High-resolution analyses of the 8 May 2003 Oklahoma City storm. Part III: An ultra-high resolution forecast experiment

Part III in this series of papers analyzing the F4 tornadic storm of 8 May 2003 will simulate the supercell in a similar manner as in previous work by the authors (Wicker et al., 2002) where grid resolution is high enough to resolve most of the turbulent eddy structures within the storm. The unique aspect of this work is that for the first time an LES simulation of a severe storm will be initialized from an analysis generated by the convective-scale ensemble Kalman filter assimilation presented in Part II by Dowell and Wicker (2004).

By combing ultra-high resolution modeling (dx < 100 m) with the EnKF storm-scale analysis we will analyze a storm-scale simulation that, at least initially, represents a “best estimate” the observed storm. This methodology is being proposed as a possible analysis technique for convective scale observations from several field programs (e.g., BAMEX and VORTEX-II). The simulated storm will be closely compared and validated via the available observations from nearby radars (not used in the data assimilation) as well as other available in situ measurements.

A close comparison between the simulated storm and the observational analysis presented in Part I by Burgess (2004) will be done, including a comparison of time-height diagrams, timing and vertical structure of the tornado formation, tornado track, intensity, etc. While it is unlikely that the simulated storm will closely match the observations for a long period of time or in such fine detail, the success of EnKF analysis for convective storm analysis (and in this case particularly) has encouraged us to attempt a detailed simulation and analysis so that we can evaluate the potential usefulness of this approach.