Evaluation of full-polarimetric parameters for vegetation monitoring in rabi (winter) season

Remote sensing data particularly Polarimetric Synthetic Aperture Radar (Pol SAR) data has shown promising results in crop monitoring and classification with reasonably high accuracy in tropical countries like India. Our study explores the use polarimetric SAR data for crop discrimination by evaluating different polarimetric parameters and various decomposition techniques using single date fine quad-pol C-band RADARSAT-2 dataset over heterogeneous agricultural area of Mehsana district of Gujarat state in India. Vegetation parameter, Radar Vegetation Index (RVI) and five decomposition techniques (e.g. H/A/α, Freeman-2, Freeman-3, Van Zyl-3 and Krogager) based on incoherent and coherent decompositions were selected to generate input images of classification. RVI shows good correlation with the crop stage at higher biomass with high value of RVI and with urban and water body with lower values of RVI respectively. H/A/α and its combinations i.e. (H-A) and (1-H)(1-A) show good results for discriminating land cover with different scattering mechanisms. Freeman-2, Freeman-3, Van Zyl-3 and Krogager are also evaluated for classification capability between various land covers and different crop types at different phenological stages. In 3-component decompositions, Freeman gives better results compared to other decomposition techniques for discriminating agricultural crops and natural vegetation. VanZyl decomposition overestimates surface component for agricultural crops and vegetation compared to Freeman 3-component decomposition. Krogager decomposition gives better seperability for man-made targets. The limitations of optical data for discrimination of crops based on biophysical parameters at various crop stages with similar spectral signature can be overcome by even single date full polarimetric SAR data with different decomposition techniques.