Identification And Hazard Assessment Of Ancient Landslides Based On Multi-source Remote Sensing Technology

With the aggravation of human engineering activities(such as railway,highway,hydropower project,etc.)and extreme weather in the eastern margin of the Qinghai Tibet Plateau,the reactivate of ancient landslides is on a sharp rise,and resulting in increasingly serious casualties and economic losses.However,due to the large scale of ancient landslides,the complex disaster background,and often suffer from later transformation or sediment cover,the original appearance characteristics of the landslides have been blurred,most of the ancient landslides are still in an unknown state.Therefore,the problem of ancient landslide and its reactivation has always been a prominent problem that perplexes the planning and construction of major projects and the safety of cities and towns.In recent years,multi-source remote sensing information technology is more and more widely used in landslide disaster research.Among them,Object-Oriented analysis(OBIA)and Synthetic Aperture Radar Interferometry(InSAR)are both efficient technologies for landslide identification and monitoring.However,how to use OBIA and InSAR technology to identify ancient landslides in complex and dangerous mountainous areas and how to effectively improve the accuracy and timeliness of landslides monitoring and early warning and risk assessment is still an urgent problem to be explored and solved.This paper focuses on the key scientific issues related to the ancient landslide and its reactivation in Dadu River Basin on the eastern edge of the Qinghai Tibet Plateau.On the basis of field investigation,multi-source remote sensing data and a variety of data processing methods are used to carry out the research on the comprehensive identification and development law of the ancient landslide.Combined with InSAR surface deformation monitoring,the early recognition research of ancient landslide reactivation is carried out,and the criterion of ancient landslide reactivation based on InSAR technology is proposed,which reveals the time-space characteristics and influencing factors of reactivation deformation of typical ancient landslides;based on the results of time-series InSAR monitoring,the dynamic risk assessment of landslide in the study area is completed.The main achievements are as follows:(1)Using high-resolution remote sensing image,the landslide information extraction and landslide image classification attribute feature discriminant analysis of Dadu River Basin are carried out.It is found that there are significant differences between the ancient landslide and the new landslide in the hue,spectrum and texture features of remote sensing image,and based on vegetation index(NDVI),brightness,terrain roughness(TRI)and gray level co-occurrence matrix entropy(GLCM),A comprehensive remote sensing identification model(GTVI)for ancient landslides is proposed.By using time series InSAR Technology(PS-InSAR and SBAS-InSAR)and D-InSAR technology to monitor the surface deformation in the study area,combined with field verification,694 landslides formed 100 years ago were identified,including 146 landslides with a volume greater than 100×10~4m~3.(2)The ancient landslides are mainly distributed along both sides of the main stream of Dadu River and Xiaojin river.There are many landslides within 5km on both sides of the fault,and With the distance from the river and the fault getting far away,the number of landslides decreases.In terms of time,the formation and evolution of ancient landslides have a good corresponding relationship with river terraces,which is significantly affected by climate change in the region.The time span is from 10ka to 40ka,concentrated in 15 to 30ka,accounting for 57.6%of the total.Neotectonic activities and earthquakes cause differences in different sections.And near the strongly active faults,ancient landslides are intensively developed.(3)On the basis of InSAR monitoring results,the displacement and velocity characteristics of typical ancient landslide reactivation are analyzed quantitatively.Based on Saito principle,the three-stage evolution process of ancient landslide reactivation is proposed;Combined with the accumulated displacement-time curve monitored by PS-InSAR,the inflection point between elastic deformation and uniform deformation is proposed as the starting point of the reactivation of ancient landslides,and the inflection point from uniform deformation to accelerated deformation is taken as the sign of the reactivation and instability of ancient landslides.Based on this,the velocity threshold for the early recognition(instability)of the reactivation of typical ancient landslides in Dadu River Basin is proposed,and it provides a new way for the early recognition of the reactivation of hidden ancient landslides in mountain and canyon areas.(4)Based on a variety of InSAR processing techniques,a demonstration study on the early recognition of the reactivation of typical ancient landslides is carried out.Using Sentinel1 and ALOS2 data,the surface displacement of Gezong ancient landslide and Jiaju ancient landslide are monitored,and the surface deformation results of millimeter scale from January 2016 to March 2018 are obtained.Through displacement analysis,surface creep characteristics analysis and landslide boundary identification,the evolution stage of ancient landslide reactivation is divided.Based on the daily rainfall data,the paper puts forward the deformation zoning,accumulation characteristics and time-space deformation model of the resurrected surface of typical ancient landslides.(5)Based on the systematic analysis of the relationship between the landslide and the influencing factors such as slope,slope direction,roughness and fracture zone,the information method(IV)and the logistic regression model(LR)were used to evaluate the vulnerability of the landslide;and the matrix model of dynamic assessment of landslide risk is established based on time-series InSAR technology,and the dynamic assessment of landslide risk in Dadu River Basin is completed.The evaluation results show that after integrating the InSAR displacement rate results for risk evaluation,the real-time evaluation of the regional slope deformation state can be carried out,and the error of landslide risk evaluation can be effectively reduced,and the accuracy and effectiveness are better through field investigation verification.The research ideas,technical methods and relevant research results in this paper have certain guiding role and reference significance for the identification of ancient landslides,the early recognition of ancient landslides reactivation and the monitoring and early warning of regional geological disasters in the complex and dangerous mountainous areas of the eastern margin of the Qinghai Tibet Plateau.