2026 IEEE JOURNAL OF SELECTED TOPICS IN APPLIED EARTH OBSERVATIONS AND REMOTE SENSING 2026 19(卷), null(期), (2300-2313页)
Accurate, seamless products of evapotranspiration (ET) are crucial for understanding the hydrological and energy cycles. Currently, there is no routinely available daily ET product based on the thermal infrared satellite data and two-source energy balance (TSEB) models over China. In this study, we developed a seamless ET dataset (also including transpiration, T and evaporation, E) for China from 2001 to 2020, with spatial and temporal resolutions of 0.01 degrees and daily, respectively. This dataset is generated through the two-source energy balance model coupled to soil moisture (TSEB-SM) and a novel reconstruction method (DNN-ETo) using deep neural networks and reference evapotranspiration. This dataset showed good agreement with ground measurements across various landcover types with an RMSE, bias, NSE, and KGE of 1.02 mm day(-1), -0.17 mm day(-1), 0.51, and 0.73. Regarding spatial distributions and interannual variations, TSEB-SM ET exhibits similar performance to previous studies, while providing more reasonable values in the arid and semiarid regions. The dataset has high potential in water resources management and climate change research. The TSEB-SM ET product is freely available online.