Monitoring And Susceptibility Evaluation Of Loess Landslide Based On SBAS-InSAR

This article uses SBAS-In SAR technology to target the loess areas of Shaanxi and Gansu,collects SAR data from March 17,2017 to June 29,2020 to carry out time series surface deformation monitoring,analyzes the monitoring results and uses logic for the susceptibility of landslides Regression model prediction.The following are the results of this research:1.Calculated the surface deformation rate and cumulative surface deformation of the Shaanxi-Gansu loess region from March 17,2017 to June 29,2020,and divided the landslide disaster points.This paper uses the surface deformation rate of-25mm/y and the cumulative deformation of-50 mm as the threshold of deformation accumulation area,and divides the area with deformation less than this threshold into deformation accumulation area.The deformation gathering areas that meet the deformation threshold,landslide characteristics and threaten objects are divided into landslide disaster points.A total of 419 landslides were monitored in the experiment.Among them,there are 359 old landslides,54 new landslides,2 slopes and 4 collapses.2.Three typical landslides are selected and their deformation laws are analyzed.Obtain the cumulative deformation of the three landslides and draw a time-series broken line chart,calculate the deformation speed at the sudden change of deformation,and compare the speed difference analysis in different regions,and it can be found that the three landslides conform to the traction landslide,the sliding landslide and the landslide with special changes..Combining the human and natural environmental changes in the corresponding area at the same time,it is concluded that human activities and sudden rainstorms are the main reasons for the sudden change of landslide settlement.3.Introduce 8 hazard factors including distance from highway,distance from river,slope,aspect,elevation,curvature,normalized water index(MNDWI),and normalized vegetation index(NDVI),and further analyze them Judgment eliminates non-important and redundant factors.Calculate the information value of each influencing factor through information entropy,use 0.6 as the threshold to filter out the susceptibility evaluation index factors with a large contribution rate of information content to the landslide,and exclude the distance to the river;then calculate the remaining 6 through Pearson correlation For the correlation coefficient between the impact factors,in the impact factor combination with the correlation coefficient greater than 0.6,the factor with the larger information entropy is retained,so the MNDWI is retained and the NDVI is eliminated.4.Use logistic regression model to predict landslide susceptibility.214 landslides and their six susceptibility index factors were counted,and 200 non-landslide data were randomly generated to form the sample point data,80% of which were used as the training set of the logistic regression model,and 20% were used as the test set.Calculating the contribution rate of the landslide susceptibility model for each evaluation index is the undetermined coefficient,and it is concluded that the contribution rate of the normalized water body index is the highest,reaching 18.19,which is the most important index factor affecting the occurrence of landslides.The accuracy of the landslide prediction rate of the logistic regression model was tested by the receiver operating curve(ROC)and reached a high prediction rate of 0.855.