Geplukt van het TWO-forum
Well, the GFS is really three models, AVN out to 84 hours, GFS out to 180 hours then MRF from 180 to 360. Last year NCEP decided to combine the three but they are still three seperate models with three seperate GRIB output files and these things take time I guess.
The AVN is a great model, it's a medium resolution 191km model which runs out to 84 hours in 3 hour time steps.
The GFS is also great, it too is a medium resolution 1 degree model which runs out to 180 hours in 3 hour time steps.
The MRF is a bit ropey. It runs from 0 to 360 hours in 12 hour timesteps but at a 2.5 degree resolution. It's really there for trends and stuff but can be way out sometimes.
The MRF is what you guys call Fantasy Island.
So, to forecast say a 5 day cold period (not a snap) I'd look at the MRF part of the GFS (180-360 hours) to see whether there are any major patterns and also how these patterns differ from yesterdays forecasts.
Only once they fall into the GFS domain (0-180 hours) should these things be taken seriously as severe events have so many variables that they need a decent resolution model to take it down to higher detail.
Quote selectie