Evaluation of EOS-04 Polsar Data for Land Cover Classification Using Machine Learning Approach

Polarimetric Synthetic Aperture Radar (PolSAR) is a powerful remote sensing technique that captures detailed information about surface scattering properties. Recently launched Earth Observation Satellite (EOS)-04 is equipped with advanced C-band PolSAR sensor provides highresolution SAR data. This study leverages EOS-04 PolSAR data to analyse scattering parameters derived from various decomposition techniques. By evaluating the effectiveness of these parameters, we identified the most informative features for classification. These optimal scattering parameters are then stacked with optical bands from Sentinel-2 imagery, and a Support Vector Machine (SVM) classifier was applied. Our results demonstrated that the SVM model utilizing integration of PolSAR and optical data achieves superior accuracy compared to using optical or SAR data alone, highlighting the significant advantages of combining these data sources for precise land cover classification.