On Scattering Model Decomposition of POLSAR Image and Its Application to the ESPRIT-Based Pol-InSAR

Abstract:
In this paper we propose a modified ESPRIT-based Pol-InSAR technique by using scattering model decomposition. The ESPRIT-based Pol-InSAR technique is one of the promising techniques to estimate forest tree-height with L-band fully polarimetric data. By using the technique, we can directly resolve interferometric phases of the ground and the canopy, respectively. However, depolarized components in dense forest canopy often cause phase bias in the estimation. In this paper, we propose to use scattering model decomposition technique as a pre-processor to remove depolarized components in the ESPRIT-based Pol-InSAR technique. By using the pre-processor, we can obtain not only accuracy improvements of local interferometric phases but also estimate shape of covariance matrix of volume scattering component in forest region. In addition, we also consider ambiguity in decomposition with scattering model based decomposition with POL-SAR data. These are verified by using the SIR-C/X-SAR data.