High-Resolution Rapid Refresh Prediction of Tornadic Supercells in the U.S. Southern Plains During May 2013
The HRRR is run hourly out to fifteen forecast hours over a domain covering the entire conterminous United States using initial and boundary conditions from the hourly-cycled RAP and is available in real-time to operational forecasters in both the private and public sectors. The hourly updating HRRR forecasts provide a measure of forecast likelihood in the form of run-to-run consistency that can be translated into probabilities of a particular weather hazard using a time-lagged ensemble with neighborhood spatial and temporal filters of diagnostic fields.
In this presentation we will evaluate HRRR forecasts of several tornadic supercell events that occurred in the southern plains of the U.S. in May of 2013 with significant impacts to communities such as Granbury, Texas on 15 May; Shawnee, Oklahoma on 19 May; Moore, Oklahoma on 20 May; and El Reno, Oklahoma on 31 May. In this evaluation we will focus on both deterministic and probabilistic HRRR forecasts. We will present both HRRR reflectivity and updraft-helicity fields as measures of forecasted convective structure to highlight the skill of the deterministic forecasts in accurately identifying regions of observed supercells.
We will also evaluate the model-forecasted local environmental kinematic and thermodynamic parameters including vertical wind shear in the lowest kilometer, low-level storm-relative helicity, most unstable parcel level, lifting condensation level and CAPE/CIN to determine the potential to discriminate between non-tornadic and tornadic supercells in the HRRR forecasts. We will synthesize both the explicitly forecasted supercell structures and the forecasted local environmental information in a time-lagged ensemble to produce an hourly probabilistic estimate of the tornado potential across regions containing numerous supercells. These probabilistic estimates will be compared with the observed tornado reports to determine the accuracy of the hazard estimation in both location and time.