Study Of Quality Control On InSAR Time Series Monitoring And Application

In recent years, with the new type of high resolution SAR satellites having beenlaunched one after another and the continues progress and development of the time-seriestechnology, interferometric synthetic aperture radar (InSAR) technology, which possessessome prominent advantages, has been widely used in the survey and monitoring of geologicaldisasters, such as earthquake, ground subsidence, landslide and debris flow. Moreover, InSARtechnology,which can provide a new way for the dynamic study of the geophysics andgeodesy research, has become a new space-geodetic technique with its unique and potentialadvantages.However, the InSAR measurements are often affected by a number of errors such asatmosphere error, DEM error, orbital error, unwrapping error and decorrelation noises, andthese errors tend to have natures of multi-source, complexity, interactive and a variety ofuncertainties, etc.. The combined effects of these random, systematic and gross errors mayseriously influence the accuracy and reliability of InSAR measurements, and thus make themimpossible to reach the theoretical accuracy of millimetre to centimetre, which seriouslyrestricted the further promotion and application of InSAR technology. So there is anincreasing demand for assessing and controlling the quality of InSAR measurements, as avariety of error existing. The assessment and improvement of the quality of InSARdeformation measurements is also an important prerequisite for the measurementspost-processing and the deformation mechanism inversion. Based on these, with a statisticalanalysis, this paper focuses on the assessment and control of the quality of InSARdeformation measurements, and gives the control method of the corresponding errors, byestablishing reliable mathematical models from the geodesy theory. The effectivelyeliminating of these errors is a guarantee for the high precision and high reliability acquisitionof the InSAR measurements.Through the study, this paper obtained the following innovative results:1) For the phase unwrapping error of InSAR measurements, this paper first studies theInSAR phase unwrapping methods, and then proposes a novel phase reconstruction methodbased on the moving window multi-quadric function to reduce the phase unwrapping error.The multi-quadric function method ensures the continuity of the unwrapped phase, while themoving window method maintains the local details of the phase. When modeling, Afitting-nodes determination method for InSAR phase data based on both the phase feature and coherence is given. And finally, a F-test method is followed to analyze the significance of thenew model.2) For the existence of orbital residual error in the InSAR data, this paper proposes arobust least squares method based on wavelets to fit and eliminate orbital residual errors,based on the linear fitting method. The wavelet multi-resolution analysis can effectivelydistinguish orbital error with other components, such as deformation signal, atmosphericdelay and other errors, whereas the robust iterative weighted least squares method can makethe orbital ramp estimation more reliable. Finally, the accuracy and reliability of the algorithmwere verified by both simulation data and real data in Xi’an area.3) For the problem of spatial covariance and atmosphere delay, DEM error estimation inInSAR time series processing, this paper, based on the study of Small BAseline Subset (SBAS)method and Multiscale InSAR Time Series (MInTS) method, gives an integrated InSAR timeseries algorithm combining MInTS and SBAS (MInTS-SBAS). This algorithm can not onlyeffectively solve the problem of spatial covariance in InSAR measurements, but also canisolate atmosphere delay and DEM error in the processing of InSAR time series. Through realdata of Xi’an area, the studies have shown that the MInTS-SBAS algorithm proposed in thispaper can effectively improve the accuracy of InSAR time series monitoring results and has abetter consistency and reliability with respect to the GPS and leveling results.4) For a large number of temporal decorrelation noise existing in InSAR time seriesprocessing, this paper gives an InSAR time series error analysis method based on Kalmanfilter. Studies have shown that the Kalman filter can not only effectively eliminate thetemporal noise of InSAR deformation time series, but also can get an optimized value of thelinear deformation rate.5) A major problem in inversion of deformation mechanism using InSAR data isthat the InSAR results often contain thousands to millions of data points. Furthermore, therealways exist errors and even some blunders, which make the data inversion is lower efficientand lower reliable. For this, this paper proposed an adaptive quadtree decomposition methodfor InSAR data reduction by taking account of the physical spatial correlation of InSAR data.This algorithm can give a dense sampling in areas where deformation gradients are high butfew where slow changes are observed. So, in the condition of preserving details ofdeformation as much as possible, this algorithm can achieve the objectives of both an efficientdata reduction and noise elimination.6)With a thoroughly analyzes of the hazard-formative factors of ground fissures inYuncheng City, this paper built, respectively, the sensitivity assessment model based on the analytical hierarchy process (AHP) method and the activity prediction model by BP neuralnetwork for ground fissures, which will play a directive role in the prevention and mitigationof ground fissures disaster in Yuncheng area.