Magnitude and controls of snow sublimation at a high-elevation Swiss alpine site

Snow sublimation remains poorly quantified globally, with published estimates spanning orders of magnitude (≈0.01mm/day to >6mm/day), corresponding to ~ 5-90% of winter snowfall. This large uncertainty limits our ability to accurately quantify water availability to downstream ecosystems, as vapor losses through sublimation reduce meltwater availability. Here, we quantify surface snow sublimation over one full winter season (November 2024 to June 2025) at a high-elevation alpine site (Weissfluhjoch Versuchsfeld, 2455 m a.s.l., Switzerland) using continuous eddy-covariance measurements from an integrated open-path gas analyzer and sonic anemometer (IRGASON). We applied an eXtreme Gradient Boosting (XGB) model to estimate surface snow sublimation and used TreeExplainer-based Shapley Additive Explanations (SHAP) to quantify the relative importance of different meteorological variables on modeled sublimation.Over the 2024-25 winter season, cumulative net sublimation was 31±21mm, equivalent to 5.1±3.6% of winter snowfall, with a mean daily rate of 0.15 mm/day, placing our estimates at the lower end of previous compilations. During the accumulation period, sublimation accounted for 48% of cumulative winter sublimation and was primarily driven by vapor pressure deficit. In contrast, 52% of cumulative sublimation occurred when the snowpack was melting. For this period, net incoming radiation emerged as the dominant statistical driver of sublimation. Notably, net radiation is only indirectly represented in current energy-balance formulations of physically-based snow models, revealing potential limitations in existing Monin-Obukhov parameterizations. Our results therefore highlight the importance of continuous eddy-covariance observations throughout winter periods for accurately constraining snow sublimation and highlight potential biases in Earth System Models that do not explicitly represent radiative controls on sublimation.