"Keith (Southend)" wrote in message
...
Out of interest what times do the GFS model update on the internet?

Hi Keith,

Take the 00z GFS as an example. The data is collected at 00z (duh!). then
the usual order of events is as follows :-

02:48 BUFR input files are loaded
02:56 Datasets are analysed and rogue or erroneous records are removed or
corrected
03:14 GFS model starts running
03:24 Metcheck/WetterZentrale etc start receiving the first of the 100
odd GRIB files for T+3
04:09 GFS complete

As for the current variability of the GFS, I was under the illusion that
there had been a problem with some of the BUFR files, but this turned out to
be a problem with the US AWIPS which is simply a way of viewing the output.
I spoke to NOAA today who confirmed that the GFS has had no problems with
data input, they praised the UKMO model for it's consistency and the HPC
forecasters simply put the GFS wobblies down to the variable weather
pattern.

Cheers,

A