| China has a vast territory,with almost all the geomorphic types on the earth,so that the types and number of geological disasters in China are the largest in the world.In the past ten years,the frequency of landslide disasters has accounted for more than half of the total number of geological disasters in our country.Due to the high frequency of occurrence,wide occurrence range,fast occurrence speed and high prediction difficulty,landslide disasters have become the biggest threat to human beings except for the earthquake.At present,most of the landslides in my country are not cataloged.Once they occur,they will inevitably cause irreversible consequences.Conventional identification methods are time-consuming and labor-intensive,and there is often a phenomenon that "the tester did not slip and the slipper did not test".Therefore,how to quickly and accurately identify the location of landslides and monitoring is an urgent problem to be solved,this paper takes the junction area between Tsingza County and Hualong County as the research area,and uses timing InSAR technology to carry out landslide hazard identification and deformation monitoring research.The main research content and conclusions are as follows:(1)Data visibility analysis and establishment of quasi-three-dimensional deformation decomposition model of landslide.Based on the visibility analysis of the slope,aspect and satellite parameters of the study area,it can be seen that the effective observation area of the ascending and descending orbit data in the study area accounts for more than 97% according to the results of the visibility zoning,and the available ascending and descending orbit data in the study area Joint monitoring.Based on the analysis of the spatial geometric relationship between the line-of-sight(LOS direction)deformation of the ascending and descending orbits and the real deformation,the quasi-three-dimensional deformation of the landslide is obtained by combining the slope and direction.(2)The landslide identification of large areas is carried out by Stacking-InSAR technology.This paper uses 20mm/a as the deformation rate threshold combined with optical images to identify the hidden danger areas of landslides,and finally 39 landslide areas are identified in the ascending and descending orbits data.Among them,11 areas identified the same landslide area,13 areas were not detected in the descending orbit 135,and 3 areas were not detected in the ascending orbit128.It is possible to use the ascending and descending orbits data to mutually verify and supplement the landslide recognition results,thereby improving the accuracy of landslide recognition and avoiding missed judgments.(3)A method based on deep learning to realize automatic identification of landslides is proposed.Using the phase change map obtained by InSAR as the training sample,a landslide recognition model was obtained and verified through U-net model training,achieving the purpose of automatically identifying the hidden danger area of landslides.(4)According to the identification results of the landslide area,two typical landslide bodies are selected for key analysis.According to the three-dimensional decomposition model of the landslide,the vertical and horizontal deformation results of the landslide are extracted to analyze the spatial deformation characteristics of the landslide;the time series deformation of the landslide is obtained using the IPTA technology to analyze its time deformation characteristics.(5)The deformation law of landslides is discussed.The results show that there is a positive correlation between the increase in landslide deformation and rainfall,and temperature is also a factor affecting landslide deformation. |
PDF Full Text Request