SAR tomography for deformation analysis in urban areas
Spaceborne Synthetic Aperture Radar (SAR) is ubiquitously used for quantitative assessment of various parameters of environmental concern, including land deformation. Advanced SAR signal processing techniques, in particular Differential SAR Interferometry (DInSAR), are used to provide an estimate of the radar line-of-sight deformation by extracting the deformation phase contribution in the observed interferogram among a pair of complex SAR images. Prior to an accurate estimation of the deformation, various other sources of unwanted phase contributions such as topography, atmosphere, etc. and phase errors have to be removed. The relatively recent advancement of DInSAR, namely Persistent Scatterer Interferometry (PSI), is better equipped to deal with temporal and geometric decorrelation and atmosphere-induced phase components found in SAR interferograms, compared to conventional DInSAR. It, therefore, allows the use of a large number of temporally diverse SAR images for deformation analysis, hence provides better deformation sampling over the course of time. PSI inherently depends on the identification of `point-like' scatterers, the so-called persistent scatterers (PSs), whose behavior is ideally analogous to point targets. The density of PSs, however, may not always be sufficient for a deformation analysis. Moreover, PSI conventionally relies on the assumption that there is a single scatterer within a resolution cell, which implies that targets in layover (as is often the case in urban areas) would not be included in a PSI-based deformation analysis.
In this project, we explore the possibility of improving the density of these PSs by developing methods to integrate SAR tomography into PSI-based deformation analysis, particularly for urban areas. A built-up city area has various man-made structures of different heights, such as houses, medium-to-high rise malls, roads, monuments, etc. Layover scenarios occur frequently. A SAR tomographic analysis will be carried out to first separate multiple scattering contributions in the same range-azimuth resolution cell, and thereby alleviate the layover problem. Subsequently, the individual phase histories of the separated scatterers will be derived, and afterwards each scatterer would individually be assessed as a PS. In this way, SAR tomography will be used to overcome the aforementioned limitations in PSI. Moreover, this work also includes the development of advanced multidimensional tomographic techniques to investigate the possibility (keeping in view the various practical limitations) of simultaneous estimation of both the scatterer location in elevation as well as the deformation.