Characterizing InSAR layover and multipath signals in dense urban environments

Abstract

As carriers of modern civilization, cities are spaces where humans live. In the background of global urbanization, an increasing number of Chinese cities are experiencing a vast amount of infrastructure construction, including bridges, buildings, tunnels, high-speed railways, airports, and dams. The gradual deterioration of infrastructures under these pressures may cause catastrophic structural failure, leading to large personal and economic losses. To reduce the risk of urban infrastructure deformation, an increasing number of technologies involve monitoring, evaluating and understanding deformation risk and how it emerges over time. Synthetic aperture radar interferometry (InSAR) can achieve all-weather, all-time, large-scale and full-coverage earth observations, and has been extensively employed for surface deformation monitoring and 3D (three-dimensional) digital surface model reconstruction in large cities. Especially with the launch of high-resolution SAR sensors, the spatial resolution of SAR images can reach 1 meter, which facilitates structural deformation monitoring of linear infrastructures and single buildings. Due to the one-dimensional side view imaging geometry of SAR, many distortions (shadow, foreshortening, and layover) inevitably exist in SAR images. These distortions lead to challenges in interpreting SAR images. In addition, the interaction between the scanning object and its surrounding objects, namely, the multipath effect, will produce ghost scatterers. Ghost scatterers blur the target image and reduce the image resolution. Almost no InSAR processing methods consider how to overcome the effects of imaging geometric distortions and the multipath effect during signal transmission. Use of the interferometric phase with various geometric distortions and the multipath effect is not reasonable as the input data for phase unwrapping. This thesis addresses some of the key issues in monitoring of urban ground subsidence and the structural deformation of large infrastructures with InSAR time series technology. The tomographic SAR (TomoSAR) and SAR simulation techniques are utilized to investigate two major application difficulties of InSAR in dense urban areas, namely, layover and the multipath effect. A comprehensive analysis of the SAR simulation results, signal characteristics in radar images, structure of the target building, and InSAR results is performed. The main research results and innovations of this thesis are as follows: 1. An InSAR phase unwrapping method based on LLL lattice reduction (proposed by three people whose names start with 'L': Lenstra A. K., Lenstra H. W. & Lovasz L.) and long-short baseline iteration is proposed. This method is applied for deformation and elevation extraction of the Lupu Bridge in Shanghai. The experiment shows that the long-short baseline iteration method is suitable for elevation extraction to overcome the high phase gradient problem if a DSM (Digital Surface Model) is not available and to improve the accuracy of the estimated deformation rates. The LLL lattice reduction algorithm is used to rapidly reduce the search radius, compress the search space, and improve the success rate of resolving the ambiguities in phase unwrapping. 2. The TomoSAR method is used to solve layover, a common difficulty in an urban SAR imaging environment. Scatterers of different elevations, which are overlaid on the same pixel, can be distinguished. Three typical TomoSAR methods are used for simulation when one and two scatterers are overlaid on one pixel. An extended 4D TomoSAR experiment, including 3D coordinates and seasonal deformation, is performed to solve an extremely difficult urban SAR imaging problem. The results show that the efficiency of the TomoSAR method, which is based on compressed sensing, is substantially improved. 3. An SAR simulation based on ray tracing is adopted to recover the signal propagation path, record the signal bounce level and dominant scattering mode, track the accurate locations of scatterers in building model coordinates and simulate a reflectivity map that is comparable to an SAR intensity image. SAR simulation is helpful for analyzing, interpreting SAR images, validating the SAR image processing algorithm, and performing SAR image geometric correction. The findings of two case studies provide further insights into the relationship between imaging geometry and SAR phase/intensity. 4. The generation, law and manifestation of the multipath effect in a radar image are investigated. Five statistical indexes of SAR images are first proposed to identify the pixels that incorporate multipath and layover. Results show that the mean amplitude and amplitude standard deviation are the key indicators for the purpose. The influence of the multipath effect on the actual InSAR solution is analyzed by eliminating the pixels that incorporate multipath. The results show that ghost scatterers are often accompanied by layover. Underground ghost scatterers can be identified and screened out using the elevations obtained by persistent scatterer interferometric SAR (PSInSAR). However, the inclusion of ghost scatterers does not affect the elevation or deformation estimation of normal scatterers.