Capability Assessment of Fully Polarimetric ALOS–PALSAR Data for Discriminating Wet Snow From Other Scattering Types in Mountainous Regions

This paper examines the capability assessment of fully polarimetric L-band data for the snow and nonsnow-area classifications. The data sets used are the fully polarimetric Advanced Land Observation Satellite-Phased Array-Type L-Band Synthetic Aperture Radar data, optical Advanced Land Observing Satellite (ALOS)-advanced visible and near-infrared radiometer-2 data close to the radar acquisition, and environmental satellite-advanced synthetic aperture radar data. Several parameters are used to discriminate the snow-covered areas from nonsnow-covered areas in the Indian Himalayan region, including backscattering coefficients, the ratio of cross/copolarized backscattering power and polarization fraction (PF) value. Supervised classification schemes are employed using polarimetric decomposition methods based on the complex Wishart classifier. The accuracy of the classification was found to be 97.95% for the Wishart-supervised classification. Among various parameters and methods, it was found that the alternative newly proposed PF scheme, based on the implementation of fully polarimetric synthetic aperture radar data, yielded the best classification result in the absence of the training samples. The PF value has been effective for discrimination of the snow-covered areas from nonsnow-covered areas, debris-covered glacier, and vegetation. The results of this investigation show that L-band fully polarimetric SAR data provide considerable improvement but may not possess the optimal capability to discriminate snow from other inherent natural and man-made scatterers in heavy snow-laden mountainous scenarios, which may require fully polarimetric S-band or C-band PolSAR measurements.