Evaluation Of Shallow Landslide Susceptibility In Southern Shaanxi And Rainfall Warning Model Study

This paper takes shallow landslides in southern Shaanxi Province as the research object,conducts field surveys and collects historical landslide data.Analyzing the sensitivity relationship between landslide hazards and various influencing factors,and using correlation analysis establishes a landslide hazard susceptibility evaluation index system in the region.Using Geographic Information System(GIS),the evaluation and zoning of shallow landslide susceptibility in southern Shaanxi was completed by using information volume model and logistic regression model.On this basis,164 cases of rainfall landslide events were screened from 540 landslide reports in 2021.The rainfall data of each landslide event were identified in the Shaanxi Provincial Geological Hazard Monitoring and Warning Information System,and the effective rainfall intensity parameters of each landslide event were calculated using the effective rainfall model.Based on the I-D power law curve of rainfall intensity(I)and critical rainfall duration(D),the effective rainfall intensity threshold curve for each susceptible area is derived.Finally,a rapid evaluation method of slope resilience in southern Shaanxi counties is proposed based on the resilience rainfall model.The main work and results of this paper are as follows:(1)A total of 12 evaluation factors were collected from Hanzhong,Ankang and Shangluo cities in southern Shaanxi.Covariance analysis of evaluation factors,correlation analysis of landslides and each evaluation factor were conducted to screen out seven evaluation factors of slope,lithology,undulation,road buffer zone,plane curvature,elevation and rainfall,and finally an evaluation index system of landslide disaster susceptibility in the region was established.(2)Using geographic information system(GIS),the informativeness model and logistic regression model were used to evaluate and partition the susceptibility of shallow landslides in southern Shaanxi.The accuracy of both models was checked by using ROC curves,and the correct rates of both models exceeded 85%,indicating that the partitioning results were more reliable.Among them,the results of the informative model are better.For the informative model,the ratio of the area of high,medium and low susceptibility zones to the total area of the region are 6.2%,45.1% and 48.7%,respectively.Overall,the area of high susceptibility zones is larger in three places,namely,Hanbin District,Ziyang County and Shiquan County,which is mainly related to the local topography,geological lithology and rainfall.(3)Using the correlation between the amount of rainfall and the number of landslides.The attenuation coefficient of the effective rainfall model was determined to be 0.8,and the number of effective rainfall days was determined by combining the landslide displacement data,and finally the rainfall parameters of 164 landslide events were calculated by using the effective rainfall model.(4)The power-law model was selected as the rainfall threshold model.And three calculation models were established with the end time on the day of landslide occurrence,the end time on the day before the landslide occurrence and the end time when the landslide started to accelerate sliding,and the I-D power-law curves of different susceptible areas were calculated respectively.After verification by examples,all three models have relatively good results.Based on the susceptibility zoning and rainfall threshold models,the warning level of shallow landslide hazard in the study area and the corresponding emergency measures are proposed.(5)The resilience rainfall model is further used to make a rapid assessment of the resilience of individual slopes and to predict the possibility of instability damage of individual slopes under rainfall conditions,with good results.Innovations of this paper:(1)In order to select rainfall intensity parameters more scientifically,when making the determination of the number of days of effective rainfall,this study combines landslide displacement data for the first time to determine the number of days of rainfall.The final attenuation coefficient is determined by calculating the correlation between rainfall and landslide number under different attenuation coefficients.(2)In order to realize early warning,when establishing the rainfall threshold model,for the first time,the cut-off time is one day before the occurrence of landslide and the cut-off time is when the landslide starts to accelerate sliding to establish the threshold model.The regional rainfall warning power law model established in this study can provide a reference for the research of rainfall warning models in other regions.