Two primary objectives in severe weather forecasting are to determine the synoptic precursors to severe weather outbreaks and to distinguish the type of outbreak, tornadic or primarily nontornadic, given these precursors. Since it is not possible to resolve tornadoes explicitly, meteorological covariates are used to aid in the categorization of outbreak type. Only recently has computer technology been sufficient for the use of mesoscale models at high resolutions (storm scale) to generate these covariates in an appropriate manner. Such covariates include storm-relative helicity, bulk shear, and lifted condensation level.

In this study, 50 tornado outbreaks and 50 primarily nontornadic severe weather outbreaks were selected based on previous studies. These cases were run using the Weather Research and Forecasting (WRF) model 24, 48, and 72 hours in advance of the synoptic time most closely associated with the outbreak valid time. NCEP Renanalysis was used for the initial synoptic-scale input for these forecasts.

The covariates found using permutation testing and their promising value in distinguishing outbreak type will be shown for all three forecast periods. Preliminary verification statistics of the forecasts will also be presented. Finally, plans for future work will be discussed.