Matthew B. Wilson
Valparaiso University, Valparaiso, Indiana
Matthew J. Bunkers
NOAA/National Weather Service, Rapid City, South Dakota
Although a probability of detection (POD) of 100% would be desirable for tornado warnings, the limitations of existing science and technology may mean that some tornado occurrences are currently not warnable, and a lower goal for POD would be more realistic, and may also help to reduce the rate of tornado false alarms. In this study, rotational velocity and circulation diameter time series are calculated from Level II radar data using GR2 Analyst for all tornadoes and tornado false alarms in the polygon warning era (post-October 1st, 2007) for the Rapid City National Weather Service Office. Environmental data from the Storm Prediction Center’s hourly mesoanalysis system are also examined for all events. These datasets are analyzed to determine which tornado occurrences are likely not warnable and to find if other useful differences between tornado cases and false alarms exist. It is found that a number of tornadoes (13-19) associated with rotational velocities below 5-10 m s-1 (20-30 kt) and mostly nonsupercellular storm modes were likely not warnable, and that if all other tornadoes in the dataset had been warned, a “practically perfect” POD for the Rapid City office would be between 77% and 86%. Although most environmental parameters did not show clearly discernable differences between tornadoes and false alarms, high values of convective inhibition were associated with false alarms cases more often than significant tornadoes, consistent with previous research. Additionally, the evolution and max values of Vr were clearly different between significant tornadoes and the rest of the dataset, in line with previous research supporting impact-based warnings.