AbstractEvapotranspiration (ET) is a critical process that regulates the transfer of heat and water between land and the atmosphere. While satellite‐based ET algorithms can provide area‐wide daily ET estimates, few are independent of local meteorological measurements. The Variant of the Moderate Resolution Imaging Spectroradiometer Standard Evapotranspiration Algorithm (VISEA) can potentially run without ground‐based observations. However, the surface energy budgets used in VISEA generally overpredict daily net radiation and related ET. To improve the accuracy of net radiation and ET, we incorporated Brutsaert's atmosphere emissivity model and corrected the routine to calculate the downward long‐wave radiation on cloudy days (which we called VISEA2023 model). The VISEA2023 model predicted net radiation and ET that agreed well with measurements at seven ChinaFlux sites, with an R2 of 0.8 and an Root Mean Square Error (RMSE) of 30 W m−2 for net radiation, and an R2 of 0.6 and an RMSE of 1 mm day−1 for ET, respectively. Additionally, the VISEA2023 model is more robust in comparison to the Variable Infiltration Capacity hydrologic model output in 10 major river basins and 80 sub‐river basins, with an R2 of 0.91 (0.78) and an RMSE of 85 (127.4) mm year−1, than MOD16, which has an R2 of 0.81 (0.82) and an RMSE of 134.8 (157.7) mm year−1, and Advanced Very High Resolution Radiometer ET, which has an R2 of 0.92 (0.62) and an RMSE of 90.1 (136.1) mm year−1. The improved VISEA2023 model can offer near‐real‐time ET on a larger scale, improving our understanding of water cycles and their relation to climate change.