Global ocean surface heat fluxes revisited: A new dataset from maximum entropy production framework with heat storage and Bowen ratio optimizations

Abstract. Ocean evaporation (latent heat flux, LE) plays a crucial role in global precipitation patterns, water cycle dynamics, and energy exchange processes. However, current bulk methods for quantifying ocean evaporation are subject to significant uncertainties. The Maximum Entropy Production (MEP) theory offers a novel approach for estimating surface heat fluxes, but its effectiveness over ocean surfaces has not been validated. This study integrates heat storage effects and four empirical Bowen ratio formulas into the MEP theory to improve ocean LE estimation. We employed multi-source data from 129 globally distributed buoy stations and seven auxiliary turbulent flux datasets for validation and comparison. We first evaluated the MEP method using observed data from buoy stations, identifying the optimal Bowen ratio formula to enhance the model. Results indicate that accounting for heat storage and adjusting the Bowen ratio significantly improve heat flux accuracy, with R²=0.99 and a root mean squared error (RMSE) of 4.7 W·m^-2 compared to observations. Subsequently, we conducted a thorough evaluation of seven global turbulent flux datasets to identify the most accurate input variables (e.g., heat storage, net radiation, surface temperature) for applying the MEP method on a global ocean scale. The enhanced MEP method provided new estimates of the annual average LE at 93 W·m^-2 and sensible heat at 12 W·m^-2 for the period 1988 to 2017. Validation against observations from 129 buoy stations demonstrated that the MEP-derived latent heat dataset achieved the highest accuracy, with a mean error (ME) of 1.3 W·m^-2, an RMSE of 15.9 W·m^-2, and a Kling-Gupta Efficiency (KGE) of 0.89, outperforming four major long-term global heat flux datasets, including J-OFURO3, ERA5, MERRA2, and OAFlux. Additionally, we examined the long-term spatiotemporal variability of global ocean evaporation, identifying a significant increasing trend from 1988 to 2010 at a rate of 3.58 mm/yr, followed by a decline at a rate of -2.18 mm/yr from 2010 to 2017. The current dataset provides a new benchmark for the ocean surface energy budget and is expected to be valuable for research on global ocean warming, sea surface-atmosphere energy exchange, the water cycle and climate change. The monthly MEP heat flux dataset for 1988–2017 is publicly available at https://doi.org/10.6084/m9.figshare.26861767.v2 (Yang et al., 2024, last access: 28 August 2024).