Multi-scale DA is applied to the 13 June 2010 case during VORTEX2. On 13 June a cold pool from overnight convection created an outflow boundary that was located near the Oklahoma-Texas Panhandle border in the afternoon. New convection developed along a cold front in the Texas Panhandle. The sub-severe convection slowly moved to the northeast and a cell moved over the intersection of the two boundaries, intensified, gained supercell characteristics, and became tornadic. This case represents a complex mesoscale environment and storm evolution that was not captured well with conventional observations or WSR-88D radars. Thus, this case presents a challenging event to analyze and predict, and could demonstrate the benefit of multi-scale DA in generating initial conditions for ensemble forecasts.
A mesoscale WRF model domain (15 km horizontal grid spacing), and a nested convective-allowing WRF model domain (3 km horizontal grid spacing), are used along with the Data Assimilation Research Testbed (DART) system for the analysis and forecast of the storm environment and convection on 13 June 2010. The impacts of radar data assimilation are investigated through comparisons of ensemble forecast performance relative to a control analysis system where only conventional observations are assimilated. Results indicate the addition of radar observations in hourly DA cycling leads to improvement in forecast skill, which is tied to better forecasts of the outflow boundary from overnight convection and subsequent convective evolution.