Assessing multi-sensor hourly maize evapotranspiration estimation using a one-source surface energy balance approach

Abstract

In this study, the performance of a one-source surface energy balance (OSEB) remote sensing (RS) of actual crop evapotranspiration (ET_a), incorporating data from different spaceborne, airborne and proximal multispectral data, was evaluated. The RS platforms in this study included Landsat-8 (30 m pixel size), Sentinel-2 (10 m), Planet CubeSat (3 m), a handheld (proximal) multispectral radiometer (MSR) (1 m) and an unmanned aerial system (UAS) (0.03 m). A 2-year data set (2020 and 2021) from two maize research sites in Greeley and Fort Collins, Colorado, USA, provided ground-based data for estimating and evaluating hourly ET_a from the OSEB algorithm. The accuracy of OSEB hourly maize ET_a estimates was evaluated using calculated hourly maize ET_a using high-frequency data collected with an eddy covariance energy balance system installed at each research site. The results indicated that the Planet CubeSat multispectral sensor (3 m), combined with on-site surface temperature data, yielded the least errors when predicting maize ET_a. The hourly ET_a estimation errors for the Planet CubeSat were MBE ± RMSE of −0.02 (−3%) ± 0.07 (13%) mm h⁻¹. These results suggest the urgent need for a specific approach to improve RS multispectral and thermal radiometric data (quality) to better support sustainable irrigation water management practices.