PolSAR calibration and reconstruction of hybrid polarimetric RISAT-1 data for pseudo quad-pol decomposition: a comparison with quad-pol

Abstract

A new approach to reconstruction of pseudo quad-polarized data from hybrid polarimetric data has been presented in this research. The algorithm is based on certain assumptions which were validated upon testing the aptness of the results and their comparison with true optical images of the region under study. This involved direct construction of the 3X3 coherency matrix from the 2X1 scattering matrices obtained from the hybrid polarimetric data. The reasonableness of the assumptions were tested by decomposing the reconstructed pseudo quad-pol data using a coherent decomposition mechanism. The data set used in this project was Level-1 FRS-1 Hybrid Polarimetric data and FRS-2 Quad-pol data of RISAT-1. Reliable scattering retrieval from SAR data involves the calibration of the data. Polarimetric calibration was performed on real and imaginary channels of the single look complex SAR data. The newly developed algorithm was implemented on calibrated data. To extract complete information of different scattering elements of any location, second order derivative of scattering matrix is the most suitable and widely used matrix. Coherency matrix of pseudo quad-pol obtained from hybrid polarimetric data using reconstruction algorithm was decomposed using Yamaguchi four component decomposition for scattering information extraction. The obtained surface, double-bounce and volume scattering were compared with the scattering elements of hybrid-polarimetric decomposition, m-alpha and decomposition of quad-pol data of RISAT-1. The comparison revealed that the results obtained were satisfactory and thus the assumptions made during the reconstruction of pseudo quad-pol data were reasonable for specific purposes. Further comparisons of results using different decompositions technique at pixel level comparison can help better understand the aptness of the algorithm.