UAV-based estimation of vineyard actual evapotranspiration using the Shuttleworth and Wallace model

A field experiment was carried out to implement the Shuttleworth and Wallace (SW) model to estimate vineyard actual evapotranspiration (ET_a) using thermal images from an unmanned aerial vehicle (UAV) and meteorological data. The vineyard (‘Cabernet Sauvignon’) was located in Pencahue Valley, Maule Region, Chile (35°20’33”S, 71°46’41”W, 86 m a.s.l.). For this study, an UAV was equipped with an infrared thermal camera (FLIR/TAU-2) in order to obtain surface temperatures at a very high resolution (6×6 cm) during the 2018-2019 growing season. Meteorological variables and surface energy balance (EB) components were measured at the time of the UAV overpass. The performance of the SW model was evaluated using measurements of ET_a obtained from an eddy covariance system (EC). In addition, estimated values of latent heat flux (LEi), net radiation (Rni) and soil heat flux (Gi) at the time of the UAV overpass were compared with ground-truth measurements from a four-way net radiometer and flux plates, respectively. Results indicated that SW model estimated ET_a with errors = 5%, root mean squared error (RMSE) = 0.37 mm day^‑1 and mean absolute error (MAE) = 0.27 mm day^‑1. Finally, instantaneous values of LEi and Rni were computed with errors of less than 10% and with values of RMSE and MAE of less than 34 W m^‑2. Results demonstrated that a thermal camera placed on an UAV could provide an excellent tool to estimate intra-vineyard spatial variability of Rni, Gi, LEi and ET_a.