In the 5-7 day range, I'm actually more curious to see the (parallel) GFSv16 than the current GFS (v15) for this Jan 8-10 system. Been some talk about it with a fantasy storm out in the 200 hr range, but everything deterministic that far out is pretty dubious. For the medium range, the GFSv16 has an extensive verification page that I encourage people to check out:
They have a powerpoint (under "Information") there that walks through a few case studies/examples of winter storms, and the main conclusions the NOAA scientists draw are:
- 500mb verification scores improved marginally (though tbh I sorta doubt how useful of a metric this is)
- In a few case studies, the GFSv16 greatly improved upon the progressive bias of the current GFS model
- The powerpoint also claims that the v15 has a warm bias at 850mb in the medium range. All I can say is I wish! I've never heard of it for our area; they use an example from the Midwest, so maybe this is a regional thing (see map). But this provides a reasonable argument that maybe the preceding airmass on the GFSv15 is too warm (it has been trending colder), which would support more widespread snow as long as the low track does not change.
Anyway, due to the window in which they tested the GFSv16, many of the example storms are from late fall, and focused more on the Midwest or NE. Now would be a good time to see how it performs for a storm in the SE, since the v15 is slated to be replaced in early February.