Statistical Comparison and Simulations of Supercells in Environments with Varying Significant Tornado Parameters

A Rapid Update Cycle-2 (RUC-2) dataset of proximity soundings from nontornadic and tornadic supercells was categorized based upon the following significant tornado parameter (STP) thresholds: STPs greater than 1, 2, 3, 4, and 5, as well as STPs between 1 and 10, and between 5 and 10. Though in a forecasting setting STPs greater than 1 generally distinguish between nontornadic and significantly tornadic (F2 or greater) supercells, it was determined that 109 out of 443 nontornadic supercell cases in the database have an STP value greater than 1. This study applied the self-organizing map (SOM) statistical clustering technique to vertical profiles of four variables—temperature, dewpoint, and ground-relative u and v wind components—for each STP threshold. These SOMs were compared to others with only u and v wind components considered. Additionally, SOMs which were created after the data had been normalized by variable and height were compared to SOMs with unaltered data to determine if either was more effective at distinguishing between storm type. It was found that the wind-only and non-normalized SOMs slightly outperformed their counterparts. Composite near-storm environments from each node of the best-performing SOM with the STP threshold between 1 and 10 were then used as the base-states for a set of idealized numerical simulations of supercells to ascertain the extent to which characteristics of the resulting storms (e.g., updraft strength, near-ground vertical vorticity) resembled the primary storm type (tornadic or nontornadic) associated with each node.