On Sep 3, 8:11*am, Dawlish wrote:
On Sep 2, 10:54*pm, "Ned" wrote:





"Dawlish" wrote in message

....

I am intrigued by your claim of making a "forecast". Are you not simply
model watching, as many of us here do, and when there is agreement at 10
days up pops your forecast. To claim the MO have come round to the same
way
of thinking as yourself is plainly ridiculous, when you don't actually
"make" a forecast, only interpret the models. Or am I missing a skill you
also have?

Only judge forecast accuracy by outcome percentage accuracy over time
Ned.

But you only "forecast" once in a blue moon,on the rare occasions when
models are basically in agreement at 10 days.What you seem to be suggesting
is that when they agree at 10 days out, there is a 75% chance they'll be
correct. *What skill do you add to the forecast method beyond highlighting
the agreement?

--
Ned

The MetO can't forecast with any reasonable accuracy, on a daily basis
what's going to happen at 10 days. Nor can anyone else; nor can I. At
present it is impossible, on a daily basis, with reasonable accuracy.
Thus, when you see that 6-15 day forecast, you have no idea if it will
be accurate, or how accurate the MetO thinks it is likely to be. It
doesn't give you an idea of how accurate they feel it will be at 10-
days plus. I wish they would and I feel it would be useful. The 6-15
dayforecast is just 120-ish words which may, or may not, be correct.
There is no indication within that of how confident the forecasters
are at those distances.

No blame in that. It's just not possible to forecast with accuracy at
that distance on a daily basis, but the MetO is under these enormous
public expectations to forecast at all distances - and be correct -
and the press office did it no favours by heightening those
expectation with some silly soundbites to "explain" their summer
forecast to the media. It fulfils the public expectation at medium
ramge, but does so with a minimum of fuss and in few words in a tiny
area of the site which is the 6-15 day forecast. It then ignores
whatever the outcome is and never returns to it and never publishes
verification figures, meaning that we have absolutely no idea whether
any forecast at 10 days is accurate. That camouflages the forecasting
difficulties very well and is why I joke about precis competitions in
morning tea-breaks.

I've developed a system where I can, on occasions (and only on
occasions) forecast at 10 days, with 75%+ accuracy. Every forecast
I've made with it (71) has been done in the public eye and judged by
others, as well as by myself and the hrashest judge has been me;
promise you. On that dataset and to that outcome accuracy, it works.
The only way to apply that is to model watch at 10-days really
assiduosly. I used to base it solely on consistency in gfs runs, but
I've tweaked that to include a need for agreement from the ECM. That's
not because I feel the gfs is the better model, it's just that the 4
runs per day, each under different peramaters renders consistency less
likely, if the outcome is to be correct. Any wobbles in that and the
forecast is rendered useless. I cannot forecast on a daily basis using
this and the system can't be used that way.

When a pattern is established; e.g. a zonal pattern, or an
anticyclonic pattern, forecasting at 10 days becomes easier, but
recently, I've done less forecasting at those times, being content to
let the pattern run and look for a change instead. Here, I said, on
31st Aug, that I was 75+ certain there would be a change to
anticyclonic weather on 10th Sept (T240 on the 31st.) and that dry and
settled weather, with above average daytime temps will affect at least
England. That will be a big pattern change and will be one not seen
since late-June/Early July. *7 days to go now and I haven't changed my
mind. I'd be happy for you to judge it at outcome.- Hide quoted text -

- Show quoted text -

Looking very good for a period of calm, settled weather as
anticyclonic influence begins to dominate the UK's weather from the
middle of next week and daytime temperatures should be warm, for
many.

It could well hang around too and tere's plenty of consistency and
model agreement, to say that in 10 days time at T240, much of the UK
will still be under the influence of high pressure. I agree with the
MetO that mist and fog could be an early morning problem in a few
areas, but most of us will be enjoying plenty of sunshine and warm
temperatures in the daytime towards mid-month. The only hiccups to the
sunshine are likely to be Atlantic fronts, dying into the high
pressure and introducing cloud to some areas from time to time.

Some nice weather coming up. A BBQ summer in September? *))