Evapotranspiration (ET) is a vital part of the hydrological cycle and the water–energy balance. To explore the characteristics of five typical remote sensing evapotranspiration datasets and provide guidance for algorithm development, we used reconstructed evapotranspiration (Recon) data based on ground and GRACE satellite observations as a benchmark and evaluated five remote sensing datasets for 592 watersheds across the continental United States. The Global Land Evaporation Amsterdam Model (GLEAM) dataset (with bias and RMSE values of 23.18 mm/year and 106.10 mm/year, respectively), process-based land surface evapotranspiration/heat flux (P-LSH) dataset (bias = 22.94 mm/year and RMSE = 114.44 mm/year) and the Penman–Monteith–Leuning (PML) algorithm generated ET dataset (bias = −17.73 mm/year and RMSE = 108.97 mm/year) showed the better performance on a yearly scale, followed by the model tree ensemble (MTE) dataset (bias = 99.45 mm/year and RMSE = 141.32 mm/year) and the moderate-resolution imaging spectroradiometer (MODIS) dataset (bias = −106.71 mm/year and RMSE = 158.90 mm/year). The P-LSH dataset outperformed the other four ET datasets on a seasonal scale, especially from March to August. Both PML and MTE showed better overall accuracy and could accurately capture the spatial variability of evapotranspiration in arid regions. The P-LSH and GLEAM products were consistent with the Recon data in middle-value section. MODIS and MTE had larger bias and RMSE values on a yearly scale, whereby the MODIS and MTE datasets tended to underestimate and overestimate ET values in all the sections, respectively. In the future, the aim should be to reduce bias in the MODIS and MTE algorithms and further improve seasonality of the ET estimation in the GLEAM algorithm, while the estimation accuracy of the P-LSH and MODIS algorithms should be improved in arid regions. Our analysis suggests that combining artificial intelligence algorithms or data-driven algorithms and physical process algorithms will further improve the accuracy of ET estimation algorithms and the quality of ET datasets, as well as enhancing their capacity to be applied in different climate regions.

Energy budget of Amazonian forests has a large influence on regional and global climate, but relevant data are scarce. A novel energy partition method based on the maximum entropy production (MEP) theory is applied to simulate evapotranspiration in Amazonia. Using site‐level eddy flux data, the MEP method shows high skill at the hourly, daily, and monthly scales. Consistent performance under different levels of land surface dryness is revealed, hinting that drought signal is appropriately resolved. The site‐level MEP‐based estimates outperform the estimates of the Moderate Resolution Imaging Spectroradiometer evapotranspiration product, which is commonly used for large‐scale assessments. At the Amazon basin scale, the two series yield similar averages but exhibit spatial differences. The parameter parsimony and demonstrated skill of the MEP method make it an attractive approach for environments with diverse strategies of water flux control.

The recently formulated maximum entropy production (MEP) model over land surfaces has been generalized to water‐snow‐ice surfaces. Analytical solutions of energy budget in terms of the partition of surface radiative fluxes into (turbulent and/or conductive) heat fluxes at the earth‐atmosphere interface are derived as functions of surface temperature (e.g., sea surface temperature). The MEP model does not require data of wind speed, air temperature‐humidity, and surface roughness. Test of the MEP model using observations from several field experiments is encouraging. Potential applications of the proposed model for understanding long‐term trends in surface heat fluxes and for closing global surface energy budget at the Earth's atmosphere are suggested.

The Weather Research and Forecasting (WRF)-Hydro model as a physical-based, fully-distributed, multi-parameterization modeling system easy to couple with numerical weather prediction model, has potential for operational flood forecasting in the small and medium catchments (SMCs). However, this model requires many input forcings, which makes it difficult to use it for the SMCs without adequate observed forcings. The Global Land Data Assimilation System (GLDAS), the WRF outputs and the ideal forcings generated by the WRF-Hydro model can provide all forcings required in the model for these SMCs. In this study, seven forcing scenarios were designed based on the products of GLDAS, WRF and ideal forcings, as well as the observed and merged rainfalls to assess the performance of the WRF-Hydro model for flood simulation. The model was applied to the Chenhe catchment, a typical SMC located in the Midwestern China. The flood prediction capability of the WRF-Hydro model was also compared to that of widely used Xinanjiang model. The results show that the three forcing scenarios, including the GLDAS forcings with observed rainfall, the WRF forcings with observed rainfall and GLDAS forcings with GLDAS-merged rainfall, are optimal input forcings for the WRF-Hydro model. Their mean root mean square errors (RMSE) are 0.18, 0.18 and 0.17 mm/h, respectively. The performance of the WRF-Hydro model driven by these three scenarios is generally comparable to that of the Xinanjiang model (RMSE = 0.17 mm/h).