I'm trying to understand why the Met Office forecasts are so
consistently wrong, and keeps changing from hour to hour sometimes !?

Last weekend - Easter - there were constantly changing forecasts for
Leeming from rain to not and back again... in the end there was no
rain to speak of - though there were storms over the Moors...

This seems to have been happening for nearly a year now, and it's very
irritating
It's not just MetO but most of the other sources too... Metcheck/BBC/
Wunderground etc...

I know it's not an exact science, but it use to be a lot more accurate
and consistent in its accuracy...

Anyone willing to try to give a bit of a layman's terms explanation ?

Nick

On Tue, 26 Apr 2011 14:36:26 -0700 (PDT), NickTheBatMan
wrote:

I'm trying to understand why the Met Office forecasts are so
consistently wrong, and keeps changing from hour to hour sometimes !?

Last weekend - Easter - there were constantly changing forecasts for
Leeming from rain to not and back again... in the end there was no
rain to speak of - though there were storms over the Moors...

This seems to have been happening for nearly a year now, and it's very
irritating
It's not just MetO but most of the other sources too... Metcheck/BBC/
Wunderground etc...

I know it's not an exact science, but it use to be a lot more accurate
and consistent in its accuracy...

Anyone willing to try to give a bit of a layman's terms explanation ?

Nick
--

Weather Forecasters that cannot forecast, plumbers they call leaky,
builders they call Bob, Motor Technicians that are grease monkey's,
Bankers that pay bonuses for failure, the list is endless - have you
forgotten that this Country has been in decline for nearly 200 years ?

On 27/04/2011 00:56, It's True wrote:
On Tue, 26 Apr 2011 14:36:26 -0700 (PDT), NickTheBatMan
wrote:

I'm trying to understand why the Met Office forecasts are so
consistently wrong, and keeps changing from hour to hour sometimes !?

Last weekend - Easter - there were constantly changing forecasts for
Leeming from rain to not and back again... in the end there was no
rain to speak of - though there were storms over the Moors...

This seems to have been happening for nearly a year now, and it's very
irritating
It's not just MetO but most of the other sources too... Metcheck/BBC/
Wunderground etc...

I know it's not an exact science, but it use to be a lot more accurate
and consistent in its accuracy...

Anyone willing to try to give a bit of a layman's terms explanation ?

Nick
--

Weather Forecasters that cannot forecast, plumbers they call leaky,
builders they call Bob, Motor Technicians that are grease monkey's,
Bankers that pay bonuses for failure, the list is endless - have you
forgotten that this Country has been in decline for nearly 200 years ?


I'm in my 50s and weather forecasts have behaved as you describe as long
as I can remember.

There is actually a solid scientific reason for it, and it was largely
discovered in the meteorological context by a chap called Lorenz
somewhen in the middle of the 20th century. Other mathematicians have
also examined this in other spheres and you may have heard the terms
"chaos theory" and the "Butterfly effect".

To forecast the weather, computers need to calculate equations
describing the behaviour of the atmosphere. These are very complicated
equations and in many cases cannot be calculated exactly. A very close,
but not exact, approximation has to be made to run the calculations in a
reasonable time.

But the real choker is that many systems - and the weather is just one
of them - are very sensitive to the starting conditions. Run those
equations forward and any error in specifying the starting conditions is
magnified until the forecast becomes worthless. This is one aspect of
what is known as "chaos" in a mathematical sense.

You will appreciate that our observational coverage of the atmosphere is
actually very poor. There are some observations at ground level, but
these are if anything fewer than in the past due to cut backs. There is
almost nothing in the upper atmosphere - when I was at university in the
1970s there were less than a dozen regular radiosonde stations covering
the whole of the UK. The situation is better now as there are ways of
measuring temperature profiles from sattelite data - but these only have
a certain accuracy and require calibration from "ground truth".

In fact, there is nowhere near enough observational data to run a
computer forecast model. What is actually done is that the output of
previous forecasts is adjusted to match what observations there are and
also other data which comes from sattelite observations. The models are
then run into the future and we hope for the best.

This is the reason why you will often see terms such as "ensembles"
mentioned in this group. To try and take account of "chaos", the major
forecasting models are now run several times for each set of data. But
each run's starting conditions are altered slightly to mimic the sort of
errors present in the combined observation / old forecast data. If the
outcome of most of the ensembles is similar, a confident forecast can be
issued. If they vary wildly, the forecaster has to use a suitably vague
form of words. With practice, you can often tell from how the forecast
is worded how accurate it is expected to be - I once saw a forecast in a
national paper which said "Mainly dry - some rain in places" which is
about as non-commital as it gets.

The usual state of play with those ensembles is that they match for a
few days, then go haywire. Once this happens, you might as well read
the tea leaves or cast the runes as put your faith in those forecasts.
But, in spite of this cut-off normally being well within the limit some
models - especially the GFS model - claim to forecast out to, there are
still many debates on here with people hoping their favourite type of
weather will arrive in a week or two.

The way I use to judge the worth of a forecast - and which does not need
the ability to read the ensemble charts although I can do that at need -
is to compare the forecasts day by day for a given period ahead. If
they are consistent, then they are probably good. If they vary day by
day, or even every six hours or so, you might as well get out the seaweed.

Because the atmosphere is so sensitive to these variations and is also
so vast that we can never accurately measure more than a small sample of
it, forecasting will never be an exact science unless some method is
discovered which is well beyond our current understanding. There will
always be a risk of incidents such as the one I once saw in a newspaper
cartoon where a man was calling the Met Office to say "I think you might
like to know that I have just finished shovelling three inches of
'partly cloudy' off my drive.".

--
- Yokel -

Yokel posts via a spam-trap account which is not read.

On Apr 27, 8:24*pm, Yokel wrote:
On 27/04/2011 00:56, It's True wrote:





On Tue, 26 Apr 2011 14:36:26 -0700 (PDT), NickTheBatMan
*wrote:

I'm trying to understand why the Met Office forecasts are so
consistently wrong, and keeps changing from hour to hour sometimes !?

Last weekend - Easter - there were constantly changing forecasts for
Leeming from rain to not and back again... in the end there was no
rain to speak of - though there were storms over the Moors...

This seems to have been happening for nearly a year now, and it's very
irritating
It's not just MetO but most of the other sources too... Metcheck/BBC/
Wunderground etc...

I know it's not an exact science, but it use to be a lot more accurate
and consistent in its accuracy...

Anyone willing to try to give a bit of a layman's terms explanation ?

Nick
--

Weather Forecasters that cannot forecast, plumbers they call leaky,
builders they call Bob, Motor Technicians that are grease monkey's,
Bankers that pay bonuses for failure, *the list is endless - have you
forgotten that this Country has been in decline for nearly 200 years ?

I'm in my 50s and weather forecasts have behaved as you describe as long
as I can remember.

There is actually a solid scientific reason for it, and it was largely
discovered in the meteorological context by a chap called Lorenz
somewhen in the middle of the 20th century. *Other mathematicians have
also examined this in other spheres and you may have heard the terms
"chaos theory" and the "Butterfly effect".

To forecast the weather, computers need to calculate equations
describing the behaviour of the atmosphere. *These are very complicated
equations and in many cases cannot be calculated exactly. *A very close,
but not exact, approximation has to be made to run the calculations in a
reasonable time.

But the real choker is that many systems - and the weather is just one
of them - are very sensitive to the starting conditions. *Run those
equations forward and any error in specifying the starting conditions is
magnified until the forecast becomes worthless. *This is one aspect *of
what is known as "chaos" in a mathematical sense.

You will appreciate that our observational coverage of the atmosphere is
actually very poor. *There are some observations at ground level, but
these are if anything fewer than in the past due to cut backs. *There is
almost nothing in the upper atmosphere - when I was at university in the
1970s there were less than a dozen regular radiosonde stations covering
the whole of the UK. *The situation is better now as there are ways of
measuring temperature profiles from sattelite data - but these only have
a certain accuracy and require calibration from "ground truth".

In fact, there is nowhere near enough observational data to run a
computer forecast model. *What is actually done is that the output of
previous forecasts is adjusted to match what observations there are and
also other data which comes from sattelite observations. *The models are
then run into the future and we hope for the best.

This is the reason why you will often see terms such as "ensembles"
mentioned in this group. *To try and take account of "chaos", the major
forecasting models are now run several times for each set of data. *But
each run's starting conditions are altered slightly to mimic the sort of
errors present in the combined observation / old forecast data. *If the
outcome of most of the ensembles is similar, a confident forecast can be
issued. *If they vary wildly, the forecaster has to use a suitably vague
form of words. *With practice, you can often tell from how the forecast
is worded how accurate it is expected to be - I once saw a forecast in a
national paper which said "Mainly dry - some rain in places" which is
about as non-commital as it gets.

The usual state of play with those ensembles is that they match for a
few days, then go haywire. *Once this happens, you might as well read
the tea leaves or cast the runes as put your faith in those forecasts. *
But, in spite of this cut-off normally being well within the limit some
models - especially the GFS model - claim to forecast out to, there are
still many debates on here with people hoping their favourite type of
weather will arrive in a week or two.

The way I use to judge the worth of a forecast - and which does not need
the ability to read the ensemble charts although I can do that at need -
is to compare the forecasts day by day for a given period ahead. *If
they are consistent, then they are probably good. *If they vary day by
day, or even every six hours or so, you might as well get out the seaweed..

Because the atmosphere is so sensitive to these variations and is also
so vast that we can never accurately measure more than a small sample of
it, forecasting will never be an exact science unless some method is
discovered which is well beyond our current understanding. *There will
always be a risk of incidents such as the one I once saw in a newspaper
cartoon where a man was calling the Met Office to say "I think you might
like to know that I have just finished shovelling three inches of
'partly cloudy' off my drive.".

--
- Yokel -

Yokel posts via a spam-trap account which is not read.- Hide quoted text -

- Show quoted text -

I can't help but agree.

"Will Hand" wrote in message
...

The 1.5Km model has been re-run for that case and results are amazing,
even at 6 hours!
It captured the convergence line and re-generating cells and very high
localised rainfall over Exmoor near Boscastle.

I'm sure there was detailed example online for the Boscastle high resolution
modeling but I couldn't find it. However, I did manage to find this
http://www.metbrief.com/Images/aug2004.jpg which demonstrates it quite well.

Jon.

"Jon O'Rourke" wrote in message
...
"Will Hand" wrote in message
...

The 1.5Km model has been re-run for that case and results are amazing,
even at 6 hours!
It captured the convergence line and re-generating cells and very high
localised rainfall over Exmoor near Boscastle.

I'm sure there was detailed example online for the Boscastle high
resolution modeling but I couldn't find it. However, I did manage to find
this
http://www.metbrief.com/Images/aug2004.jpg which demonstrates it quite
well.

Jon.

Cheers Jon.
There was a hundred+ page internal document written about Boscastle which
showed the effect of increasing resolution from global right down to 1km
nicely. IIRC that diagram was in the report. It is my view that we are now
much better placed strategically to predict the probability of these genuine
extreme rainfall events occurring. Moreover things will continue to improve
as our models get even better.

Cheers,

Will
--

On Apr 28, 1:52*pm, "Jon O'Rourke" wrote:
"Will Hand" wrote in message

...

The 1.5Km model has been re-run for that case and results are amazing,
even at 6 hours!
It captured the convergence line and re-generating cells and very high
localised rainfall over Exmoor near Boscastle.

I'm sure there was detailed example online for the Boscastle high resolution
modeling but I couldn't find it. However, I did manage to find thishttp://www.metbrief.com/Images/aug2004.jpgwhich demonstrates it quite well.

Jon.

........and how much has forecasting at 10 days improved over the last
30 years?

8/10 cats say your accurate as ****

On 28/04/2011 9:00 PM, Dawlish wrote:

.......and how much has forecasting at 10 days improved over the last
30 years?

On Thu, 28 Apr 2011 13:00:53 -0700, Dawlish wrote:


.......and how much has forecasting at 10 days improved over the last 30
years?

According to February's Weather, ECMWF forecasts for 500hPa heights
appear to be as accurate at 10 days as they were at 7 days thirty years
ago.



--
Graham Davis, Bracknell, Berks. E-mail: change boy to man
To consider the Earth the only populated world in infinite space is as
absurd as to assert that in an entire field sown with millet only one
grain will grow. - Metrodoros, 300BC

On Apr 29, 8:35*am, Graham P Davis wrote:
On Thu, 28 Apr 2011 13:00:53 -0700, Dawlish wrote:

.......and how much has forecasting at 10 days improved over the last 30
years?

According to February's Weather, ECMWF forecasts for 500hPa heights
appear to be as accurate at 10 days as they were at 7 days thirty years
ago.

--
Graham Davis, Bracknell, Berks. E-mail: change boy to man
To consider the Earth the only populated world in infinite space is as
absurd as to assert that in an entire field sown with millet only one
grain will grow. - Metrodoros, 300BC

Thanks Graham. Maybe a reason to re-subscribe, as I'd like to see the
article and any stats. you can get the first page he

http://onlinelibrary.wiley.com/doi/1...a.736/abstract

But further than that its subscription time.

In reality, forecasting a week ahead was not accurate 30 years ago and
that's about where we still are 30 years later at 10 days. As Philip
said in his Telegraph blog only last week; forecasting more than even
7 days ahead is often not to be relied on. 30 years ago, it was even
less accurate.