Research On The Monitoring And Inversion Of Different-scale Complex Surface Deformation With Multi-temporal InSAR

As a new earth observation technique,Interferometric Synthetic Aperture Radar(InSAR)has been widely applied in various surface deformation monitoring due to its large space coverage,high resolution and high precision in recent years.There are also several technical limitations,such as spatial and/or temporal decorrelation,DEM error and atmospheric delay,which make it difficult to obtain reliable deformation results.To overcome these errors,some advanced multiple-temporal InSAR techniques emerged,e.g.PS-InSAR and SBAS-InSAR,which mainly focus on the point targets with stable scattering characteristics in the long sequence of SAR images.These techniques not only improve the accuracy of InSAR-derived deformation,but also make revolutionary change in the development and application of InSAR.However,many technical problems,e.g.the rank defect caused by multiple independent subsets in conventional SBAS,effective recognition of high-coherent targets in low coherent condition and large-scale deformation monitoring with InSAR and the corresponding precision verification,still need to be solved in order to obtain high-precision results.Besides,multi-source factors,such as faults movement,magmatic activity,groundwater pumping,underground mining,landslide and ground fissure,may affect the surface deformation simultaneously in many areas,but the superimposed complex deformation can only be obtained from InSAR observations.Taking the complex surface deformation over Taiyuan basin as an example,this paper will study the key techniques in the deformation monitoring with high precision,large coverage and separate the different-scale deformations,such as land subsidence and fault movement,combined with the geophysical model using multi-temporal and multi-sensor SAR images,which will extend the breadth and depth of the application of multi-temproal InSAR.The main researches in this paper are as follows:(1)The rationale and characteristics of conventional D-InSAR and different types of multi-temporal InSAR have been analyzed and summarized systematically.Based on this,we suggest that many factors such as the features of research object,the geology and geomorphology,the quality and quantity of SAR images should be considered in the actual deformation monitoring with multi-temporal InSAR,which will provide a reference for the method selection in the subsequent research.(2)Based on the deformation model of conventional SBAS-InSAR,we firstly analyze the shortcoming of singular value decomposition(SVD)when the interfermetric combinations are separated into multiple subsets and deduce the corresponding deformation datum.Then,we put forward a constrained SBAS-InSAR model,which takes the detected deformation period as the constraint condition to inverse the time series deformation.The simulation data and real Envisat ASAR data in Southern California has verified the proposed model can be effectively solve the datum bias caused by multiple subsets to obtain more real time series.(3)Considering the advantages of PS-InSAR and SBAS-InSAR,we mainly focus on the research on the Interferometric Point Target Analysis(IPTA)technique with short baseline and propose a coherent point target detection method with multiple thresholds based on the amplitude deviation,coherence coefficient threshold and spectral correlation among sublook SAR images,which can effectively solve the dependence on the quantity of SAR images in the conventional PS-InSAR.We also propose a suitable nonlinear deformation estimation method.To overcome the limitations of computational efficiency and error propagation,we proposed the patch-wise InSAR method and solved the datum problem of the different InSAR results.To mosaic the multiple InSAR results obtained from adjacent tracks,the coordinate system transformation and the offset compensation have been solved.The redundant observations in the adjacent tracks have been compared to cross-validate the accuracy of InSAR measurements.The result indicated that the RMSE is less than 5.0 mm/a in Taiyuan basin.(4)We studied the spatiotemporal variations in the surface deformation over Taiyuan using IPTA method with short baseline and conduct the correlation analysis between land subsidence and groundwater change.According to the principle of effective stress,the storativity of aquifer system over Taiyuan has been calculated and the spatial distribution is aslo presented.The results show that the deformation corresponds with the locations of groundwater pumping well.(5)In order to overcome the limitation that one-dimensional deformation in the line-of-sight(LOS)direction can hardly reveal the real surface deformation,we studied the two-dimensional time series model based on the ascending and descending SAR images and applied the method to obtain the 2-D deformation results.Based on the uniform elastic half-space rectangular dislocation model,we put forward a simplified model,which is suitable for the inversion of the deformation induced by groundwater pumping,and then separated the deformation contribution of land subsidence and fault movement.The observed fault movement is 5mm/a,which has a good consistency with the previous data.(6)To solve the problem that the distribution of high coherent point target is uneven and conventional deformation monitoring method can hardly obtain reliable results,we proposed a refined PS-InSAR method by considering the spatial correlation of adjacent high coherent point target.Simulated and real data show that PS baseline network model based on Delaunay triangulation is superior to conventional GAMMA IPT method and is more suitable for the linear engineering deformation monitoring,e.g.high-speed railway.