"Paul Hyett" wrote in message
...
On Tue, 1 Sep 2009 at 12:25:43, Dawlish wrote in
uk.sci.weather :

36 hours on and still looking good for a settled and warmer period of
weather next week, for England, at least. Good agreement between the
ECM and gfs. The MetO are coming round to the idea, but won't fully
commit themselves on today's 6-15 day update. I don't really blame
them. It's very easy to make mistakes forecasting at 10 days.

http://www.metoffice.gov.uk/weather/...t_weather.html

If only that set-up would wait 3 months...
--
Paul Hyett, Cheltenham (change 'invalid83261' to 'blueyonder' to email me)

Any forecast beyond 5 days needs to be taken with a huge pinch of salt at
this time of year, despite any apparent consistency. Hurricanes can develop
quickly and unexpectedly and have great potential for altering the strength
of jets and their orientation as they become extra-tropical. Also there is
at least one upper trough disruption before the forecast high pressure
period and models do not handle trough disruptions well. Confidence in an
anticyclone at this stage has to 50% at most I'd say.

Will
--

On Sep 2, 8:25*am, "Will Hand" wrote:
"Paul Hyett" wrote in message

...

On Tue, 1 Sep 2009 at 12:25:43, Dawlish wrote in
uk.sci.weather :

36 hours on and still looking good for a settled and warmer period of
weather next week, for England, at least. Good agreement between the
ECM and gfs. The MetO are coming round to the idea, but won't fully
commit themselves on today's 6-15 day update. I don't really blame
them. It's very easy to make mistakes forecasting at 10 days.

http://www.metoffice.gov.uk/weather/...t_weather.html

If only that set-up would wait 3 months...
--
Paul Hyett, Cheltenham (change 'invalid83261' to 'blueyonder' to email me)

Any forecast beyond 5 days needs to be taken with a huge pinch of salt at
this time of year, despite any apparent consistency. Hurricanes can develop
quickly and unexpectedly and have great potential for altering the strength
of jets and their orientation as they become extra-tropical. Also there is
at least one upper trough disruption before the forecast high pressure
period and models do not handle trough disruptions well. Confidence in an
anticyclone at this stage has to 50% at most I'd say.

Will
--

I don't know about 5 days, but any forecast beyond a week usually
needs to be taken with a huge pinch of salt at most times of the year
Will - especially when it involves forecasting a change of pattern; in
this case from very unsettled to very settled weather.

However, I have maintained for several years that there are times when
it is possible to forecast with 75% accuracy out to the far reaches
of 10 days, where forecasting usually becomes guesswork. I could be
wrong, of course (about 25% of the time), but I think this is one of
those times where it is possible. I haven't seen anything yet to
change my mind about the 10th and the MetO have certainly come around
to the same way of thinking as myself, about next week and the time
around the 10th in particular.

"Most parts are expected to become mainly dry, bright and warm around
the middle of next week" - today's 6-15 day tea-break precis! 8))

"Dawlish" wrote in message
...

However, I have maintained for several years that there are times when
it is possible to forecast with 75% accuracy out to the far reaches
of 10 days, where forecasting usually becomes guesswork. I could be
wrong, of course (about 25% of the time), but I think this is one of
those times where it is possible. I haven't seen anything yet to
change my mind about the 10th and the MetO have certainly come around
to the same way of thinking as myself, about next week and the time
around the 10th in particular.

"Most parts are expected to become mainly dry, bright and warm around
the middle of next week" - today's 6-15 day tea-break precis! 8))


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?

--
Ned

In article ,
Ned writes:

"Dawlish" wrote in message news:2d8c8221-e967-4149
...

However, I have maintained for several years that there are times when
it is possible to forecast with 75% accuracy out to the far reaches
of 10 days, where forecasting usually becomes guesswork. I could be
wrong, of course (about 25% of the time), but I think this is one of
those times where it is possible. I haven't seen anything yet to
change my mind about the 10th and the MetO have certainly come around
to the same way of thinking as myself, about next week and the time
around the 10th in particular.

"Most parts are expected to become mainly dry, bright and warm around
the middle of next week" - today's 6-15 day tea-break precis! 8))


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?


As far as 10 days out, are the Met Office doing anything very different
from interpreting the models? I can't see what else they could be doing.
Of course, they may well have access to data that Dawlish does not,
notably ensembles for some models that only make their operational runs
available to the general public, so that their forecasts ought to be
better than his.
--
John Hall "Do you have cornflakes in America?"
"Well, actually, they're American."
"So what brings you to Britain then if you have cornflakes already?"
Bill Bryson: "Notes from a Small Island"

On Sep 2, 8:34*pm, John Hall wrote:


As far as 10 days out, are the Met Office doing anything very different
from interpreting the models?
I can't see what else they could be doing.

--
John Hall * *

- Show quoted text -


Only in the sense that when Chopin sat down at the piano he was
merely interpreting Chopin.

Tudor Hughes, Warlingham, Surrey.

On Sep 2, 8:25*pm, "Ned" wrote:
"Dawlish" wrote in message

...

However, I have maintained for several years that there are times when
it is possible to forecast with 75% accuracy out to the far reaches
of 10 days, where forecasting usually becomes guesswork. I could be
wrong, of course (about 25% of the time), but I think this is one of
those times where it is possible. I haven't seen anything yet to
change my mind about the 10th and the MetO have certainly come around
to the same way of thinking as myself, about next week and the time
around the 10th in particular.
"Most parts are expected to become mainly dry, bright and warm around
the middle of next week" - today's 6-15 day tea-break precis! 8))

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?

--
Ned

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

"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

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.

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? *))

On Wed, 2 Sep 2009 at 08:25:40, Will Hand
wrote in uk.sci.weather :

If only that set-up would wait 3 months...

Any forecast beyond 5 days needs to be taken with a huge pinch of salt
at this time of year, despite any apparent consistency. Hurricanes can
develop quickly and unexpectedly and have great potential for altering
the strength of jets and their orientation as they become extra-tropical.

ISTM that even the most vigorous storms bounce off high pressure cells
like peas off an elephant, though...
--
Paul Hyett, Cheltenham (change 'invalid83261' to 'blueyonder' to email me)