On Jan 28, 12:20*pm, "Martin Rowley"
wrote:
... For the last two months, I've been verifying the GFES ensemble
extremes output available he-http://www.met.rdg.ac.uk/~brugge/forecast/
[ I'll be doing a full analysis sometime late February and posting it
to Roger so he can review and publish as he sees fit.]

Today's issue has the following minimum & maximum values for Heathrow
for the 9th to 12th February inclusive:-

MIN(18-06Z): -4, -5, -5, -6

MAX(06-18Z): 0, 0, -1, -1

For Hurn, same period

MIN: -6, -6, -7, -6

MAX: 0, 0, -1, 0.

For such low daytime maxima, the model must think there is a bit of
snow on the ground!

These are based on the ensemble output, NOT the deterministic (or
operational) output.

It'll be interesting to see how they verify! One word of caution
though, my preliminary assessment of these data is that when a *major*
change of type takes place, the scheme has difficulty in settling down
to achieve the 'correct' solution, often not sorting itself out until
5 or 6 days before the event; this isn't a surprise to me, as
ensembles, though useful, shouldn't be regarded as the 'Holy Grail' of
forecasting - there's always that elusive Amazonian butterfly to play
its part ;-)

Martin.

--
Martin Rowley
West Moors, East Dorset (UK): 17m (56ft) amsl
Lat: 50.82N * Long: 01.88W
NGR: SU 082 023

I always thought Martin, that ensembles at least in part take care of
uncertainties in inputs, caused by for example butterflies and
elephants.
What they do not help with as regards NWP, are the errors involved in
finite differencing of the differential equations, and
parameterisations of all sorts of feedbacks at various scales, not
least the subgrid scales.
As you quite rightly say, beyond 5-6 days uncertainties grow
alarmingly, particularly in certain situations.
It also depends on how demanding the consumer is as regards the
forecast. A few 100 km can mean a lot to Jo public.

Len
Wembury, SW Devon