Improving the calibration-free complementary evaporation principle by linking with the Budyko framework

Abstract. While it has performed well in predicting terrestrial evapotranspiration (ET_a) in many gauged locations over the world, the calibration-free complementary relationship (CR) depends on a questionable assumption that the Priestley-Taylor coefficient (α_e) is spatially constant over an extensive area. In this work, we evaluated the predictive performance of this convenient method, which only requires atmospheric inputs, against in-situ flux observations and water balance estimates (ET_wb) in Australia. We found that the CR method with a spatially constant αe derived from fractional wet areas did not perform as highly as previous studies would suggest, underperforming three advanced ET_a models in closing basin-scale water balance. This problem was remedied by linking the CR method with a traditional Budyko equation that allowed upscaling of optimal αe values from gauged basins to ungauged locations. The CR method with the αe upscaled by the atmospheric inputs and the mean precipitation (P) better reproduced the grid ET_wb available over the entire continent, and outperformed the three ET_a models. This study suggests that the fixed αe could lead the CR method to biased ET_a estimates, and it needs to be constrained by climate conditions to better close local water budgets.