| Due to the influence of the Indian plate subduction and compression on the Eurasian continent,the western Sichuan Province is adjacent to the world’s highest altitude plateau,the Qinghai-Tibet Plateau,with steep terrain and frequent geological disasters,especially in the southwestern mountainous areas.Under the combination of multiple factors such as large elevation differences,complex geological structures,and river erosion,and disturbances such as precipitation and human engineering activities,landslide disasters frequently occur.This article aims to explore a comprehensive method for identifying landslide hazards in the southwestern mountainous areas of Sichuan Province based on the advantages of InSAR and remote sensing technology,which can overcome the difficulties of traditional ground identification and investigation of landslide development points.By constructing InSAR deformation indicators and landslide optical interpretation indicators,and combining with the most concerned factors of landslide hazards,such as threatened objects,a comprehensive method for identifying landslide hazards based on multi-source remote sensing data was proposed to achieve accurate identification of landslide hazards in the southwestern mountainous areas.Information entropy model and multi-model coupling method were used to compare and evaluate the susceptibility of landslides in the study area,and a relatively suitable hazard assessment model was obtained.In order to be able to provide reference for landslide potential hazard identification and evaluation in other similar areas,and to provide a basis for decision making in geological disaster prevention and management and monitoring and early warning in Sichuan Province and even in the southwestern mountainous areas.The main content is as follows:(1)Determination of landslide InSAR deformation and optical remote sensing interpretation signs.Based on D-InSAR and SBAS-InSAR technologies,the average annual surface deformation rate in the study area is between-67.49-120.96 mm/y.Combining with the deformation breeding law of landslides,a landslide InSAR deformation sign is established based on the criteria of "color of the patch,deformation density,and deformation position",supplemented by a threshold of the absolute value of 10 mm/y of deformation velocity in the radar line-of-sight direction.Based on the texture and color difference of optical remote sensing images and the typical deformation mechanism of landslides,an optical remote sensing interpretation sign of landslides is established from the aspects of the back wall and the shape and structure of the landslide body.(2)Develop a comprehensive landslide hazard recognition method based on a control layer of threat objects and terrain slope,a base layer of InSAR deformation indicators and landslide optical interpretation indicators,breaking through the traditional landslide recognition method of "using only a single remote sensing data and other data as auxiliary".Construct a "slope-deformation rate-optical interpretation" landslide hazard recognition matrix model,changing the traditional landslide recognition process of "relying entirely on subjective experience to extract quantifiable indicators",improving recognition accuracy,and reducing missed detection rates.(3)Based on the comprehensive remote sensing recognition method mentioned above,experiments were conducted in the southwestern mountainous areas of Sichuan Province,and a total of 469 landslide hazards were identified.Among them,242 landslide hazards(52%)had InSAR deformation features,194 landslide hazards(41%)had optical morphology(deformation)features,and 33 landslide hazards(7%)had both InSAR deformation and optical morphology(deformation)features.Through field verification,all 469 identified landslide hazards were confirmed as actual landslide hazards,with a total of 437 confirmed cases,and the identification accuracy reached 92.96%.This proves the feasibility and reliability of the comprehensive remote sensing recognition method mentioned above.(4)Based on the InSAR time-series deformation results,two typical landslides in Fuyang Village,Panzhihua City and Pingweng Village,Muli County were selected for deformation analysis,and the results showed that the InSAR deformation can reveal the detailed deformation process of landslides in different time periods,and the accelerated deformation of Fuyang Village landslide is closely related to the summer rainfall.(5)Based on the comprehensive remote sensing identification results and the terrain and geological conditions of the study area,ten evaluation factors such as elevation,slope,aspect,and terrain relief were selected.Information value model(I),information value coupled analytic hierarchy process(I-AHP),and information value coupled logistic regression model(I-LR)were used to carry out landslide susceptibility assessment,and the evaluation results were verified through rationality tests and receiver operating characteristic(ROC)curves.The results show that:(1)The evaluation accuracy of the information value coupled logistic regression model(I-LR)is higher than that of other models,and it is suitable for landslide susceptibility assessment in the southwestern mountainous area of Sichuan Province.(2)The high-risk areas in the study area are distributed between elevations of 1000-3000 m and slopes of 15-30°,concentrated in cultivated land and construction land,and on slopes within 2000 m of roads and faults and within 1000 m of rivers.(3)The landslide hazards in the study area are mainly distributed in the western mountainous canyon area and the steep slope areas on both sides of the Jinsha River and Yalong River in the southwest,reflecting the greater susceptibility of landslides to river erosion and human engineering activities. |
