C-band Sentinel-1 data for estimating the basal crop coefficient and evapotranspiration of winter wheat

Estimating crop evapotranspiration (ETc) is of primary importance for irrigation management. The model commonly used for this purpose is the FAO-56 approach which consists of accurately estimating the basal crop coefficient K_cb. Historically, K_cb is derived from optical indices such as NDVI giving its sensitivity to vegetation cover fraction and to the Leaf Area Index. Nevertheless, optical data are disturbed by the presence of clouds. In this context, the objective of this work is to investigate the potential use of all-weather radar data as a substitute of NDVI to derive K_cb. The study is conducted over two winter wheat fields (Field 1 and Field 2) in Morocco, monitored during two agricultural seasons 2016-2017 and 2017-2018. Each field is equipped with an eddy covariance station allowing the estimation of ETc every 30 minutes. In addition, a weather station was installed over an alfalfa plot near the study fields. First, the backscattering coefficient and the interferometric coherence ( ρ at VV polarization) are derived from Sentinel-1 data with a 6-day revisit time and a spatial resolution of 10 m. Second, empirical relationships have between established between K_cb, on one hand, and the interferometric coherence and the polarization ratio, on the other hand and the results are also compared to the classical K_cb-NDVI (derived from Sentinel-2) method. The results show that good statistical metrics are obtained between K_cb and NDVI (R=0.77 and RMSE=0.14 for Field 1). Similar results are obtained also using ρ (R=0.76, RMSE=0.18). Finally, the K_cb is estimated from the calibrated relationships on one season and then used to estimate ET_c. The results demonstrate reasonable estimates of ET_c on Field 1 (R=0.70, RMSE=0.75 mm/day and bias=-0.18 mm/days) using K_cb-ρ. By contrast, a significant overestimations is highlighted both with  (bias=0.73 mm/day) and NDVI (bias=1.46 mm/day) over Field 2. Interestingly, the K_cb-ρ relationship is more consistent in the estimation of ET_c when changing from one field to another. These outcomes open new perspectives for the estimation of ET_c from radar data as a potential substitute of NDVI in case of persistent cloud cover.