Evaluation of WRF forecasts of tornadic and nontornadic outbreaks occurring in the spring and fall when initialized with synoptic-scale input

Recent studies have suggested that the Weather Research and Forecasting mesoscale model (WRF) is able to discriminate tornado outbreaks from primarily nontornadic outbreaks consistently at least 72 hours in advance of their occurrence and when initialized with coarse, synoptic-scale data. However, many primarily nontornadic outbreaks occur during the summer months, when tornado outbreaks rarely are observed. It is reasonable to hypothesize that the WRF may not be able to distinguish the outbreak types if the list of cases is limited to those outbreaks occurring in the spring and fall seasons, when both types of outbreaks are common.

In this study, WRF simulations of 50 tornado outbreaks and 50 primarily nontornadic outbreaks are conducted, using the National Centers for Environmental Prediction (NCEP)/National Center for Atmospheric Research (NCAR) reanalysis datasets. Objective and subjective analyses of the 24-hour forecasts, in comparison to previous research in which the primarily nontornadic outbreaks were not limited to the spring and fall seasons, indicate that there is only limited degradation in the model's ability to discriminate outbreak types when limiting the cases to those occurring in the spring and fall. These results suggest that synoptic-scale processes play a substantial role in the occurrence or absence of tornado outbreaks. The analysis of these cases also indicates that the outbreak types frequently consisted of noticeably different synoptic and subsynoptic environments. Finally, low-level storm-relative severe weather parameters proved to be especially helpful in discriminating outbreak types. A discussion of the utility of various severe weather parameters to distinguish outbreak types in this study will be presented.