SAR tomography for forest height retrieval based on compressive sensing and post-processing

Forest height retrieval is a major tool for estimating forest biomass, which plays a key role in global carbon studies. TomoSAR using multiple baselines to form an elevation synthetic aperture which has the capability to retrieve the structural information of the observed objects. However, its image quality is limited by the distribution and extent of the baselines. Compressive sensing provides super resolution power and allows to recover details not accessible otherwise. Since the structure of a boreal forest appears at two layers at P-band, compressive sensing method can be used to retrieve the height. In this paper we introduce a compressive sensing based method to retrieve height of the boreal forest within the Krycklan River catchment, Northern Sweden, investigated at P-band during the ESA campaign BioSAR 2008. Because compressive sensing occasionally causes outliers and slightly underestimates the amplitudes, a two-step post-processing method is followed. The experiment results verify the performance of the proposed method for retrieving the boreal forest height.