Upscaling evapotranspiration from the instantaneous to the daily time scale: Assessing six methods including an optimized coefficient based on worldwide eddy covariance flux network

Remote sensing plays an irreplaceable role in mapping evapotranspiration (ET) at the regional scale. Instantaneous ET can be derived by multiple ET models based on various remotely sensed data at the time of satellite overpass. Therefore, a reliable temporal upscaling method from instantaneous ET to daily ET is of great importance. In this study, six widely used upscaling methods for estimating daily ET from instantaneous ET were compared under different ecosystems and meteorological conditions based on worldwide eddy covariance flux network. In total, 139 flux sites based on eddy ccovariance systems at the global scale were used, and the 9:00-15:00 overpass times of the main growing season (May to September in the Northern Hemisphere and November to March in the Southern Hemisphere) were chosen as the study period. Five assessment metrics (coefficient of determination (R-2), root mean square error (RMSE), mean absolute error (MAE), index of agreement (d), and relative error (RE)) were selected to evaluate the accuracy and applicability of the six upscaling methods. The results showed that the modified evaporative fraction (MEF) method and the sine method (only at specific times) performed best for most ecosystems, but the optimal instantaneous time was quite different for different ecosystems. The MEF method worked best at 12:00 for forests and shrublands and attained the smallest MAE at 13:00 for grasslands and savannas. For wetlands and croplands, the MEF method performed best at 14:00. More attention should be paid to the selected instantaneous time when using the sine method because of its high sensitivity to time of day. The modified coefficient of the MEF method, which varied with instantaneous time and ecosystem, ranging from 1.04 (grassland) to 1.36 (wetland), was also examined. The comprehensive results of this study could be useful for mapping daily ET from instantaneous ET.