Warn-on-Forecast sensitivity experiments with the 19 May 2013 Norman-Shawnee, OK, tornadic supercell

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 4 November 2014: 4:45 PM
Madison Ballroom (Madison Concourse Hotel)
Corey K. Potvin, CIMMS/NSSL, Norman, OK; and L. J. Wicker, M. M. French, and D. W. Burgess

On 19 May 2013, a tornadic supercell that produced EF-4 damage near Shawnee, Oklahoma was observed by five Doppler radars (KOUN, KTLX, TOKC, MPAR, and NOXP) at very close range (~ 20 km, see figure). In addition, TOKC, MPAR, and NOXP collected partial low-level volumes at 1-1.5 min frequencies, and KOUN, NOXP, and KTLX collected dual-polarization radar data. The exceptional degree of verification enabled by this radar dataset presents an excellent opportunity to test ensemble system design choices for Warn-on-Forecast. To do this, we assimilate KTLX and TOKC reflectivity and velocity data using a WRF-LETKF system and then generate 1-2 h ensemble forecasts. Triple-Doppler retrievals are used for verification. Our initial focus will be on analysis and forecast sensitivity to model grid spacing, microphysics scheme, and homogeneous vs. heterogeneous initialization.