Curious if they use NWS weather station data or if they collect their own.

It is pretty much an independent system. There are only a handful of organizations on the planet operating an independent global weather model. For historical reasons the US operates two of them. A good argument can be made that the European model is better than either, so it might make sense for the US to combine some of those global weather model resources.

unless "all private except certain secure friends" is not in the spirit of US Federal Science?

Isn't it true that open methods beat closed expensive methods many times.. in the case of weather prediction, it is trivial to benchmark predictions.

For how long? Like seriously, how long until that department/company/office is shut down