Research On Continental-scale InSAR Surface Deformation Monitoring Method

The rapid expansion of China’s urban agglomerations and the continuous advancement of urbanization have consumed some underground resources such as groundwater,coal,oil,natural gas,minerals,etc.,making the surface of urban in an unstable state,resulting in continuous,periodic,complex Sexual subsidence.In addition,China has a vast territory with complex and diverse geological structure,topography and landforms and there are three-level topographic steps.Various geological disasters such as landslides,debris flows,earthquakes,and glacier movements with surface deformation.Urban surface subsidence and other geological disasters threaten people’s personal safety and property safety,bring challenges to social security management,and bring huge obstacles to the sustainable development of cities.Therefore,normalized and intelligent monitoring of surface deformation information under large spatial scales is a necessary prerequisite and basic requirement for evaluating and preventing surface subsidence and geological hazards.In the past two decades,InSAR technology has received attention and InSAR technology has continued development.It has rapidly transitioned from traditional DInSAR technology to time-series InSAR technology,realizing long-term,highprecision deformation monitoring of interested areas and provides technical support for continental-scale deformation monitoring.The development of the field and direction of SAR-related technologies has also promoted the rapid development of SAR satellites and imaging technologies,and multi-constellation and multi-mode have become the development trend.The free opening of Sentinel-1 A/B wide-format data to users in2014 marked the arrival of the era of big SAR data,providing data support for continental-scale surface deformation monitoring.In recent years,scholars from various countries have done a lot of work in the application of continental-scale surface deformation monitoring,making contributions to continental-scale surface deformation.However,with the expansion of the monitoring range and the increase in the amount of processed data,there are still some problems to be solved in the monitoring of surface deformation at the continental scale.This paper focuses on the problems existing in the InSAR processing of Sentinel-1 data at the continental scale,in order to improve the ability and reliability of large-scale surface deformation monitoring.The main research contents and innovative work of this paper are as follows:(1)In this paper,an adaptive distributed coherent scatterer InSAR technology is proposed based on the extraction and joint processing of coherent pixels and distributed pixels,which solves the problem of insufficient number of targets for high-precision surface parameter inversion by coherent pixel technology in non-urban areas.This method realizes high-precision and fast inversion of surface deformation under large spatial scale and various surface scattering characteristics,and provides an idea for multi-view fast processing of distributed scatterers.Firstly,the processing method of traditional coherent pixel technology is described,and the limitations of coherent pixel technology and distributed scatterer InSAR technology are listed out through experiments.Based on its limitations,an adaptive distributed coherent scatterer InSAR technology is proposed,which extracts candidate distributed pixels(Distributed Scattering Pixel,DSP)under multi-view and homogeneous pixel clusters of candidate pixels through a double-layer detection window,and then estimate coherence on distributed pixels according to homogeneous pixel clusters.During phase optimization,the interferometric phase result of the maximum likelihood estimator based on eigendecomposition is used as the initial value of the phase triangulation algorithm(PTA)phase estimation method to improve the speed and accuracy of phase estimation.Finally,the DSP and coherent pixels are combined to invert the surface deformation parameters by network construction,so as to increase the monitoring points and improve the scope of application of the algorithm.The mountainous area of Southwest China and the local area of Shigatse,Qinghai-Tibet Plateau were selected as experimental areas.The results show that the density of monitoring points has increased by about 5-10 times,the phase quality has been improved,and the root mean square error between the InSAR results of adaptive distributed coherent scatterers and the leveling data has been(RMSE)is 5.45mm/year,and the error is about 9.2% of the maximum observation,which verifies the validity,applicability and reliability of ACDP-InSAR technology.(2)Based on Sentinel-1 data features,this paper proposes a method for nonlinear deformation extraction at large spatial scales and proposes a progressive distributed coherent scatterer InSAR technology based on the idea of sequential adjustment.It solves the problem of low computational efficiency caused by the need to rebuild a new SAR stack for processing after adding new SAR data.It realizes the rapid processing of new data and provides technical support for the needs of near real-time surface deformation monitoring at large spatial scales.As an important part of the fast processing of new data in time series InSAR technology,nonlinear deformation extraction needs to be studied.Based on the data characteristics of Sentinel-1 data,a large-scale nonlinear surface deformation sequence inversion method suitable for Sentinel-1 data is proposed.The important steps of the algorithm,such as discrete coherent target interpolation,spatial phase unwrapping,trend phase estimation,phase reference correction,weighted singular value decomposition,and atmospheric phase estimation methods have been deeply studied and algorithms is compared and verified by actual Sentinel-1 data processing.The idea of sequential adjustment is briefly described.Based on the summary of sequential estimator and progressive SBAS processing technology,a progressive distributed coherent target scatterer time-series InSAR processing method based on sequential idea is proposed.The time-series InSAR technology is divided into two processing modules:linear deformation calculation and deformation sequence calculation.The methods for fast processing of newly added data in these two modules are respectively studied,and finally a set suitable for adaptive ACDP-InSAR is formed.Then new data processing method is compared and analyzed by Sentinel-1 data.The research shows that the root mean square error of progressive ACDP and ACDP is ±3.5mm,and the processing speed of new data is 4.5 times that of ACDP,which proves the effectiveness of the proposed algorithm.(3)In this paper,based on the establishment of a double-layer triangulation network between the corresponding points in the overlapping area of adjacent Sentinel-1 frames,the problem that the coordinate system of adjacent SAR frames and the InSAR datum are not unified is solved.The network solves the problem of inconsistent InSAR datums of multiple SAR maps at the continental scale,preliminarily realizes the control of the InSAR datum at the continental scale,and provides technical support for InSAR post-processing at the continental scale to improve the reliability of InSAR results.The problems existing in the method and the necessity of improvement are explained through the combing of the traditional mosaic methods.Geocoding of InSAR results based on polynomial fitting of coordinate transformation parameters.Based on the verification of geocoding accuracy,the method of searching for the corresponding points is determined,and according to the characteristics of the data,a fast search scheme for the corresponding points in the overlapping area of Sentinel-1 data is proposed.After obtaining corresponding points,construct a three-dimensional doublelayer triangulation network for corresponding points.In the horizontal network,the geometric reference correction can be realized and the reliability of the solution results can be checked;the InSAR reference correction value of the adjacent Frame can be calculated in the vertical network.Making China as an example,369 frames in China were numbered and spatially analyzed.In order to prevent the transmission of errors during the datum correction process,the national frames were divided into multiple datum unified areas with main datum based on the spatial regional relationship.A triangular network is established for the national datum control area,and the overall adjustment value of each datum control area is obtained through weighted least squares to correct,resulting in control the InSAR datum at the national scale.(4)This paper solves the problem of human intervention in the process of time series InSAR based on data statistics,building a retrieval database,and adaptive parameter extraction,and realizes the unsupervised automatic processing of time series InSAR,which provides convenience for continental-scale surface deformation monitoring.The Sentinel-1 data within China in 2020 was processed using the supercomputing platform of the Earth Big Data Project of the Chinese Academy of Sciences,and the risk and impact of land surface subsidence across the country were assessed based on spatial analysis of multi-source spatial data.Based on the analysis of the key steps of human-computer interaction in timeseries InSAR processing,the automatic processing of InSAR processing is realized by data statistics,building retrieval database,and self-adaptive parameter extraction.The SAR data is automatically transferred from the storage space to the processing space by establishing a data retrieval table;the final interferometric combination is obtained by adaptively setting the parameters of the interferometric pair;the appropriate coherent target is selected by the statistics of the average coherence coefficient;the edge weight threshold is set through the network edge weight statistics;the solution reference point is automatically selected through the automatic search method;the quality of the stage results is checked through the data statistics.Methods guarantees fully unsupervised,robust processing of incoming data to accommodate the high complexity of the processing chain.The InSAR processing module is deployed on the supercomputing system of the Earth Big Data Platform of the Chinese Academy of Sciences to realize the InSAR automatic processing of the national Sentinel-1 data in 2020.After postprocessing the calculated results,collect multi-source spatial data to statistics and analysis national subsidence.