Peter Duniho wrote:

Three or four days in advance? Forget it. It is impossible to KNOW what
the weather will be that far in advance.

No, don't forget it. Yes, it's impossible th *know*, but there are some
astonishingly accurate mathematical models out there. Depending on the
situation, their predictions are valid for several days, even
cloudbasewise, and they are becoming better as the models are refined
and the computing power grows. Of course, you'll never *know* in advance.

An example of a usually very accurate two days prediction (yes, only two
days, but they are working on it) can be found he
http://www.unibas.ch/geo/mcr/3d/mete...nmm2/BASEP.htm

I don't know an easily accessible source for such predictions concerning
the USA, but I would asssume that they exist.

Stefan