Key Issues Study Of InSAR Topography Reconstruction And Deformation Monitoring Based On TerraSAR-X Data

Interferometric Synthetic Aperture Radar (InSAR) is a new type of space-to-earth observationtechnology, which is developed on the basis of SAR in the late1990s. It utilizes the SARphase information to obtain accurate topographic and deformation information in various timespan. With the unique advantages of monitoring tiny surface deformation with its all-weather,day-and-night, broad-covering, and high degree of automation monitoring grounddeformation capacity, InSAR technique has become a wonderful tool to monitor geologicaldisaster. The SAR satellites of European Space Agency(ESA) and Japanese Space Agencystopped running one after another. Also with the constantly improvement requirements ongeological disaster monitoring accuracy and reliability, and the more and more subtle objects,the high resolution radar satellite was proven to be an inevitable trend. Germany’s highresolution radar satellite TerraSAR-X launched successfully in2007, that marked thesatellite-borne SAR study entering a new era for their short revisit time, high spatialresolution, short wavelength, multi-polarization, multi-mode and high precision deformationmonitoring ability. Therefore it will be provided with strong practical significance andscientific value to do a research on constructing high precision and resolution digital elevationmodel(DEM) and monitoring deformation especially for complicated geo-hazards.As for the more short wavelengths and higher resolution, TerraSAR-X data processing notonly have to face the common track error, external DEM error and the atmosphere error, butalso is confronted with the more serious time and baseline decorrelation, phase unwrappingproblems of large deformation gradient, the DEM error introduced by city high-rise building,as well as the atmospheric error. The errors all above are undoubtedly limiting accuracy andreliability of monitoring results.Based on the above problems, this paper comprehensively study the high resolution DEMreconstruction and high precision deformation monitoring from TerraSAR-X data processingof the main error according to TerraSAR-X data characteristics. The main research contentsand results are as follows:1. The paper analyzes the main errors and weakened measures in TerraSAR-X DEMreconstruction, talks about in detail the influence of the baseline error, deformation phaseerror, the phase unwrapping error. The phase unwrapping error is easy to appear in bigelevation gradient. The deformation effect may arise a hight deviation if the researched region is suffering surface deformation during the SAR data acquisition, which increases withdeformation value. Besides the random error such as atmospheric error restricted the accuracyof DEM. Based on these errors, this paper puts forward six key technical steps of InSARDEM reconstruction:①The control points (GCP) baseline estimation algorithm wascombined with precision orbit data to refine the perpendicular baseline, correcting thesystematic deviation caused by baseline error term.②Aiming at the deformation phase error,deformation model was established, which eliminated the elevation deviation aroused bysurface deformation during the SAR data acquisition.③The "remove and restore" phaseunwrapping method based on low precision external DEM was proposed, that reduced phasediscontinuous caused by phase unwrapping error in steep area, ensuring the veracity of DEM.④We discussed in detail the selection method of weight factor in Combinedmulti-interferogram algorithm, deduced and realized the weight factor determination methodbased on baseline information and coherence as well as the wavelet analysis considering theatmosphere factor. The multi-interferogram algorithm effectively weakened the influence ofatmospheric delay and phase noise and improved the precision of final InSAR DEM.⑤Against to a large number of null area in InSAR DEM, we realized a new kriginginterpolation method considering SRTM DEM, which adjusted the horizontal and verticalvariation functions through structure analysis results of known point, effectively filling thenull area.⑥We put forward an absolute calibration method based on GPS elevation data as avertical datums.The CR technology was firstly used to confirm the accurate position of GPSin the SAR coordinate. Then the least square fitting method calibrated InSAR elevation toWGS84datum. Based on the above key issues we established a thorough high resolutionDEM data processing flow under the complex environment. The new algorithm was tested inXi’an by six TerraSAR images, we finally got a6m resolution DEM in WGS84, whoseelevation precision is4m by comparing with the1:50000terrain data.2. This paper discusses the DEM error in high resolution InSAR deformation monitoring,introduces the characteristics of some external DEM, provides selection criteria of externalDEM in high resolution data processing. The deformation monitoring analysis in Xi’an cityshow that the TerraSAR DEM established by our new method reduced the DEM errorinfluence on deformation.3. Directing at the various error sources in TerraSAR-X deformation monitoring, we studiedthe key issues of monitoring ground subsidence and fissures disaster in Xi’an city and ShanxiQingxu using high coherent point target analysis technique. The high signal-to-noise ratio coherent point were selected by amplitude dispersion thresholding and interferometric phasestability analysis, which ensured the reliability of phase information. According to thetemporal and spatial correlation features of phase, we applied a high-pass filter in temporaldomain and a low pass filtering in spatial domain to interferometric phase. DEM error,atmosphere error and orbit error were evaluated and then removed, that guaranteed theaccuracy of phase unwrapping. The three-dimensional phase unwrapping was used to repairreal phase value, we eventually achieved the ground subsidence annual rate and timeseriesresults of Xi’an with a millimeter scales by comparing with Envisat monitoring resultsand GPS results. Sufficient density of pixels were identified by spatial coherence iterativeanalysis of interferometric phase in farmland area to overcome the decorrelation. We appliedpermanent scatter and small baseline subsets technique to Qingxu to obtain ground subsidenceannual rate, and two monitoring results were in good agreement. The regularity of groundsubsidence space distribution in ground fissures area were analyzed through profile analysis.