Capability of Decomposition Methods for Identification of Crops and Other Land-cover Targets using Hybrid Polarimetric SAR Data

The main objective of current study is to investigate the potential of decomposition methods for monitoring crops and discrimination of other land cover targets using c-band hybrid polarimetric Risat-1 SAR images. There are two study areas namely Burdwan and Bharatpur in India chosen for analysing various existing decomposition methods in this paper. The Risat-1 hybrid polarimetric SAR Single look complex (SLC) data by ascending observation mode were utilized in our experiment and acquired in the month of December 22nd, 2014 and August 3rd, 2016 from parts of Burdwan and Bharatpur area respectively. The Stokes classical parameters G(0), G(1), G(2) and G(3) are derived from hybrid SAR images for further analysis. From these Stokes parameters, the relative phase, degree of polarization, orientation, ellipticity and polarization angle are calculated. Furthermore, four decomposition techniques namely m-delta, m-alpha, m-chi and modified m-chi are performed and expressed in the form of odd bounce, even bounce and volume components for monitoring crops and surrounding land cover targets in our study areas. The preliminary results have been observed from supervised classification on the basis of decomposition methods for identification of various crops, barren land, urban areas and waterbodies in the study sites. It has been shown that volume component among all decompositions is over estimated in comparison to odd and even bounce components. Risat-1 hybrid SAR data is found to be more suitable, convenient and cost-effective for discrimination of various land cover targets whereas cloud free optical data is a prime hindrance to the crop inventory.