"John Hall" wrote in message
...
In article ,
Lawrence Jenkins writes:

"Martin Rowley" wrote in message
...
http://www.metoffice.gov.uk/corporat...20100305b.html


Thats good to see they will now concentrate on AGW disaster weather
scenarios a hundred years hence.



Forecasting weather and forecasting climate are two very different
things. It may be well be easier to forecast the climate for the next
few decades than it is to forecast the weather for the next three
months.
--
John Hall
"Acting is merely the art of keeping a large group of people
from coughing."
Sir Ralph Richardson (1902-83)



Yes true John and they can't seem to do neither. It's amazing what you can
get away with when it ain't your money

"John Hall" wrote in message

Very different, because it was for a /particular/ three month period.
They said something along the lines of: "we expect this coming winter to
have a 70% chance of being warmer than the 1971-2000 average". That
makes it a weather forecast. If they had said "we expect 70% of winters
in the next decade to be warmer than the 1971-2000 average" then in
would have been a climate forecast.


Yes, but forecasting climate in the long-term can be easier than
forecasting weather in the shorter term. Take those two example
forecasts above. With the first, if the winter wasn't warmer than
average most people would view it as having been wrong. It certainly
wouldn't have provided them with useful guidance. But the second
forecast would have provided useful guidance by the end of the decade if
most winters were warmer than average, even if the number was not
precisely 7 out of 10 but 6 or 8 (or arguably even if it was 9 or 10).


That's because, averaged over the whole globe, there seems very little
possibility of that happening unless the understanding embodied in the
models is fatally flawed. I suppose that a sudden unpredicted - and
possibly unpredictable - increase in volcanic activity or a reduction in
solar activity could lead to a reduction, though, at least temporarily
--
John Hall

Thanks for some good answers there John. The difference between a seasonal
forecast and a climate forecast is now somewhat clearer in my mind.

A few points of concern though; When seasonal forecasts are made (whether or
not by the met office), I presume they are based on recent 30 year averages
or perhaps longer and recent trends within that - or they should be anyway.
And are usually made only one season in advance of course. IE I haven't seen
a forecast yet for the winter or summer of 2014....

Is it wrong to therefor say that decadal forecasts should be using 30
decades worth of data (which obviously they can't) as well as recent trends
to forecast with similar accuracy the next decade? And that using less
decadal data hinders the accuracy or value of that forecast? And likewise,
is it sensible to make a forecast for 4 decades from now with such a limited
database and a plethora of potential variables, most of which are not new
and have influenced the climate greatly in the past? IE is a current decadal
climate forcast for the 2050's not akin to making a seasonal forecast now
for 2014, but with even less data?

Let's assume the met office issued the exact same winter forecast for the
next nine winters, and alongside that, global climate predictions say that 7
out of 10 winters should be warmer than average in north west europe. And in
the event nine winters were warmer than average, Haven't they both got it
wrong? Conversely of course, I could have said only half of the winter's
were warmer than average. Who is less wrong on that occasion, and are they
more wrong than if there had been 9?

I think my wider point is that some people, very mistakenly, take for
granted that climate forecasts for 2050 are as accurate as 2 day weather
forecasts and hugely more reliable than a seasonal forecast. And perhaps
worse than this, some scientists advocate that line of thinking and won't
hear a word against it. I would like to understand why this has became such
a common assumption? Some variables such as volcanic or solar activity,
cannot of course be forecast. But the one variable that is a consequence of
continued global warming is an increase in water vapour in the atmosphere,
which strangely people forget about, or even ignore, and will likely have a
massive effect compared to current climate change not only upon the climate,
but equally on humanity. Indeed, I wonder to what extent this variable has
impacted on the incorrect met office seasonal forecast?

As you can tell, I'm far from convinced that forecasting the climate in
decadal terms is any easier than a seasonal weather forecast, but one thing
is for sure, it takes a hell of a lot longer to find out if it's achievable.

And if anyone else would like to chip in, please try to do so in the same
intellectual fashion as John has :P

Alex.

In article ,
Alex Stephens Jr writes:
Thanks for some good answers there John.

Thanks.

The difference between a seasonal
forecast and a climate forecast is now somewhat clearer in my mind.

A few points of concern though; When seasonal forecasts are made (whether or
not by the met office), I presume they are based on recent 30 year averages
or perhaps longer and recent trends within that - or they should be anyway.

I think the Met Office seasonal forecasts specifically said that they
were relative to the 1971-2000 average.

And are usually made only one season in advance of course. IE I haven't seen
a forecast yet for the winter or summer of 2014....

Yep. A /weather/ forecast - as distinct from a /climate/ forecast - that
far in advance could be no better than guesswork, I think. (And a
climate forecast for 2014 would have to say no more than that the
climate would be much as it is now, as four years is too short a period
for it to change much. Which doesn't rule out the possibility of 2014
having either unusually warm or unusually cold weather in the UK.)

Is it wrong to therefor say that decadal forecasts should be using 30
decades worth of data (which obviously they can't) as well as recent trends
to forecast with similar accuracy the next decade?

Did you really mean 30 decades? I don't think that as long a record as
that is necessary. What you need to put in the models as a starting
point is as accurate a representation of the current climate as possible
- for which thirty years of data should be enough. There's a trade off:
use too short a reference period, and the short term "noise" of weather
can distort the picture, but too long a period and the underlying
climate may have changed appreciably during that time. I suspect that's
why conditions for 1971-2000 are usually used as a base-line. Even then,
some warming is evident during that period.

Having said that, you would obviously want to check, and maybe
calibrate, your model by running it for the past, to compare its results
with what actually happened. Clearly the longer period that you can do
that for, the better. Since climate is a global thing, and you can't
really look at one part of the world in isolation, you ideally need
global data, which I suspect would limit you to the last 100-150 years.

And that using less
decadal data hinders the accuracy or value of that forecast? And likewise,
is it sensible to make a forecast for 4 decades from now with such a limited
database and a plethora of potential variables, most of which are not new
and have influenced the climate greatly in the past? IE is a current decadal
climate forcast for the 2050's not akin to making a seasonal forecast now
for 2014, but with even less data?

Let's assume the met office issued the exact same winter forecast for the
next nine winters, and alongside that, global climate predictions say that 7
out of 10 winters should be warmer than average in north west europe. And in
the event nine winters were warmer than average, Haven't they both got it
wrong?

I think you get into areas where it's hard to know what is meant by
"right" and "wrong" unless you define your criteria precisely in
advance. And even then a scientist might view a forecast as having been
right where a typical member of the public might view it as wrong.
Personally I would say that a forecast of 7 out of 10 winters warmer
than average would be pretty successful if 9 out of 10 were warmer,
since the /weather/ can't be forecast that far ahead so there's always
the chance of essentially random factors causing rather more - or rather
fewer - warm winters than expected.

Conversely of course, I could have said only half of the winter's
were warmer than average. Who is less wrong on that occasion, and are they
more wrong than if there had been 9?

About the same, I would say.

I think my wider point is that some people, very mistakenly, take for
granted that climate forecasts for 2050 are as accurate as 2 day weather
forecasts and hugely more reliable than a seasonal forecast.

If you look at 2050 in isolation, then you are effectively turning the
climate forecast into a weather forecast, and it is likely to be worse -
less accurate and much less detailed - than an attempt to forecast the
next season in 2010. But if you regard it as a climate forecast for the
period 2045-2055, then the picture should be different. It won't tell
you about individual years, but it will tell you about the
characteristics of the period. Forty years is a far way away, but it
should be possible to say something like 95% confidence that mean
temperature for 2045-55 will be between 1 and 2 degrees (say) higher
than it is now. And for that far ahead, that's really all governments
and big corporations need for long-term planning. It's not like seasonal
forecasts, where you might want to know if you should take your holiday
at home or abroad, or as a council whether you should stockpile more
salt.

And perhaps
worse than this, some scientists advocate that line of thinking and won't
hear a word against it. I would like to understand why this has became such
a common assumption? Some variables such as volcanic or solar activity,
cannot of course be forecast. But the one variable that is a consequence of
continued global warming is an increase in water vapour in the atmosphere,
which strangely people forget about, or even ignore, and will likely have a
massive effect compared to current climate change not only upon the climate,
but equally on humanity.

It's certainly taken account of in climate modelling.

Indeed, I wonder to what extent this variable has
impacted on the incorrect met office seasonal forecast?

Not at all, I would have thought, since the amount of water vapour
currently in the atmosphere is known and the effect of GW isn't going to
noticeably affect it over a period as short as three months.

As you can tell, I'm far from convinced that forecasting the climate in
decadal terms is any easier than a seasonal weather forecast, but one thing
is for sure, it takes a hell of a lot longer to find out if it's achievable.

And if anyone else would like to chip in, please try to do so in the same
intellectual fashion as John has :P
--
John Hall
"Acting is merely the art of keeping a large group of people
from coughing."
Sir Ralph Richardson (1902-83)

"John Hall" wrote in message
...


If you look at 2050 in isolation, then you are effectively turning the
climate forecast into a weather forecast, and it is likely to be worse -
less accurate and much less detailed - than an attempt to forecast the
next season in 2010. But if you regard it as a climate forecast for the
period 2045-2055, then the picture should be different. It won't tell
you about individual years, but it will tell you about the
characteristics of the period. Forty years is a far way away, but it
should be possible to say something like 95% confidence that mean
temperature for 2045-55 will be between 1 and 2 degrees (say) higher
than it is now.

Thanks again for an excellent reply John.
95% certainty on a figure ±0.5ºC at a range of 40 years ± years.
And presumably precipitation estimations must be equally as accurate.
I have to say that is utterly phenomenal. I had no idea the climate was so
easily modelled.
It is indeed easier than a 2 day weather forecast apparently.

Wow.... Nope, I can't get my head around it. It doesn't fit. Not the actual
figure, but the confidence percentile, it's way too high with too little a
margin for error on such a timescale. And you don't need a PhD in statistics
to work that one out. :-)

Eeee, who's guilty of this crime? Name and shame.

Alex.

In article ,
Alex Stephens Jr writes:

"John Hall" wrote in message
.. .


If you look at 2050 in isolation, then you are effectively turning the
climate forecast into a weather forecast, and it is likely to be worse -
less accurate and much less detailed - than an attempt to forecast the
next season in 2010. But if you regard it as a climate forecast for the
period 2045-2055, then the picture should be different. It won't tell
you about individual years, but it will tell you about the
characteristics of the period. Forty years is a far way away, but it
should be possible to say something like 95% confidence that mean
temperature for 2045-55 will be between 1 and 2 degrees (say) higher
than it is now.

Thanks again for an excellent reply John.
95% certainty on a figure ±0.5ºC at a range of 40 years ± years.
And presumably precipitation estimations must be equally as accurate.
I have to say that is utterly phenomenal. I had no idea the climate was so
easily modelled.

I hope that in saying that you're going by something more than my "Forty
years is a far way away, but it should be possible to say something like
95% confidence that mean temperature for 2045-55 will be between 1 and 2
degrees (say) higher than it is now." The "say" was supposed to indicate
that the "between 1 and 2 degrees" was being used purely to illustrate
the point I was making, rather than to suggest that is what the models
are actually saying (though I would think that it would be in the right
ball park). I have a tendency to pontificate, but I'm certainly not an
expert in the field of either GW or computer modelling, and I wouldn't
want to have mistakenly given you that impression.

It is indeed easier than a 2 day weather forecast apparently.

Wow.... Nope, I can't get my head around it. It doesn't fit. Not the actual
figure, but the confidence percentile, it's way too high with too little a
margin for error on such a timescale. And you don't need a PhD in statistics
to work that one out. :-)

Eeee, who's guilty of this crime? Name and shame.


See above. Nobody may be saying exactly that, as my "it should be
possible to say something like" was meant to indicate. But averaging
over a 10 year period such as 2045-55, it seems to me quite plausible to
forecast the mean temperature for the period pretty accurately with a
high degree of confidence.
--
John Hall
"Acting is merely the art of keeping a large group of people
from coughing."
Sir Ralph Richardson (1902-83)

"John Hall" wrote in message
...
In article ,
Alex Stephens Jr writes:

"John Hall" wrote in message
. ..


If you look at 2050 in isolation, then you are effectively turning the
climate forecast into a weather forecast, and it is likely to be worse -
less accurate and much less detailed - than an attempt to forecast the
next season in 2010. But if you regard it as a climate forecast for the
period 2045-2055, then the picture should be different. It won't tell
you about individual years, but it will tell you about the
characteristics of the period. Forty years is a far way away, but it
should be possible to say something like 95% confidence that mean
temperature for 2045-55 will be between 1 and 2 degrees (say) higher
than it is now.

Thanks again for an excellent reply John.
95% certainty on a figure ±0.5ºC at a range of 40 years ± years.
And presumably precipitation estimations must be equally as accurate.
I have to say that is utterly phenomenal. I had no idea the climate was so
easily modelled.

I hope that in saying that you're going by something more than my "Forty
years is a far way away, but it should be possible to say something like
95% confidence that mean temperature for 2045-55 will be between 1 and 2
degrees (say) higher than it is now." The "say" was supposed to indicate
that the "between 1 and 2 degrees" was being used purely to illustrate
the point I was making, rather than to suggest that is what the models
are actually saying (though I would think that it would be in the right
ball park). I have a tendency to pontificate, but I'm certainly not an
expert in the field of either GW or computer modelling, and I wouldn't
want to have mistakenly given you that impression.

It is indeed easier than a 2 day weather forecast apparently.

Wow.... Nope, I can't get my head around it. It doesn't fit. Not the
actual
figure, but the confidence percentile, it's way too high with too little a
margin for error on such a timescale. And you don't need a PhD in
statistics
to work that one out. :-)

Eeee, who's guilty of this crime? Name and shame.


See above. Nobody may be saying exactly that, as my "it should be
possible to say something like" was meant to indicate. But averaging
over a 10 year period such as 2045-55, it seems to me quite plausible to
forecast the mean temperature for the period pretty accurately with a
high degree of confidence.
--
John Hall
"Acting is merely the art of keeping a large group of people
from coughing."
Sir Ralph Richardson (1902-83)

Sorry John, I was neither accusing you or anyone else. But if someone had
drawn a conclusion of such certainty over a 35-45 year period it would be
very simple to discredit it.
Even projecting the maximum range of the previous 40 years along with trend
(roughly 0.9ºC higher than the mean of this decade, or 1.6ºC higher than the
current available thirty year mean ±1.9ºC ) would lead us to a figure far
inferior than a 95% confidence value (but far greater than the narrow one
above) - and yet still allows for the possibility of a stall in global
warming from current values.
IE it cannot be ruled out with anything like 95% certainty
Alex.