Time-Series InSAR Technology And Its Application In Landslide Deformation Monitoring

China is one of the most serious countries threatened by geological disasters in the world.Every year,hundreds of people are killed by disasters and the economic loss amounts to 1 billion yuan.In the total amount of disaster,landslide disaster accounted for two-thirds.How to monitor landslide geological hazards with minimum cost and high efficiency has always been a concern of researchers.Differential Interferometric Synthetic Aperture Radar(DInSAR)is a microwave telemetry technology developed in the last ten years.Because of its advantages of non-contact,all-weather,high resolution,low manpower and material cost,it is favored by researchers in the field of geological disaster monitoring.However,this method is affected by time,space and atmospheric conditions,so its monitoring accuracy is limited.In order to overcome these limitations,time-series InSAR technology,such as Persistent Scatterer Interferometric Synthetic Aperture Rada(PS-InSAR),was developed,providing a new means for geological hazard monitoring.Starting from the basic concepts of InSAR and DInSAR,this paper studies the theories and methods related to the application of timing InSAR,analyzes the main components and causes of DInSAR differential phase,introduces the mainstream time-sequence InSAR technology and related software,details the theory and method of the open-source software DORIS for interferogram production and open-source time-sequence InSAR software StaMPS.On this basis,some major problems in the application of time-sequence InSAR technology are studied.A new generation Sentinel-1A interferometric wide-format SAR image was used to study the surface deformation characteristics of the large-giant landslide at the south side of Gaojiawan village,Hongshui town,Ledu district,Haidong city,Qinghai province around January 2016.The main research work and conclusions are as follows:(1)Using DORIS and StaMPS open source software,studied the characteristics of InSAR data processing platform under Linux system.Based on the "named pipeline" of Linux system,a simple and convenient method for parallel optimization of data processing of open source InSAR software is proposed.This method can run efficiently without much modification to the source program.Verified by the experiment,compared with traditional scene serialization processing method,by using this parallel method to time-sequence InSAR software DORIS-StaMPS,data processing efficiency reached 335.164% on average.At the same time,this parallelization method has strong portability and can be used for parallel optimization of open source timing InSAR data processing platforms such as GMTSAR to meet the needs of InSAR large data processing.(2)In view of the new generation TOPS imaging model,this paper studies the differences in the main image selection methods based on space-time baseline and doppler baseline with different time baseline proportions,When a single burst is used as the main image selection basis.Aiming at the problem that the main image selection is not unique when the single burst is the main image selection basis and the research area spans multiple adjacent burst,this paper proposes a main image selection method that integrates multiple burst baselines and considers the dominant part of the research area by taking Sentinel-1A interferometric wide-format SAR images as an example.This method takes into account both the need for high ESD registration and the need for high coherence in the research area in TOPS imaging mode,so that the main image selection is unique.By comparing the time-sequence InSAR data set composed of the optimal main image selected by this method and the main image selected by a single burst,the number of PS points selected in the research area was 2.1738% more than that in the same processing conditions.(3)For time series InSAR,different time series data sets can reflect different characteristics of surface deformation in the study area.The data set of long time series usually aggravates the temporal uncorrelation of SAR data set with the growth of time,leading to the decrease of available PS points in the research area,the loss of deformation details and the difficulty of unwrapping.However,the potential trend deformation can only be distinguished from the sparse PS points.Although it is difficult to process data in the short time series data set,especially the atmospheric delay error is not easy to eliminate,but under the error acceptable condition,the coherence of the study area is improved and the available PS points are increased,so as to retain the detailed characteristics of surface deformation as much as possible.In practical application,the different time series of the same data set should be comprehensively applied to the surface deformation of relatively complex surface,so as to obtain the deformation characteristics of different time scales in the study area.(4)In order to study the possible fault activity and surface deformation characteristics of gaojiawan landslide body,different time series data sets were divided from the lifting rail data.The results show that there are two major faults in Gaojiawan landslide,and one of which intersects with Zhangjiazhuang tunnel of lanzhou-xinjiang high-speed railway.On January 18,2016,the fault above Gaojiazhuang tunnel started violent movement,which caused the uneven opening of the upper and lower plates by about 13 mm,and led to the cracking of the arch roof of the tunnel below.The slope near the south side of Gaojiawan village is protected by the south side of the mountain beam,which blocks the sliding pressure from the mountaintop fault.The mechanical uplift was earlier than the large deformation time of the fault,and the maximum uplift occurred on January 18,2016.Up to the SAR data cut-off time in this paper,the Gaojiawan landslide body as a whole is still in an unstable state of peristaltic slide.