Comparison of tropical forest Above Ground Biomass estimation techniques based on Polarimetric and Tomographic SAR data acquired at P band
Résumé
To understand and quantify the role of forests on carbon cycle and its impact on climate change, it is critical to accurately quantify and monitor changes in forest Above-Ground-Biomass (AGB). New advances in radar technologies have enabled the development of methods to map forest structure in 3-D. These techniques are based on the use of Polarimetric SAR Tomography (PolTomSAR) that enables the estimation of vertical profiles through a tridimensional (3-D) mapping of the backscattered power. The estimated vertically-spaced slices of backscattered power may contain information relative to two uncorrelated layers the ground defined as a surface backscatterer and the canopy modeled with its volume vertical structure. Thus, an accurate AGB information may be extracted from the power measured at a domain adapted to the canopy layer and less affected by the ground echo mainly due to the double bounce contribution. An algebraic synthesis of the forest allows a Sum of Kronecker Product (SKP2-) decomposition of the data covariance matrix which makes possible a polarimetric filtering of the ground echo. Moreover, the processing of a vertical domain adapted to the TomSAR vertical resolution and to the tree height allows a spatial filtering of the ground echo. The proposed tomographic biomass estimator is compared to the existing ones using P-band data acquired by the SETHI sensor over the Paracou data site in French Guiana in 2009.