Detailed comparison of observed and modeled tornadogenesis

Previous work by the authors (22nd Severe Storms Conference) examined WSR-88D radar observations of the 8 May 2003 Oklahoma City, OK tornadic supercell and model representations produced by assimilating radar observations into a numerical cloud model. Overall evolution of the modeled storm was compared to the radar observations. Furthermore, the feasibility of initializing a high-resolution forecast model with the assimilation results was tested.

This work closely examines the time interval of the formation of a violent (F4) tornado from the 8 May storm and compares high-resolution model data to high-resolution radar observations. The high-resolution model data are produced by ensemble Kalman filter assimilation of radar observations into the NSSL Collaborative Model for Multi-scale Atmospheric Simulation (NCOMMAS). NCOMMAS output at 500-m grid spacing and 1-minute intervals are compared to OKC-area TDWR radar observations. The very-near-range-to-the-storm TDWR data feature horizontal and vertical resolution of about 200 m and updates every minute. Low-level TDWR radial velocities are combined with KTLX WSR-88D radial velocities to produce dual-Doppler horizontal wind fields at 5-minute intervals to compare observed and modeled horizontal wind vectors. Areas of agreement and disagreement between observed and modeled tornadogenesis processes will be summarized.