An insight into the sensitivity of fully polarimetric SAR data to biomass of pearl millet crop

Abstract

The development of polarimetric decomposition techniques improves the capability to understand the complex scattering mechanisms occurring in the dryland crops. The present study is focused on assessing the sensitivity of coherent, incoherent, model and Eigen vector-based decomposition techniques from multitemporal Radarsat-2 polarimetric dataset to scattering mechanisms occurring in the critical growth stages of pearl millet crop. Yamaguchi volume component along with its temporal response was observed to be sensitive and reliable due to its dynamic biomass characterization. Overestimation of volume component and earlier saturation of biomass from 4 kgm−2 were observed limitations in Freeman 2 & 3 component decomposition when dealing with higher vegetation crops. The decrease in volume scattering over increasing biomass proved Van Zyl response contrary to the established understanding. Panicle stage was identified as the critical stage for discriminating pearl millet from similar structured maize and sorghum crops. In this stage, alpha angle generated from Cloude-Pottier decomposition almost reached 550 indicating the increase in contribution of dihedral scattering due to vertical prominent panicles in the pearl millet fields. The various Touzi parameters (α1, Ψ1, φ1 and τ1) were observed to have sensitivity towards changes in crop canopy structures, which will require further research to verify. The proposed Δ(F3-Y4) was the most suitable Volume Difference Index (VDI) for characterizing biomass of erectophile crops as the dynamic biomass response was found even beyond 8 kgm−2.