An eigenvalue decomposition based full polarimetric SAR framework for leaf area index retrieval using entropy-anisotropy features

Abstract

Present study proposes a novel Full Polarimetric Radar Vegetation Index (FPRVI) for accurate Leaf Area Index (LAI) retrieval using RISAT-1A full-polarimetric SAR data. The eigenvalue-based polarimetric decomposition parameters entropy (H) and anisotropy (A) have been employed to develop this index. Sensitivity of H, A, and FPRVI was investigated over a number of phenological phases, whereas Wishart H–α classification revealed a transition from volume scattering at the booting stage to double-bounce scattering with wheat grain development. When utilized as a vegetation descriptor in the Water Cloud Model (WCM), FPRVI exhibited strongest correlation to HV backscatter (r = 0.65). Additionally, FPRVI demonstrated a significant robustness with varying soil moisture. WCM-based LAI retrieval showed the best performance for soil moisture value 0.4 (r = 0.66; RMSE = 1.35 m²/m²). These outcomes demonstrate the potential of FPRVI for accurate LAI estimation and future retrieval of crop biophysical parameters across diverse agro-climatic conditions.