Spaceborne PolSAR Tomography for Forest Height Retrieval

Synthetic aperture radar (SAR) data with polarization diversity and multibaseline acquisition has the capability to resolve the structural information of the target. Three-dimensional (3-D) imaging using SAR tomography is an advance and recent technology that focuses on backscattered echoes not only in the 2-D plane defined by the azimuth and range coordinates, but also in cross-range direction. Forests act as a semitransparent media for radar wavelengths, and thus allow us to gather the vertical distribution of this volumetric scatterer. This paper investigated the potential of polarimetric SAR tomography (PolTomSAR) to retrieve the scatterer distribution of forest area using C-band Radarsat-2 spaceborne SAR dataset. Teak patch of Haldwani forest was chosen as the test site to validate the robustness of TomoSAR algorithms. Fourier, beamforming, Capon, Capon like, and polarimetric capon were implemented in this study and backscatter power variations with their corresponding vertical profile for a particular azimuth bin were derived and analyzed. It was observed that Fourier suffered from high sidelobes that were reduced by beamforming and Capon. Limitations of beamformer were highlighted and well resolved by Capon. However, both beamforming and Capon yielded low backscattering power at different height levels. The Capon-like algorithm substantially improved the backscatter intensity at different forest height levels. Polarimetric capon surpassed the Capon algorithm and was able to estimate better heights by incorporating various polarization channels. Height map was generated for fully polarimetric Capon and validated with the ground truth data.