The good thing with that is that's not necessarily something the CAMs should have a good handle on at this range I don't think. Probably should lean on the globals until they lock in on a synoptic scale setup, and then use the mesoscale models for localized features once things are settled up top.