Evaluating the Accuracy and Skill of the SPC's Significant Severe Convective Outlooks

In the early 2000’s, the Storm Prediction Center (SPC) added probabilistic convective outlooks to their suite of regular forecasts available to the public.  Included in these new products were forecasts of “significant severe” events: tornadoes of magnitude E/F2 or greater, hail of at least 2 inches in diameter, and wind gusts of 74 mph or higher.  In this study, the accuracy and skill of these forecasts of high impact severe weather events are assessed during 2005–15 for each individual hazard in the SPC’s day 1 forecasts, as well as for all hazards combined in the day 2 and day 3 forecasts.  Special attention is paid to the frequency at which these forecasts are skillful, and whether data suggest that forecasts of significant severe events should be made more frequently.  The latter is achieved through the use of practically perfect forecasts, which are designed to simulate forecasts that would made by a forecaster with full knowledge of the event beforehand, and are created by applying a 2-D Gaussian smoother to severe reports.  A method by which practically perfect forecasts are used to determine when a forecast of significant severe weather would have been justified, especially when no outlook was issued, is demonstrated.  These high impact severe weather forecasts will also be compared to the SPC’s standard probabilistic convective outlooks for differences in skill between the two forecasts.