High-resolution Persistent Scatterer Radar Interferometry And Its Application To Monitoring Subsidence Of High Speed Railway

Geographical environment and geological condition greatly influence the construction of the high-speed railway, especially for the stability of the roadbed and bridge. However, land subsidence along and around the railway generally arised from thenatural or anthropic activities, such as excessive exploitation of groundwater. In addition, high-speed trains operation and rail vibration ususlly creat the pressure to the roadbed and bridge so as to result in the deformation, especially in vertical direction, i.e. the subsidence. As the above mentioned issues will directly affect the train operation safety, in this paper, an advanced InSAR technique termed PS InSAR (PSI) is implemented to monitor and further study the land subsidence of high-speed railway.Interferometric synthetic aperture radar (InSAR) is a common technique for the extraction of ground subsidence, especially suitable to large areas. However, the subsidence measurements derived by InSAR are often degraded by two major factors, i.e., spatial and temporal decorrelation and atmospheric artifacts, which often occur in the areas with slower subsidence, e.g. vegetated areas. To resolve the problem, PSI has been recently employed. However, two major problems for the technique are that the selection of PS needs numerous SAR images so as to result in high cost and the limitation for the detection of deformation in the area with few available SAR images.In order to improve the applicability and feasibility of PSI, an improved PS selection method, i.e. coherent point identification is proposed and implemented in this thesis. Since the selection of PS can be performed on a minimum of two SAR images, the shortage of great demand of SAR image sources in application can be overcomed. In addition, it is difficult to indentify available and adequate natural PS in the areas, e.g. vegetated or agriculture areas, this may result in failure in subsidence detection. In order to increase the PS density, the man-made reflector (CR), which is grouped into two parts, e.g. fixed and separated CR, is utilized to form the subsidence monitoring network by connecting the CR and natural PS.In this thesis, high resolution TerraSAR-X SAR images are utilized as test data. When applying the PSI techniques to such data, there is usually no external DEM with comparable resolution to effectively remove the topography related phase components. This may result in the estimation error of deformation parameter in case of using the low resolution DEM. Therefore, the interferograms with ultrashort baselines (USB) is proposed and used in this study. Based on the USB interferometric pairs, the PSI approach is characterized by several features in PS identification, PS networking and PS solution by a least squares (LS) estimator. As the USB-based interferograms are not sensitive to topographic effects, an external digital elevation model (DEM) is no longer required for differential processing so as to simplify the TCP modeling and decrease the number of parameters to be estimated in TCP solution.The USB-based PSI method will be validated with23high resolution TSX images acquired between29April2009and11November2010over Tianjin. In the experimental evaluation, a comparision between the subsidence measurements derived by the USB-based PS solution and those derived by the PS solution based on the full interferometric combinations is conducted and the data form the ground-based leveling measurements are used as benchmarks to validate the results by the proposed method. The final results indicated that the accuracy of subsidence rates and subsidence between leveling and the USB-based TCP solution are2.0mm/yr and2.6mm.10high resolution TSX images acquired between20August2008and30January2010over Shanghai are also utilized and further validate the proposed USB-based PS method. As there are no leveling measurement data in the area, a variance component estimation approach is adopted to assess the accuracy of deformation results. The results showed that the discrepancy of minimum PS subsidence rates between priori variance components and estimated variance components is-8.88mm/yr, and the discrepancy of maximum rates is0.79mm/yr. the precision of the subsidence rates is improved2.45mm after the implementation of the VCE. It is denoted that the VC-matrix of the estimated subsidence rates is more realistic and can be used to evaluate the precision of the results.In contrast to traditional subsidence monitoring approaches, the new approach can significantly improve the precision and efficiency of deformation monitoring and has a high reliability for the assessment of the impacts of land subsidence. The approach has great practical significance and can be applied in many fields, e.g. monitoring the deformation along and around the railway to assure the train operation safety.