Improvements to an intermediate complexity atmospheric model for high-resolution downscaling in very complex terrain.

Abstract:

Snow deposition patterns in complex terrain are heavily dependent on the underlying topography. In reality, this topography affects precipitating clouds at the kilometer-scale, and causes changes to the wind field at the sub-kilometer scale, resulting in altered advection of falling hydrometeors. Snow particles are particularly sensitive to changes in the near-surface flow field due to their low density. Atmospheric models which run at the kilometer scale cannot resolve the actual heterogeneity of the underlying terrain, resulting in precipitation maps which do not capture terrain-affected precipitation patterns. Statistical downscaling techniques are then unable to recreate the atmospheric dynamics that are active at the sub-kilometer scale and do not maintain the inter-variable dependencies. Thus, snow-atmosphere interactions such as preferential deposition are often not resolved in precipitation data used as input to snow models. To bridge this spatial gap and resolve snow-atmosphere interactions at the sub-kilometer scale, we couple an intermediate complexity atmospheric model (ICAR) to the COSMO NWP model. Applying this model to sub-kilometer terrain required changes to ICAR’s assimilation of forcing data, computational grid, and boundary layer scheme. In particular, the near-surface flow now accounts for surface roughness and topographically induced speed up. Our approach maintains the accurate large-scale precipitation patterns from COSMO but resolves the dynamics induced by terrain at the sub-kilometer scale. These downscaled precipitation patterns are validated using data from meteorological stations and wind and snow deposition data from a field campaign in the Swiss Alps (RACLETS). We find that solid-precipitation patterns at the ridge scale are better resolved in the high-resolution version of ICAR than the current ICAR or COSMO models. This updated version of ICAR presents a new tool for dynamic downscaling of NWP output as forcing data to snow models.