Snow transport favors strong sublimation and may therefore have an important effect on the surface mass balance of polar and high-mountain regions. Recently, small-scale models such as large-eddy simulation (LES) with Lagrangian snow particles have improved the understanding of snow transport processes and revealed shortcomings in the parameterizations of large-scale models. This study leverages LES simulations to assess and improve current parameterizations of sublimation and snow transport. Measurements from the S17 site, East Antarctica, are used to define realistic model parameters and boundary conditions and verify the plausibility of the simulations. Various parameterization options are tested in a simple one-dimensional model inspired by the large-scale model CRYOWRF. When parameterizing the moisture and heat fluxes for given mass and number mixing ratios of particles, four improvements lead to a good agreement with the LES simulations: (a) A low surface friction velocity, (b) at least one grid level in the saltation layer, (c) prognostic humidity and temperature values at all heights, and (d) a correction term in the sublimation formula of Thorpe and Mason. The correction term accounts empirically for transient particle temperatures in the lowest 0.3 m of the atmosphere but requires further validation in a wider range of conditions. When estimating the particle mixing ratios in the one-dimensional model, an improved vertical discretization is critical. Overall, the proposed improvements change the latent heat flux by up to 102 W m^-2 (or 67%). To reduce the remaining errors, the saltation-suspension interface and near-surface particle speed should be better parameterized.