Résumé
Abstract. Snow microstructure – characterized by density, grain size, grain shape and arrangement – fundamentally determines snowpack macroscopic properties. Despite this critical role, many land surface models (LSMs) lack explicit representation of snow microstructure. This limitation has become increasingly critical as future spaceborne missions for snow water equivalent measurement demand advanced modelling systems capable of accurately estimating snowpack properties, including microstructure, across diverse climatic and vegetation regions. The Soil Vegetation and Snow (SVS) LSM, used by Environment and Climate Change Canada for operational land surface and hydrological predictions, has been substantially upgraded to address these challenges. SVS version 2.0 (SVS2) incorporates the detailed multilayer Crocus snowpack model, enabling distinct simulations of snowpack evolution in both open terrain and forested areas within each grid cell. Crocus within SVS2 has been upgraded from its original alpine design with three major enhancements to handle Canada's varied snowpack conditions: an advanced albedo parameterization that accounts for spatial variability in light-absorbing particle deposition, new physical parameterizations tailored to Arctic snow characteristics, and a refined canopy model for forest environments. Significant improvements in simulations of near-surface density predictions are evident along a latitudinal transect from southern Quebec to the Canadian Arctic, while challenges remain in simulation of density and specific surface area in basal snow layers. SVS2 achieved substantial gains in snow melt-out timing accuracy, reducing prediction errors by over 50 % compared to the alpine Crocus version and surpassing two established snow reference products (ERA5-Land and ERA5-Crocus). These enhancements position SVS2 as a substantial advancement for future operational snow modeling applications across Canada.