Description of a WRF-based Ensemble Data Assimilation System for Warn-on-Forecast

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Tuesday, 4 November 2014: 9:00 AM
Madison Ballroom (Madison Concourse Hotel)
Kent H. Knopfmeier, CIMMS/Univ. of Oklahoma, NOAA/NSSL, Norman, OK; and D. M. Wheatley, T. A. Jones, and G. J. Creager

One of the primary goals of the NOAA Warn-on-Forecast (WoF) research program is to provide accurate short-range (0-1 h) probabilistic forecasts of severe convective storms. A WoF storm-scale ensemble data assimilation (DA) system would comprise a very-high-resolution (O 1-km or less), event-dependent grid with a convection-allowing (grid spacing ~ 3-km) ensemble serving as a background. Preliminary work toward such a system has sought via the assimilation of conventional, radar, and high-resolution satellite observations to produce 3-km ensemble grids that contain both realistic mesoscale (i.e. drylines) and convective (i.e. supercells/MCSs) features.

The ensemble DA system currently under development at NSSL employs the Weather Research and Forecasting (WRF) model in a one-way nested framework with a 3-km inner domain nested within a 15-km parent domain, which covers the continental United States. Members of the 0000 UTC Global Ensemble Forecast System (GEFS) forecast cycle provide initial and boundary conditions. Initial soil and sea-surface temperature (SST) data at 0000 UTC are provided by the Earth System Research Laboratory-Rapid Refresh (ESRL-RAP, i.e. RAPv2) model and the Real-Time Global SST High-Resolution analysis. The WRF physics options are also varied amongst the ensemble members to fully populate the ensemble and address deficiencies in the model physics. Routinely available observations of altimeter setting, temperature, dewpoint, and horizontal wind components from land and marine surface stations, Mesonets, rawinsondes, and aircraft as well as satellite winds are assimilated using an ensemble Kalman filter (EnKF) contained within the Data Assimilation Research Testbed (DART) software package at hourly intervals on both domains simultaneously until just prior to convection initiation. Storm-scale DA of radar and high-resolution satellite observations then begins on the 3-km inner domain only for the remainder of the convective episode.