Evaporation from snow surface: A multi-model evaluation with the FLUXNET2015 dataset

Abstract

Evaporation from snow surfaces (Ess) is a critical component of snow mass balance, and its accurate estimation is essential for the water balance at high latitudes and altitudes. Selecting an appropriate model is important for achieving an accurate estimation of Ess. However, there have been few studies focused on the choice of Ess models. In this study, we evaluated six commonly used Ess models at twelve eddy covariance sites in the FLUXNET2015 dataset. The six models considered include an empirical method (EM), two bulk aerodynamic (BA) methods, two energy-based models (maximum entropy production (MEP) model and Priestley-Taylor (PT) model), and a combination model (Penman model). The results show that the performance of the six models differs significantly, and no single model consistently outperforms the others across all sites. Overall, the Penman method performs the best on both the hourly and daily scales with Kling-Gupta efficiency (KGE) coefficients of 0.75 and 0.65 on the two timescales, respectively. Therefore, we recommended the Penman model as the preferred model for Ess estimation. In the absence of net radiation data, we suggested using the EM model as it noticeably outperforms the other four models. The reason for the poor performance of the two BA models, with KGE values less than 0.4, may be attributed to their unreasonable parameterization of atmospheric stability. On the other hand, ignoring or simplifying the aerodynamic term is likely responsible for the poor performance of the MEP and PT models.