Research On The Theory And Algorithm Of Short Spatio-Temporal Baseline PS-DInSAR

In recent years, the urban subsidence issues have increasingly become prominent in the world and caused huge losses to people's lives and property. In view of this problem, the governments and agencies around the world have carried out settlement monitoring intensively. Leveling and GPS are considered as the primary detecting tools. With the development of radar technology, the differential synthetic aperture radar interferometry (DInSAR) has become to an important way of detecting ground deformation with high efficiency and wide coverage. However, its applications are largely limited due to spatio-temporal decorrelation and atmospheric artifacts. Therefore DInSAR based on permanent scatters (PS-DInSAR) is proposed to mitigate the negative influences.On the basis of investigating the principles of DInSAR, this thesis aims at improving the strategy and algorithm of data processing and analysis in conventional PS-DInSAR for decreasing the influences of both decorrelation and atmospheric artifacts. The research content and relevant conclusions presented in this thesis are outlined as follows.1. Interferometric combination is formed by considering both temporal- and spatial-baseline threshold. The experiment is performed with use of 39 ERS-1/2 C-band SAR images (acquired from 1992 to 2000) related to the Phoenix region of the United States. We set the thresholds as 4 years and 120 m, respectively, and obtain 86 effective interferograms. In comparison with the conventional approach of interferometric combination by sharing the unique master image, the time sampling rate of the former is increased significantly.2. The amplitude dispersion index thresholding algorithm is used to detect PS automatically, and thus connecting them by Delaunay triangulation. The linear deformation and DEM error can be extracted from PS networking model by searching within the solution domain.3. In view of free interferometric combination, the time series of residual phases which include nonlinear deformation and atmospheric delay can be reconstructed by singular value decomposition. The nonlinear deformation phases can be derived by first high-pass filtering in time and low-pass filtering in space and then removing atmospheric delay phases.4. The subsidence analysis over Phoenix by PS algorithm shows that the largest linear subsidence is up to 3.2 cm per year, and the accumulated subsidence is up to 33 cm in 10 years. There are two obvious settlement bowls in the study area. The derived results are in good agreement with the measurements obtained previously. This demonstrates that the short baseline PS-DInSAR can be efficiently applied to detect regional ground deformations.