High resolution analyses of the 8 May 2003 Oklahoma City storm. Part II: EnKF data assimilation and forecast experiments

The Ensemble Kalman Filter (EnKF) is currently being tested as a method for initializing storm-scale numerical forecast models. After initial testing in Observing System Simulation Experiments (OSSEs), the EnKF technique is now ready for testing in real-data cases. We have chosen the 8 May 2003 Oklahoma City storm as a test case because it was observed by the rich instrument network in central Oklahoma. During the afternoon of 8 May 2003, several convective cells formed along a dryline in west-central Oklahoma. One of these cells eventually developed into the supercell that produced an F4 tornado in Oklahoma City.

We are currently assimilating Doppler velocity and reflectivity observations from the KOUN 10-cm research radar into the NSSL Collaborative Model for Multiscale Atmospheric Simulation (NCOMMAS). We employ ensembles with ~50 members and model grids with ~1-km horizontal grid spacing. The focus of the experiments is to determine how well the development and movement of the Oklahoma City storm can be predicted if the model is initialized with observations during only the “first-echo” stage of the convective cells. We use the EnKF to assimilate approximately 5 volumes of radar data over a period of 30 min into NCOMMAS. Then, we produce 60- min forecasts initialized with the results of the EnKF assimilation.

Preliminary results are so far encouraging – both the growth of the Oklahoma City storm and the weakening of other nearby cells are predicted. At the conference, we will discuss how the assimilation and prediction results depend on a number of factors, including the model grid resolution, the method for populating the initial ensemble, the number of radar-data volumes that are assimilated, and the model microphysical parameterizations.