Radar reflectivity estimation using multiple SAR scenes of the same target: technique and applications

An investigation on techniques to estimate the underlying reflectivity in a time series of SAR images is presented in this paper. The topic has interesting implications for a wide range of radar remote sensing applications where the time evolution of the ecosystem is of importance. Focus will be given here to a thematic application-monitoring of the tropical forest-which is one of the major R/D activities within the European Commission TREES project, and which offered the motivation to develop the approach presented. The theory behind two statistical estimators, which were previously proposed by one of the authors, and that are used in the present study, is briefly summarized. Implementation issues which are specific to the case at hand are then discussed. Finally some results are presented, based on simulated SAR images and an ERS-1 time series acquired over the tropical forest in West Africa. The results highlight the performance of the filters, in terms of speckle suppression and restoration of the image texture. At this stage of the authors' research they can already conclude that the approach has a high potential for radar remote sensing applications; the key point is that it allows for the reconstruction of the underlying radar reflectivity time evolution at the full spatial resolution of the signal but with a dramatic improvement of the signal to noise ratio; the way is thus paved for reaching unprecedented results in the visual or automatic interpretation of SAR imagery.