SAR tomography using staring and high-resolution spotlight data from the TanDEM-X pursuit monostatic mode

Abstract

Synthetic aperture radar (SAR) tomography has been intensively used in urban scenarios to reconstruct the elevation position of each coherent scatterer overlaid within a 2-dimensional focused azimuth-range pixel [1] [2]. For this purpose, observations sampled at different spatial frequencies w.r.t. a reference ‘master’ scene are needed. Modern algorithms, e.g., SL1MMER proposed by Zhu and Bamler (2010), have been developed to achieve both robustness and super-resolution. Due to the intrinsic highly heterogeneous nature of urban areas with buildings and other infrastructures, SAR imaging mode with the highest resolution possible is usually favored. This insures a small number of scatterers inside each resolution cell and thus increases the probability of successful recovery. However, data takes over time seem to become obsolete, as older satellites are replaced by their successors which are equipped with upgraded sensors. Besides, new methodologies emerge which contribute to draining more performance out of available hardware. The former could be that the German satellite TerraSAR-X retires and hands its job over to the foreseen Next Generation TerraSAR-X HD, which will possibly be characterized by larger range bandwidth, while steering the radar beam more sharply to synthesize a larger aperture could be an example for the latter. Anyway, SAR data with the same resolution will be discontinued, as long as better resolution is desired. Naturally, the same problem could also arise purely due to the change of public interest within the life span of a satellite. In this abstract, we propose a new method to jointly carry out tomographic processing on SAR data from the same pipeline but with different resolutions. The data set used in this study will be described in Section 2. Section 3 is dedicated to the demonstration of the mathematical model. An outlook will be given in Section 4.