A fine knowledge of the physical properties of the snowpack and of its evolution is a prerequisite for a large number of applications impacting human activities, such as avalanche forecasting, water resources management or projections of future climate evolution. Observations are usually sparse in time and space, and therefore unable to capture the high spatio-temporal variability that characterizes these properties. To address this limitation and to provide projections of the evolution of the snowpack physical state on various timescales, many numerical models have been developed over the years. They range from single-layer snow schemes in which the snowpack is represented as a single ephemeral soil layer featuring specific properties, to detailed snowpack models dealing explicitly with the snow microstructure and enabling an explicit description of the vertical distribution of physical properties.

However, even the most detailed snowpack models currently available suffer from several weaknesses. First, they do not adequately represent water vapor transfer within the snowpack. Yet, the latter is a critical process in polar regions where it accounts for most of the mass transfer occurring within the snowpack, leading to errors in modeled density profiles which impact the ground thermal regime and projected permafrost evolution. Second, many of the processes that they include, in particular those related to the evolution of the snow microstructure, are based on empirical laws calibrated against observations performed in alpine settings. This leads to a lack of universality of these models and limit their domain of applicability. Third, because they were not though as such from the very first stage of their development several decades ago, the available detailed snowpack models lack of modularity, which hinders their use for applications requiring a lighter description of snow in which only a limited number of the processes implemented are relevant.