Thanks for digging in Dave.  I am definitely curious what is going on in your specific cases here (something for the URMA folks to address). But I do want to mention that the MinT analysis from URMA for this night was very similar to MatchObsAll (bottom two plots). MOA presumably matches all the cold stations that you discuss.  Since the final analysis looks good, URMA has probably been a good source for MaxT/MinT bias-correction.  The NBM forecast minimum temperatures in this case were closer to the observations than NDFD at most stations (top two plots).  I'm not sure why the hourlies you were looking at were so different, but clearly the final answer from URMA was a good one.