Backscatter Coefficients to Scattering Powers: Improving Dual-Pol Crop Classification

This study proposes a novel method for crop classification using dual-pol scattering power components (dihedral-like, surface-like, and unpolarized), comparing their performance against traditional backscatter coefficients. In tests with simulated NISAR data, the proposed method achieves 92 % accuracy with a kappa coefficient of 0.88, significantly outperforming backscatter-based classification, which shows 80 % accuracy and a kappa coefficient of 0.71. Furthermore, the method consistently outperforms the backscatter coefficient-based classification accuracy by at least 10 % across all crop growth stages, demonstrating the effectiveness of scattering powers for precise crop monitoring.