There was a noticeable spike in ProbSevere in LMK – between 1938UTC (6%) and 1940UTC (66%) then back down to 7% at 1950UTC.  Satellite rates were n/a – and the KLVX radar was in near optimal sampling distance (50 nm) from the storm of interest.

So – we are unsure why the ProbSevere spiked?  This would be a good example to pose to the developers.  I attached the 1938 and 1940Z images below.