Amongst all the discussion of the MO's winter forecast, one very
important point has been missed. Unfortunately, the original
forecast seems no longer to be available, but the updated version,
issued on 22 Dec, says:

QUOTE
The forecast is based on our prediction that the North Atlantic
Oscillation (NAO) will be negative this winter.
UNQUOTE

Note the use of the word "will", and the absence of any
percentage probability. Unlike other parts of the forecast which
are given probabilities, this statement will be able to be judged
come the end of February.

No judgment should, of course, be made at this stage, but
these figures indicate what sort of February may be required for
the statement to verify:

Month.....Iceland/Azores....1961-90 norm......NAO-index
Dec 2005 ...... 18.1 .................17.0 ..................+1.1
Jan 2006 ....... 28.7 ................. 17.4 .................+11.3
required
Feb 2006 ........ 2.2 ................. 14.7 ................. -12.5

I have used mean sea-level pressure at grid-points 65N25W
and 40N 30W to represent Iceland and the Azores, because
these are easily available to me, and I have used 1961-90 as
it is believed that this is probably the standard period used in
developing the method, although I cannot find in the documents
covering methodology (or the forecast itself) on the MO
website any specification of the period. The only quoted
period I have found is 1950-1998 for the hindcast study itself.
In that hindcast study it is said that the NAO index used there
is based on data at the 500mbar level, but it also insists that there is
little difference between that index and the surface NAO index.

Out of interest, the coldest Februarys of the last 130 years
had the following NAO indexes[sic]:
1986 -17.7
1963 -11.2
1956 -12.7
1947 -32.7
1895 -32.2
.....so it is possible ....

Philip Eden

Thank you Philip.
So have they made a forecast of what it will be for February? Since it was
possible to make a forecast for winter, early on, one would presume they can
make one for Febraury - hopefully with greater accuracy.

Dave
"Philip Eden" philipATweatherHYPHENukDOTcom wrote in message
...
Amongst all the discussion of the MO's winter forecast, one very
important point has been missed. Unfortunately, the original
forecast seems no longer to be available, but the updated version,
issued on 22 Dec, says:

QUOTE
The forecast is based on our prediction that the North Atlantic
Oscillation (NAO) will be negative this winter.
UNQUOTE

Note the use of the word "will", and the absence of any
percentage probability. Unlike other parts of the forecast which
are given probabilities, this statement will be able to be judged
come the end of February.

No judgment should, of course, be made at this stage, but
these figures indicate what sort of February may be required for
the statement to verify:

Month.....Iceland/Azores....1961-90 norm......NAO-index
Dec 2005 ...... 18.1 .................17.0 ..................+1.1
Jan 2006 ....... 28.7 ................. 17.4 .................+11.3
required
Feb 2006 ........ 2.2 ................. 14.7 ................. -12.5

I have used mean sea-level pressure at grid-points 65N25W
and 40N 30W to represent Iceland and the Azores, because
these are easily available to me, and I have used 1961-90 as
it is believed that this is probably the standard period used in
developing the method, although I cannot find in the documents
covering methodology (or the forecast itself) on the MO
website any specification of the period. The only quoted
period I have found is 1950-1998 for the hindcast study itself.
In that hindcast study it is said that the NAO index used there
is based on data at the 500mbar level, but it also insists that there is
little difference between that index and the surface NAO index.

Out of interest, the coldest Februarys of the last 130 years
had the following NAO indexes[sic]:
1986 -17.7
1963 -11.2
1956 -12.7
1947 -32.7
1895 -32.2
....so it is possible ....

Philip Eden

"Philip Eden" philipATweatherHYPHENukDOTcom wrote in news:43e0c344$0
:

I have used mean sea-level pressure at grid-points 65N25W
and 40N 30W to represent Iceland and the Azores

Philip

I did something similar comparing the relative strengths of the high in the
area around the Azores and in Scandinavia since 1987, I'll post later in
the week on this thread when I'm less busy !

Richard

Philip, this confuses and surprises me somewhat:

Felly sgrifennodd Philip Eden philipATweatherHYPHENukDOTcom:
Note the use of the word "will", and the absence of any
percentage probability.

To be fair, there was a probability if you looked up the forecasting
methods, where they said that the May SST anomalies (?I think) predict
with 2/3 certainty the NAO. That's from memory, and I am open to
correction!

Month.....Iceland/Azores....1961-90 norm......NAO-index
Dec 2005 ...... 18.1 .................17.0 ..................+1.1
Jan 2006 ....... 28.7 ................. 17.4 .................+11.3
required
Feb 2006 ........ 2.2 ................. 14.7 ................. -12.5

So why is it that the weather that has dominated here and in continental
Europe recently is exactly what one would have expected from a negative NAO
(is it not?). It certainly matches closely what I made a point of reading
about when I saw the NAO forecast.

Adrian
--
Adrian Shaw ais@
Adran Cyfrifiadureg, Prifysgol Cymru, aber.
Aberystwyth, Ceredigion, Cymru ac.
http://users.aber.ac.uk/ais uk

"Adrian D. Shaw" wrote in message
...
Philip, this confuses and surprises me somewhat:

Felly sgrifennodd Philip Eden philipATweatherHYPHENukDOTcom:
Note the use of the word "will", and the absence of any
percentage probability.

To be fair, there was a probability if you looked up the forecasting
methods, where they said that the May SST anomalies (?I think) predict
with 2/3 certainty the NAO. That's from memory, and I am open to
correction!

You may well be right, but I am quoting from their web page
wich purports to discuss the reasoning behind the forecast.
I think that is fair. Norman and others have pointed out that
it is not possible to verify a forecast which depends on
probabilities until one has a sufficiently large number of them.
The statement that I quote will be verifiable on March 1.

Month.....Iceland/Azores....1961-90 norm......NAO-index
Dec 2005 ...... 18.1 .................17.0 ..................+1.1
Jan 2006 ....... 28.7 ................. 17.4 .................+11.3
required
Feb 2006 ........ 2.2 ................. 14.7 ................. -12.5

So why is it that the weather that has dominated here and in continental
Europe recently is exactly what one would have expected from a negative
NAO
(is it not?). It certainly matches closely what I made a point of reading
about when I saw the NAO forecast.

That is evident, but your expectations are flawed. This is also why the
point I am making has been missed.

Is a forecast which turns out to be correct for the wrong reason
still correct? To be fair to other posters, it has once or twice -
quite correctly - been mentioned that the correlation between
winter weather over Europe and/or the UK on the one hand, and
the NAO index on the other, is far from exact. No all -ve NAOi
winters are cold and dry. Not all cold and dry winters have a
-ve NAOi.

Philip

"Adrian D. Shaw" wrote in message
...
Philip, this confuses and surprises me somewhat:

Felly sgrifennodd Philip Eden philipATweatherHYPHENukDOTcom:
Note the use of the word "will", and the absence of any
percentage probability.

To be fair, there was a probability if you looked up the forecasting
methods, where they said that the May SST anomalies (?I think) predict
with 2/3 certainty the NAO. That's from memory, and I am open to
correction!

So why is it that the weather that has dominated here and in continental
Europe recently is exactly what one would have expected from a negative
NAO
(is it not?). It certainly matches closely what I made a point of reading
about when I saw the NAO forecast.

Let me throw another thought into the discussion. A -ve winter NAOi
does not automatically guarantee a -ve winter CET. (OK, I know that's
not quite what was offered, but it's what people are expecting). A quick
flick through the individual monthly sea-level pressure charts for the
last 30 years suggests that the relationship between NAOi and CET
may also be something like 2/3 (maybe a little more, but not much).

So, if we accept that figure, the possible outcomes a
44% colder than average with a -ve NAOi
22% colder than average with a +ve NAOi
22% warmer than average with a -ve NAOi
11% warmer than average with a +ve NAOi

So, if the forecast is right for the wrong reason, is it a success?
If it is wrong, but for the right reason, is that a success?

I can see the press release forming as I speak. Either of those will
be a "partial success" or a "qualified success". So there is an 89%
probability that the MO's winter forecast will be spun as a success
or a partial success.

Just try it, Wayne.

Philip Eden

Let me throw another thought into the discussion. A -ve winter NAOi
does not automatically guarantee a -ve winter CET. (OK, I know that's
not quite what was offered, but it's what people are expecting). A quick
flick through the individual monthly sea-level pressure charts for the
last 30 years suggests that the relationship between NAOi and CET
may also be something like 2/3 (maybe a little more, but not much).

So, if we accept that figure, the possible outcomes a
44% colder than average with a -ve NAOi
22% colder than average with a +ve NAOi
22% warmer than average with a -ve NAOi
11% warmer than average with a +ve NAOi

So, if the forecast is right for the wrong reason, is it a success?
If it is wrong, but for the right reason, is that a success?

I can see the press release forming as I speak. Either of those will
be a "partial success" or a "qualified success". So there is an 89%
probability that the MO's winter forecast will be spun as a success
or a partial success.

Just try it, Wayne.

Philip Eden


The problem I have with these % probability forecasts, is when the
probability is less than 50%. If it doesn't happen, the forecast was
correct, in that it predicted correctly that something didn't happen.

So, in Winter instead of the forecastor saying "There is a 70% chance it
will stay mild" he/she says "There is a 30% chance of cold weather
returning".

Graham

Felly sgrifennodd Philip Eden philipATweatherHYPHENukDOTcom:
Let me throw another thought into the discussion. A -ve winter NAOi
does not automatically guarantee a -ve winter CET. (OK, I know that's
not quite what was offered, but it's what people are expecting). A quick
flick through the individual monthly sea-level pressure charts for the
last 30 years suggests that the relationship between NAOi and CET
may also be something like 2/3 (maybe a little more, but not much).

So, if we accept that figure, the possible outcomes a
44% colder than average with a -ve NAOi
22% colder than average with a +ve NAOi
22% warmer than average with a -ve NAOi
11% warmer than average with a +ve NAOi

So, if the forecast is right for the wrong reason, is it a success?
If it is wrong, but for the right reason, is that a success?

Interesting, thanks for these figures!

The Met Office figures still add up though, using your 2/3 relationship.

They forecast a -ve NAO with 2/3 probability, therefore by inference also
a +ve NAO with 1/3 probability.

The forecast of a cooler than average winter for central Europe was
based on:

44% chance of colder than average with -ve NAO
22% chance of colder than average with +ve NAO

= 66% chance of colder than average winter.

Assuming that actually is how they arrived at their 2/3 chance (maybe an
assumption too far), then the forecast was right (for Central Europe, at
least).

But what is right, when we're talking percentages, rather than black and
white/yes and no? Really, it just provides more data to plug back into
the statistics for another year. But to put it another way, I doubt that
many Europeans would say that the forecast was misleading.

Adrian
--
Adrian Shaw ais@
Adran Cyfrifiadureg, Prifysgol Cymru, aber.
Aberystwyth, Ceredigion, Cymru ac.
http://users.aber.ac.uk/ais uk

"Philip Eden" philipATweatherHYPHENukDOTcom wrote:



So, if the forecast is right for the wrong reason, is it a success?
If it is wrong, but for the right reason, is that a success?


When I was forecasting, I think I would've claimed a right forecast for the
wrong reason as a success - not that it ever happened, of course. There
were enough occasions when the forecast was right but the weather was wrong
for me not to feel guilty.

--
Graham Davis
Bracknell

In message , Adrian D. Shaw
writes
Felly sgrifennodd Philip Eden philipATweatherHYPHENukDOTcom:
Let me throw another thought into the discussion. A -ve winter NAOi
does not automatically guarantee a -ve winter CET. (OK, I know that's
not quite what was offered, but it's what people are expecting). A quick
flick through the individual monthly sea-level pressure charts for the
last 30 years suggests that the relationship between NAOi and CET
may also be something like 2/3 (maybe a little more, but not much).

So, if we accept that figure, the possible outcomes a
44% colder than average with a -ve NAOi
22% colder than average with a +ve NAOi
22% warmer than average with a -ve NAOi
11% warmer than average with a +ve NAOi

So, if the forecast is right for the wrong reason, is it a success?
If it is wrong, but for the right reason, is that a success?

Interesting, thanks for these figures!

The Met Office figures still add up though, using your 2/3 relationship.

They forecast a -ve NAO with 2/3 probability, therefore by inference also
a +ve NAO with 1/3 probability.

The forecast of a cooler than average winter for central Europe was
based on:

44% chance of colder than average with -ve NAO
22% chance of colder than average with +ve NAO

= 66% chance of colder than average winter.

Assuming that actually is how they arrived at their 2/3 chance (maybe an
assumption too far), then the forecast was right (for Central Europe, at
least).

But what is right, when we're talking percentages, rather than black and
white/yes and no? Really, it just provides more data to plug back into
the statistics for another year. But to put it another way, I doubt that
many Europeans would say that the forecast was misleading.
Adrian, when I was in the Office we used to work on a zonal index which
I understood to be the mean of the surface pressure difference between
the Azores and Reykavik. I would be interested to see the upper air
anomalies this winter. Whether they be 500MB, Total thickness or 300MB.
I was used to working at 300MB while CFO (as it was then) used 250MB
both for obvious reasons. (Jet stream).
Although I have no data I think I can still suggest that the
semi-sinusoidal upper pattern over the eastern northern Atlantic has
been fairly static in January. I suppose the NAO is based on the upper
air, in January the upper ridge was maintained by a series of confluent
upper troughs relaxing to the south of Iceland thereby maintaining a
semi-permanent upper ridge and the surface high over northern UK.
As you may gather from my cautious wording - I stand to be corrected.
Cheers
Paul
--
'Wisest are they that know they do not know.' Socrates.
Paul Bartlett FRMetS
www.rutnet.co.uk Go to local weather.
400FT AMSL 25Miles Southwest of the Wash