Scattering Parameters from Sentinel-1 SAR Data for Crop Growth Assessment

Accurate and high-resolution spatio-temporal information on crop growth is essential for agronomic management and grain yield estimation. Data provided by the dual-pol Synthetic Aperture Radar (SAR) on board Sentinel-1 is an effective way for crop mapping and monitoring from space independent of weather conditions. In this study, we propose a target characterization parameter,
$\theta_{\text{xP}}$
that utilizes the 2D Barakat degree of polarization and the elements of the covariance matrix representing dual-pol SAR data. We also propose an unsupervised clustering scheme using
$\theta_{\text{xP}}$
and the scattering entropy,
$H_{\text{xP}}$
. We utilize time-series Sentinel-1 data of wheat fields over a Canadian test site to show the sensitivity of
$\theta_{\text{xP}}$
to the development of crop morphology at different phenological stages. Similarly, the effectiveness of the unsupervised
$H_{\text{xP}}/\theta_{\text{xP}}$
clustering plane is also examined using a sequence of temporal clustering plots. This innovative clustering framework will be convenient for the operational use of Sentinel-1 SAR data for agricultural applications.