The High Resolution Rapid Refresh (HRRR) is a 3-km, convection permitting model, run hourly in real-time at the Global System Division (GSD) of the NOAA Earth System Research Laboratory (ESRL). The WRF-ARW-based HRRR is run out to fifteen hours over a domain covering the entire coterminous United States (CONUS), using initial and boundary conditions from an hourly-cycled 13-km mesoscale model, formerly the Rapid Update Cycle (RUC), and currently the Rapid Refresh (RR). Both the RR and RUC include a diabatic digital filter-based radar reflectivity data assimilation procedure to improve specification of the divergent component of horizontal wind in areas of precipitation.

In this presentation, we will include a brief evaluation of real-time HRRR forecasts covering the tornado outbreaks on 14-16 April 2011, 26-27 April 2011 and the Joplin, MO tornado on 22 May 2011. Deterministic forecast evaluation will include accuracy measures of timing, location and convective structure. Timing and location will be evaluated using multi-scale model reflectivity verification. The evaluation of convective structure will include diagnostic fields characterizing low and mid-level updraft rotation, graupel production and maximum 10 m wind speeds as compared with local storm reports.

Time-lagged ensemble HRRR forecasts will be evaluated as probabilistic forecasts in estimating the likelihood of severe weather phenomena including tornado production. The time-lagged ensemble probabilistic forecasts will be compared with non-tornadic supercellular events to evaluate ensemble skill in identification of tornado potential.

Finally, some preliminary evaluation of experimental HRRR forecasts leveraging radar data assimilation at the 3-km scale will be presented in comparison with the real-time HRRR forecasts of the tornado events.