Improving meso-gamma scale NWP of winter weather with advanced ground-snow analysis and downscaling

In collaboration with the Army Test and Evaluation Command (ATEC), NCAR has developed and has been operating a real-time four-dimensional data assimilation and forecasting (RTFDDA) system at seven test ranges across the U.S. The modeling system was built upon MM5 and WRF models with an enhanced "observation-nudging"-based continuously FDDA process. It is capable of providing multi-scale, rapid-cycling analyses and short-term forecasts to support range routine tests. Analyses and forecasts for the range local regions are produced using fine mesh domains with grid increment of 1-3 km, which resolve details of local underlying forcing due to complex terrain and land surface thermal and stress heterogeneities. Ground snowcover, with its unique physical properties (insulation, melting, and large albedo), sometimes dominates the land-surface forcing that controls the model surface and boundary layer weather analyses and forecasts. In clear-sky days after snowy weather, patches of snowcover lead to significant thermal contrasts across the snowcover boundaries, which affect local circulations and often results in daytime snow-breezes blowing from the cooler snow-covered grounds under weak synoptic situation. This effect is particularly evident in mountain desert areas such as the Army Dugway Proving Ground (DPG), Utah, where patches of snow can often last for a few days and the daytime solar heating contrast between snow patches and neighboring desert grounds is very large. Differences of the day-time surface temperatures between snow grounds and bare desert grounds can be as large as 8-12°C. Thus, to accurately define the initial snow-cover fraction and snow water equivalent (SWE) are critical for mesoscale analyses and forecasts of winter weather.

In this paper, an advanced high-resolution snow analyses scheme that composites multi-source of snow measurements and coarser analysis products is developed. The data sources include: 4 km satellite-derived daily Northern Hemisphere NOAA/NESDIS snow mask, daily 1 km SWE from the National Operational Hydrologic Remote Sensing Center (NOHRC), and the EDAS SWE analysis (for regions outside the NOHRC covered area). The SWE used to initialize the NWP model is derived as follows. EDAS (outside the U.S.) and NOHRC SWE values are interpolated to the NWP model grid. In snow-free areas, as depicted by NESDIS snow mask, all SWE is removed. In other areas, the model-interpolated SWE is corrected to be at least the snow depth that would give the satellite-derived snow cover fraction (using a standard SWE to snow cover fraction relationship).

Numerical experiments are conducted for two 2007 winter cases with the DPG RTFDDA model system to demonstrate the impact of the new snow analysis scheme. RTFDDA produces significantly-improved surface and boundary layer weather analyses with the new snow analyses than those interpolated from the coarse NCEP operational GFS and/or NAM models.