We describe in this paper the development of a double-moment model of blowing snow and its application to the Canadian Arctic. We first outline the formulation of the numerical model which solves a prognostic equation for both the blowing snow mixing ratio and total particle numbers, both moments of particles that are gamma-distributed. Under idealized simulations, the model yields realistic evolutions of the blowing snow particle distributions, transport and sublimation rates as well as the thermodynamic fields at low computational costs. A parametrization of the blowing snow sublimation rate is subsequently derived. The model and parametrization are then applied to a Canadian Arctic tundra site prone to frequent blowing snow events. Over a period of 210 days during the winter of 1996/1997, the near-surface relative humidity consistently approaches saturation with respect to ice. These conditions limit snowpack erosion by blowing snow sublimation to ~ 3 mm snow water equivalent (swe) with surface sublimation removing an additional 7 mm swe. We find that our results are highly sensitive to the proper assimilation of the humidity measurements and the evolving thermodynamic fields in the atmospheric boundary layer during blowing snow. These factors may explain the lower values of blowing snow sublimation reported in this paper than previously published for the region.