on the main evolution and damage characteristics of landslides in Ganzhou City,Jiangxi Province,this study reconstructs a typical physical and geological model of red clay landslides with universal significance.According to the local rainfall characteristics,the cumulative heavy rainfall once every 50 years is combined with different slope excavation distances and slope excavation angles,and then the saturated unsaturated seepage theory and the non-linear finite element method are used to calculate the landslide stability and pore water pressure changes by GeoStudio 2007 software,to reveal the instability law and disaster mechanism of red clay landslides under the coupled conditions of heavy rainfall and engineering slope excavation.The main research results are as follows:(1)Based on the catalog date of landslides in southern Jiangxi province,then classify the environmental factors in three factor combinations and use interval frequency ratio to analyze the relationship between each factor and landslide.The results show that different environmental factors have different degrees of effects on landslide occurrence.And combing with the landslide susceptibility results for each combination of conditions,it can be obtained that the three factors of rainfall as well as human activity-related road density and gully density can well improve the model prediction performance and have a very important influence on the development of landslides.(2)The three different environmental factors groups in the study area are processed for landslide susceptibility model construction and accuracy testing,then three machine learning models,SVM,LR and RF,are selected for probabilistic prediction of landslide susceptibility,and analysis the landslide susceptibility results.The results show that: the predictive performance of the three factor groups in each model,from high to low were the Improved factors,the Original add rainfall factors,the Original factors.The model performance from high to low is RF,SVM,LR.The prediction accuracy of RF is the highest among the three models,and the Improved factors-RF model is the optimal combination,which reflects the model performance extremely well.(3)By comparing the landslide susceptibility prediction results of Improved factors,Original add rainfall factors and Original factors,it is found that the prediction accuracy of the landslide susceptibility model is improved by the group of continuous environmental factors considering road density and gully density,compared to the group of discrete environmental factors considering distance from road and distance from river.Road and gully density are more suitable for landslide susceptibility modeling,which can effectively improve the accuracy of factor characterization,and it has a clearer physical meaning and is more closely related to landslides.(4)In the case of no rainfall,landslide stability shows a continuous decrease trend with the increase of slope excavation distance and angle.When the slope excavation distance is certain,the stability of landslide will show a convex decrease trend with the increase of excavation angle.When the slope excavation angle is certain,with the increase of excavation distance landslide stability shows a uniform decrease trend.(5)Under heavy rainfall conditions,the decline in stability of landslides is greater than that under conditions of no rainfall.The landslide stability also shows a decreasing trend of rapid first and then slow under the rising and then falling rainfall conditions.In addition,the landslide stability tends to decrease rapidly under heavy rainfall conditions as the distance and angle of the slope excavation increases.(6)Under the condition of no rainfall,with the increasing excavation distance of the slope,the rate of change of the stability coefficient of landslide is also increasing.When the excavation distance is certain,the greater the excavation angle,the rate of change of stability shows a decreasing trend.The rate of change of landslide stability coefficient is positively correlated with the rainfall intensity under the strong rainfall condition of rising first and then falling second,that is the greater the rainfall intensity,the greater the rate of change of stability coefficient.(7)Under continuous rainfall conditions,the pore water pressure along the potential slip surface will be redistributed,and the pore water pressure is also increasing.The relationship between the pore water pressure and the stability of the landslide presents an obvious reverse correspondence.