Thanksgiving Storm Hype - Models Reverse

I posted the other day that the GFS was thinking about moderate snow in my area (central Pennsylvania) late on Thanksgiving, but the DGEX was convinced there would be none (other than flurries behind the front). Today, they have switched. The DGEX now predicts heavy snow will start in Pennsylvania and New York late Thanksgiving Day (see below), traverse the states, then shy away into northern New England.

THANGSGIVING EVENING

Again, forgiving the populated I-95 area which will see mostly rain, this could be a significant travel problem for people returning home late on Thanksgiving Day. But this morning's GFS keeps light snow behind the front well separated from the line of rain (and in fact both are so light they disappear at times). It's also a lot quicker, having the front move off the coast by Thanksgiving morning.

THANKSGIVING MORNING

How confident is the GFS of its own forecast that far out? Not very. If we look at the Ensembles, where they run the model multiple times with slightly different conditions, it's by no means sure of where the "540" line (often the rain/snow line at the surface) will be Friday morning (this is its lowest dip of the ensembles over Pennsylvania, though a couple think it could be in Canada or Virginia):

The exciting thing is that the NMM model can now see out to Wednesday afternoon, when the storm is forming in the Plains, which means that by the time I write this from work on Monday, we'll have the opinion from the NMM as well. Since it's the preferred short-term model, I'll be more likely to believe it than either the GFS or DGEX.