Evapotranspiration estimates from eddy covariance and from UAV imagery

This study compares evapotranspiration (ET) estimates obtained from 42 hours (7 9 July 2023) of continuous eddy covariance measurements with ET estimates derived from UAV thermal and multispectral imagery collected in 12 flight missions completed in the same period. Meteorological conditions were stable during the measurement campaign with mostly clear sky, weak wind and air temperatures fluctuating diurnally between +8 to +30C. The eddy covariance estimates averaged over 30 minute intervals corresponded to the source area covered mostly by oat field and freshly cut meadow. The algorithm for ET estimation was based on the Priestley-Taylor scheme applied in the ECOSTRESS Level-3 Evapotranspiration product, however, with a number of variables (components of radiative energy balance, relative humidity, ground heat flux) obtained from direct measurements. The NDVI and SAVI indices obtained from multispectral UAV images were used to estimate fractions of absorbed and intercepted photosynthetically active radiation. UAV-based ET estimates obtained at ca. 8 cm spatial resolution were averaged over the eddy covariance footprint area and interpolated for the times of EC-based ET measurements. Our results show that the approach combining UAV-based thermal and multispectral imagery with point measurements of meteorological variables and energy balance components might provide robust spatial ET estimates for agricultural areas of the size covered by one UAV mission, this is of the order of up to tens of hectares. This research was funded by National Science Centre, Poland, project WATERLINE (2020/02/Y/ST10/00065), under the CHISTERA IV programme of the EU Horizon 2020 (Grant no 857925) and the "Excellence Initiative - Research University" programme at AGH University of Krakw.