On Sep 18, 8:41*am, "Col" wrote:
Gavino wrote:
"Lawrence13" wrote in message
....
Only time will tell Paul, I mean do you know how hard it is to
forecast ten days ahead? Oh of course you do cos you got it all
wrong.

And you, I suppose, did not ????

Of course some of Dawlish's forecasts will turn out to be 'wrong'.
He does not claim to be a fortune teller.
As I understand it, he claims 80% confidence in his forecasts.
The whole point of his methodology is that is a probabilistic one (as
indeed must be any 10 day forecast, if the producer is honest about
it).
Its usefulness cannot be judged on a single forecast, but on the
overall success rate, which will never be 100%.

An 80% success rate from forecasts only made when they are
judged as being most likely to be correct. This isn't always made clear.

Unfortunately the Met office can't duck out of making forecasts
when it's too difficult to call, and they can't just say 'don't know'
even if that is a reasonable assessement of the situation
--
Col

Bolton, Lancashire
160m asl- Hide quoted text -

- Show quoted text -

In all seriosness Col I truly wish the Met Office would say that
sometimes e.g we just don't know, there could be heavy falls just
about anyway but then again some could see nothing. I'd much prefer
that to the blanket severe weather warnings that go out every time we
have potential severe weather alerts.

Lawrence13 wrote:
On Sep 18, 8:41 am, "Col" wrote:

Unfortunately the Met office can't duck out of making forecasts
when it's too difficult to call, and they can't just say 'don't know'
even if that is a reasonable assessement of the situation
--
Col

Bolton, Lancashire
160m asl- Hide quoted text -

- Show quoted text -

In all seriosness Col I truly wish the Met Office would say that
sometimes e.g we just don't know, there could be heavy falls just
about anyway but then again some could see nothing. I'd much prefer
that to the blanket severe weather warnings that go out every time we
have potential severe weather alerts.

To be fair they quite often do say that kind of thing wrt heavy showers.
You may catch one and it could be heavy, or you could just as easily
end up staying dry. This is effectively a 'don't know' for any given
location, but then so is a probabalistic forecast of '50% chance of
showers'.

I was thinking more along the lines of the monthly forecasts where
they know full well the models are all over the place at 10 days but
they are still committed to making a forecast.
--
Col

Bolton, Lancashire
160m asl

On Sep 18, 8:41*am, "Col" wrote:
Gavino wrote:
"Lawrence13" wrote in message
....
Only time will tell Paul, I mean do you know how hard it is to
forecast ten days ahead? Oh of course you do cos you got it all
wrong.

And you, I suppose, did not ????

Of course some of Dawlish's forecasts will turn out to be 'wrong'.
He does not claim to be a fortune teller.
As I understand it, he claims 80% confidence in his forecasts.
The whole point of his methodology is that is a probabilistic one (as
indeed must be any 10 day forecast, if the producer is honest about
it).
Its usefulness cannot be judged on a single forecast, but on the
overall success rate, which will never be 100%.

An 80% success rate from forecasts only made when they are
judged as being most likely to be correct. This isn't always made clear.

Unfortunately the Met office can't duck out of making forecasts
when it's too difficult to call, and they can't just say 'don't know'
even if that is a reasonable assessement of the situation
--
Col

Bolton, Lancashire
160m asl- Hide quoted text -

- Show quoted text -

I always make that clear Col. Always. You must have seen this before,
but I'll repeat it.

I believe the times when accurate forecasts can be made at 10 days is
limited, however, I also believe there are times when there it is
possible to recognise consistency and agreement between 2 main models
and I use that to make these occasional forecasts with 80% confidence.
Of course the MetO can't "duck out" of making forecasts. That's why
they don't bother making any accuracy data public and make the
forecast deliberately short and over a range from 6-15 days. An
approach involving telling people when their confidence in a forecast
is higher at 10 days would be far preferable to what we have now.
People have very little idea of whether what they read on the 6-15 day
forecast is likely to be correct, or not, under the present system.
Most probably don't really care, as a result and dismiss them as
unlikely to be correct.

The MetO knows the difficulties of forecasting at 7 days+, as we all
do. I just wish they'd recognise this and engage their public more
effectively by admitting that their are times when their forecasts are
unlikely to be correct and explaining why. Why can't they say "We
don't know" when they actually don't know?? Isn't that better that
leading the public on by issuing a forecast that they simply can't be
sure of how accurate it's likely to be? What's the actual use of that
6-15 day forecast when it gets to 1 week distance? A general idea?
Is that the best the MetO can do? Sometimes, I believe, you can have a
far better idea than that and I'll bet there are times when The MetO
forecasters are far more confident at 7-10 days than at other times.
The general public simply isn't told that and I don't think that is
good enough.

It's time for a change of approach from the MetO to probablistic
forecasting at 7 days+.

PS Hang on, I notice agree with Lawrence here (though this isn't about
severe warmings in this particular instance). That doesn't happen
often, but on this count, I think we are expressing a wish that many
people would have of the MetO.

Admit when you don't actually know and express your uncertainty in
ways with which the public can engage, is what I feel (and what I've
told them).

Dawlish wrote:
On Sep 18, 8:41 am, "Col" wrote:


I always make that clear Col. Always. You must have seen this before,
but I'll repeat it.

You usually do, but there have been a few instances recently
where you have not.
On this very thread for example you were talking about the number
of 'correct' forecasts you made and your 'headline' 80% success rate.
No mention was made of the fact that you only make these forecasts
when the models are performing favourably for you.

To the uninitiated, that is very misleading.
--
Col

Bolton, Lancashire
160m asl

i think scotland yard call it "fraud"

On 18/09/2011 12:38 PM, Col wrote:
that you only make these forecasts
when the models are performing favourably for you.

To the uninitiated, that is very misleading.

On Sep 18, 12:38*pm, "Col" wrote:
Dawlish wrote:
On Sep 18, 8:41 am, "Col" wrote:

I always make that clear Col. Always. You must have seen this before,
but I'll repeat it.

You usually do, but there have been a few instances recently
where you have not.
On this very thread for example you were talking about the number
of 'correct' forecasts you made and your 'headline' 80% success rate.
No mention was made of the fact that you only make these forecasts
when the models are performing favourably for you.

To the uninitiated, that is very misleading.
--
Col

Bolton, Lancashire
160m asl

If I miss every now and again, I apologise. It's like putting warning
notices up saying "steep drop" every 20 metres on the cliff edge, in
case someone misses it. The implication that you are likely to fall
and break your neck is always there, notice, or not. *))

On Sep 18, 12:38*pm, "Col" wrote:
Dawlish wrote:
On Sep 18, 8:41 am, "Col" wrote:

I always make that clear Col. Always. You must have seen this before,
but I'll repeat it.

You usually do, but there have been a few instances recently
where you have not.
On this very thread for example you were talking about the number
of 'correct' forecasts you made and your 'headline' 80% success rate.
No mention was made of the fact that you only make these forecasts
when the models are performing favourably for you.

To the uninitiated, that is very misleading.
--
Col

Bolton, Lancashire
160m asl

Oh. Also, the models don't perform favourably for me, or anyone. They
are just released several times a day.

yellow belly **** poor coward fraudster

also apply

now go way , you lost, loser

On 18/09/2011 1:47 PM, Dawlish wrote:


Oh. Also, the models don't perform favourably for me, or anyone. They
are just released several times a day.

"Col" wrote in message
...
Gavino wrote:
Of course some of Dawlish's forecasts will turn out to be 'wrong'.
He does not claim to be a fortune teller.
As I understand it, he claims 80% confidence in his forecasts.
The whole point of his methodology is that is a probabilistic one (as
indeed must be any 10 day forecast, if the producer is honest about
it).
Its usefulness cannot be judged on a single forecast, but on the
overall success rate, which will never be 100%.

An 80% success rate from forecasts only made when they are
judged as being most likely to be correct. This isn't always made clear.

Unfortunately the Met office can't duck out of making forecasts
when it's too difficult to call, and they can't just say 'don't know'
even if that is a reasonable assessement of the situation
--
Col

Why not? Surely that would be the honest thing to do?
Otherwise the 'forecast' is just a guess (even if an educated one), and has
little value since no-one knows how much confidence to put on it.
It would be more useful if they could say something like "Current
atmospheric conditions are such that, at 10 days, the weather could do X, Y
or Z, with probabilities A, B and C, which we will refine as we get nearer
the time".

Dawlish wrote:
On Sep 18, 12:38 pm, "Col" wrote:
Dawlish wrote:
On Sep 18, 8:41 am, "Col" wrote:

I always make that clear Col. Always. You must have seen this
before, but I'll repeat it.

You usually do, but there have been a few instances recently
where you have not.
On this very thread for example you were talking about the number
of 'correct' forecasts you made and your 'headline' 80% success rate.
No mention was made of the fact that you only make these forecasts
when the models are performing favourably for you.

To the uninitiated, that is very misleading.
--
Col

Bolton, Lancashire
160m asl

Oh. Also, the models don't perform favourably for me, or anyone. They
are just released several times a day.

By the models performing favourably for you I meant you were able,
on occaison, to make forecasts as you describe he

"I believe the times when accurate forecasts can be made at 10 days is
limited, however, I also believe there are times when there it is
possible to recognise consistency and agreement between 2 main models
and I use that to make these occasional forecasts with 80% confidence."
--
Col

Bolton, Lancashire
160m asl