Here’s a case where the ProbTor model significantly increased in an area where tornado potential was extremely limited. Below are animations of the radar velocity at the closest radar (KGRK), the low-level Merged AzShear, and the time series of the ProbSevere product. The cell in question was small and behind the main line of storms in the more stable airmass, yet the ProbTor model jumped to 79% for several minutes. The time series seemed to indicated this was due to an increase in AzShear yet the Merged Low-Level AzShear product showed no distinct signature (same for Mid-Level AzShear not shown).

UPDATE:

After consulting with the fantastic Mr. Cintineo, he found that the false flags were likely from a jump in the Low-Level Rotation tracks in MRMS (shown below). These false signatures, however, were not making it into the aforementioned AzShear product.

#ProtectAndDissipate