Discrimination of basmati and non-basmati rice types using polarimetric target decomposition of temporal SAR data

Abstract

The present study distinguishes the growing areas of basmati and non-basmati rice types using polarimetric target decomposition technique on temporal Synthetic Aperture Radar (SAR) data. Multi-temporal quad-pol RADARSAT-2 data of part of the Indo-Gangetic plains were acquired to analyse the contribution of different scattering components (double bounce, singe bounce and volume scattering) at various crop growth stages of both rice types. A decision tree-based framework has been proposed to segregate both rice types and other major land use-land cover classes by capturing the temporal variations in different scattering components. Both rice types were separated in the study area with user's accuracy of 85.19% and 82.93% for non-basmati and basmati rice respectively.