On Jun 19, 10:58*pm, "Jon O'Rourke" wrote:
wrote in message

...

Have to say that the current forecast on meto.gov.uk looks a little
more pessimistic than I would expect, particularly from the GFS chart
but even the Met Office's own chart. My interpretation of the GFS
chart for central southern England would be bright and warm for much
of the day, with light rain after about 7pm. Even the Met Office chart
would suggest to me only light/sporadic rain during daylight hours,
with pressure reasonably high at around 1016mb. Is there some other
factor here not evident on the pressure charts?

This Saturday I hold an annual outdoor event which almost always is
adversely affected by the weather (4 years out of 7 so far - the late
June Wimbledon/Glastonbury curse....) so if it happens again I think I
will move it permanently!

Nick

Well, the forecast for the SE issued this afternoon mentions "Cloud and
outbreaks of rain spreading northwards on Saturday" - and it may well be
that it brightens up in the afternoon.

Interesting to note the amount of press the GFS continues to get. While it's
routinely scrutinised by most forecasters it's far from being the best
model, both in terms of deterministic and ensemble output, at least in IMHO.
I guess the availability, frequent updates and generally good presentation
on the web have made it the model of choice for many.

Jon.

Jon, Tudor,

The only reasons I choose the gfs to use in my forecasting is that it
has output beyond T240 and output is produced 4 times a day. If
another model did the same, I'd use it and I agree with everything Jon
says. The NOAA comparison figures for the last month clearly show that
to 6 days.

5 days http://www.emc.ncep.noaa.gov/gmb/STATS/html/acz5.html
6 days http://www.emc.ncep.noaa.gov/gmb/STATS/html/acz6.html

Archives: http://www.emc.ncep.noaa.gov/gmb/STATS/STATS.html

The archives show the same. Tudor - the ECM model comes out as being
consistently the best. GFS and the UKMO model perform reasonably
equally over time at 5+6 days, perhaps the UKMO shades second place.

After 6 days, I feel that accuracy declines considerably and swiftly.
I do wish NASA would provide comparison figures for 10 days, but they
don't; so we have to speculate. I feel there would be little
difference and we'd be down well below 0.5, in terms of correlation -
towards "fairly weak" in correlation test terms. It's only a guess
though. At 10 days, I don't think any perform that well.

Paul