Performance of tomographic SAR estimation techniques at low image resolution

Abstract

SAR tomography is a remote sensing technique that allows three-dimensional imaging of volumetric targets. It allows, for example, to analyse the vertical structure of vegetation layers or other targets with significant penetration of the sensor's radiation (e.g., dry soil and ice layers). Due to its unique capabilties and potential, SAR tomography has gained significant attention during the last years and has become an established technique for analysis of all kinds of three-dimensional backscattering scenarios. For future space-borne SAR sensors operating in longer wavelength, like BIOMASS or TanDEM-L, it is planned to employ tomographic imaging modes on an operational basis. Since the tomographic resolution is achieved not in the vertical direction but perpendicular to the line of sight, the sensor's range resolution has significant influence on the capability of resolving horizontally layered structures, like for example forest scenarios. This paper reviews the different state-of-the-art signal processing methods for three-dimensional SAR imaging with respect to their performance in case of low image resolution, with focus on imaging parameters as to be expected for upcoming space-borne missions. The expected imaging capabilities are demonstrated using real air-borne data acquired by DLR's E-SAR sensor. Special attention is also paid to illustrating the potential of the techniques to operate with low numbers of input images and under conditions of temporal decorrelation.