P-LSHv2: a multi-decadal global daily land surface actual evapotranspiration dataset enhanced with explicit soil moisture constraints in remote sensing retrieval

Abstract. Accurately quantifying the impact of soil water availability on evapotranspiration (ET) is curcial for improving ET retrieval accuracy. However, most global satellite-derived ET datasets do not explicitly incorporate soil moisture constraints, leading to significant uncertainties, particularly in water-limited regions. In this study, we propose an enhanced soil moisture constraint scheme that effectively captures soil moisture’s influence on vegetation transpiration and soil evaporation using a quantile-based approach. Unlike previous methods, this scheme relies solely on soil moisture data, reducing uncertainties associated with heterogeneous soil hydraulic properties. We integrated this approach into the process-based land surface ET/heat fluxes algorithm (P-LSH, or P-LSHv1), developing an improved version, P-LSHv2. Using observations from 106 global flux towers, we calibrated biome- and climate-specific parameters and quantified moisture constraints across diverse climates and land cover types. P-LSHv2 achieves notable improvements in ET estimation, with a reduced Root Mean Square Error (RMSE) of 0.67 mm d^-1; and an increased correlation coefficient (R) of 0.81, outperforming its predecessor, P-LSHv1, particularly in arid regions. Comparative analyses show that P-LSHv2 surpasses the Penman-Monteith-Leuning model and the Global Land Evaporation Amsterdam Model in capturing soil moisture anomalies' effects on ET, enhancing global ET accuracy. Leveraging the P-LSHv2 algorithm, we have produced a long-term global daily ET dataset spanning 1982–2023, providing a valuable resource for research on terrestrial water and energy cycles and climate change. The dataset is freely available at https://doi.org/10.11888/Terre.tpdc.301969 (Feng Jin, 2025).