Interception of snowfall by the trees is the main challenge for snowpack simulations under forests

The Crocus snowpack model is now coupled with the new vegetation scheme MEB (Multiple Energy Balance). This new system represents all the main snow-vegetation interactions by including 1 layer of high vegetation and 1 layer of litter. This work presents the first evaluations of this new system with a new dataset collected on a spruce forest at Col de Porte (1325 m a.s.l. Chartreuse massif, French Alps) during the winters 2016-2017 and 2017-2018. We illustrate that the processes linked with the interception of snowfall by the trees are the most challenging to simulate. The initial representation of this process in the model was based on empirical parameterizations from the literature with observations in Canada. This version fails to reproduce the very strong impact of interception on our dataset. We demonstrate that this error is much more significant than the spatial variability of the snowpack in the forest and than the uncertainties in the other processes of the model (including radiative effects of the canopy). Numerical experiments show that the melting of intercepted snow is likely to be highly underestimated. This suggests that a more physical parameterization of this process is required to improve snowpack simulations in this environment.