On Dec 11, 8:43*pm, Dave Cornwell wrote:
wrote:
On Dec 11, 5:52 pm, Dave Cornwell wrote:
how accurate (in %age terms) is the GFS at 120h ?
-----------------------------
Who knows for sure? 60% ??
Dave

The secret with GFS is not to take any particular frame literally, but
to monitor the time period you are interested in on a run by run
basis. The general trend on the previous three runs is for an intense
low to be somewhere over northern Scotland between 18Z Thursday and
06Z on Friday.

The 12Z run has the centre a little further south, but the low centre
continues to be forecast at below 950mb (and maybe 940 for a time).

Therefore I would say that the accuracy on a slightly larger spatial/
time scale is higher than 60%, maybe as high as 80 in this case.

---------------------
I think the answer should have been accuracy of what, really? e.g,
position of low, wind speed, temperature. So many factors all with
different levels of accuracy.
Dave- Hide quoted text -

- Show quoted text -

That's true. The stats from NCEP are determined by the accuracy of the
500hpa contour chart, I believe.

Fred; as far as that goes, model accuracy could be judged as pretty
good at 5 days. gfs has performed well recently, but ECM is almost
always ahead, averaged over a month.

Dave's right though. What really counts is the weather one experiences
on the ground. Now the accuracy there is a whole different kettle of
fish!