Understanding and Correcting Non-Equilibrium Biases in Surface Flux Equilibrium-Based Evaporative Fraction Estimation.

Estimating evapotranspiration (ET) in areas with limited data remains a challenge, yet it is crucial for water and agricultural management. The Surface Flux Equilibrium (SFE) method is a promising approach, as it estimates the evaporative fraction (EF) using only temperature and humidity observations, assuming that the surface and air reach a steady balance when drying from sensible heat and moistening from ET become similar. In this study, we explore the conditions in which the assumptions of the SFE theory do not hold and identify the variables that can help correct the resulting biases using the ERA5 reanalysis dataset over the Indian landmass. We find that in water-limited regions, the temperature difference between the land and atmosphere can be used to correct biases in SFE-based EF estimates when there are longer intervals between two rainfall events, while relative humidity can be used to correct biases in areas with more frequent rain. In energy-limited regions, net radiation controls the surface flux imbalance and can therefore be used for bias correction. Incorporating these region-specific variables into machine learning models significantly improves SFE’s EF estimates. Our results highlight the value of identifying and using physical indicators to enhance the accuracy of SFE under non-equilibrium conditions.