http://news.bbc.co.uk/2/hi/science/nature/8462890.stm

Colin Youngs
Brussels

On Jan 16, 2:19*pm, "Colin Youngs" wrote:
http://news.bbc.co.uk/2/hi/science/nature/8462890.stm

Colin Youngs
Brussels

Just read it Colin; you beat me to posting it. *))

This is a very balanced piece. I agree with most of it, though at
times, it mixes seasonal and annual forecasting, which are two
actually two very different things. In terms of seasonal forecasting,
I believe they should either abandon it, or explain the experimental
nature of the forecasts and the likelihood that they may not prove
correct, far better than they presently do.

Even with the MetO's difficulties in forecasting seasonal weather
correctly, they are as good as, or better than eveyone else, except
using hindcasting odds. If anyone feels they are not, then present us
with the longer-term forecast accuracy of the person, or organisation,
you think is better. That's all you have to do. At least the MetO are
prepared to discuss their track record and don't hide it, whilst
basing their "expertise" on a few remembered successful forecasts and
forgetting the rest. They all count. They really do.

The real difficulty is that, IMO, *no-one* can forecast seasonal
weather with confidence which is backed by outcome forecast accuracy
stats. If there is; show us, don't just bleat about the MetO not being
able to do it when no-one else can. It's an area at the edge of
possibility. It's not something from which MetO accuracy should be
expected.

On 16 Jan, 15:29, Dawlish wrote:
On Jan 16, 2:19*pm, "Colin Youngs" wrote:

http://news.bbc.co.uk/2/hi/science/nature/8462890.stm

Colin Youngs
Brussels

Just read it Colin; you beat me to posting it. *))

This is a very balanced piece. I agree with most of it, though at
times, it mixes seasonal and annual forecasting, which are two
actually two very different things. In terms of seasonal forecasting,
I believe they should either abandon it, or explain the experimental
nature of the forecasts and the likelihood that they may not prove
correct, far better than they presently do.

Even with the MetO's difficulties in forecasting seasonal weather
correctly, they are as good as, or better than eveyone else, except
using hindcasting odds. If anyone feels they are not, then present us
with the longer-term forecast accuracy of the person, or organisation,
you think is better. That's all you have to do. At least the MetO are
prepared to discuss their track record and don't hide it, whilst
basing their "expertise" on a few remembered successful forecasts and
forgetting the rest. They all count. They really do.

The real difficulty is that, IMO, *no-one* can forecast seasonal
weather with confidence which is backed by outcome forecast accuracy
stats. If there is; show us, don't just bleat about the MetO not being
able to do it when no-one else can. It's an area at the edge of
possibility. It's not something from which MetO accuracy should be
expected.

This from Biased BBC:

"...I asked, last week, how long it would be before the intrepid Roger
Harrabin came up with a defence of the Met office, after his Yorkshire-
based colleague, Paul Hudson, dared to suggest that Accuweather's Joe
*******i (among others) was more accurate with his weather-forecasting
than the Met and its £170m global warming lying machine (aka a
supercomputer). Well, it's taken him all week. And if you can
understand his back-flipping, contortionist - nay, fantastical -
reasoning, you deserve a prize. As I see it, our friend Mr Harrabin
believes that when the Met Office is wrong, they are actually right,
because they are nearly right; and that in any case, it doesn't
matter, because it's getting much hotter, and their supercomputer can
see that, whereas the day-to-day incidences of freezing etc, don't
really count because they are part of the 'frying tonight' overall
trend - and on that, of course, the Met Office is always right. As for
those who doubt any of this, well, according to Mr Harrabin, he
doesn't give a damn, because they don't count, and of course, they
can't count (unlike the Met). Something like that. Me? I'll stick with
Mr *******i. His writing style might not be the most elegant, but his
message is crystal clear and honest. The Met Office are warmist
crooks..."

Fair comment, I think.

CK

|"Dawlish" wrote in message
...
On Jan 16, 2:19 pm, "Colin Youngs" wrote:
http://news.bbc.co.uk/2/hi/science/nature/8462890.stm

Colin Youngs
Brussels
|
|Just read it Colin; you beat me to posting it. *))
|
|This is a very balanced piece. I agree with most of it, though at
|times, it mixes seasonal and annual forecasting, which are two
|actually two very different things. In terms of seasonal forecasting,
|I believe they should either abandon it, or explain the experimental
|nature of the forecasts and the likelihood that they may not prove
|correct, far better than they presently do.
|
|Even with the MetO's difficulties in forecasting seasonal weather
|correctly, they are as good as, or better than eveyone else, except
|using hindcasting odds. If anyone feels they are not, then present us
|with the longer-term forecast accuracy of the person, or organisation,
|you think is better. That's all you have to do. At least the MetO are
|prepared to discuss their track record and don't hide it, whilst
|basing their "expertise" on a few remembered successful forecasts and
|forgetting the rest. They all count. They really do.
|
|The real difficulty is that, IMO, *no-one* can forecast seasonal
|weather with confidence which is backed by outcome forecast accuracy
|stats. If there is; show us, don't just bleat about the MetO not being
|able to do it when no-one else can. It's an area at the edge of
|possibility. It's not something from which MetO accuracy should be
|expected.
|

But the Met Office issue a number of "seasonal" forecasts for each season,
so which one do they count when they verify?

To take a very topical example, their original forecast for this current
winter was 50% mild and only 20% cold (these being the upper and lower 33%
of the temperature distribution - the middle 33% being "average"). This
would have been the forecast on which the local highway authorities might
have based their salt stockpiles. The government recommended 6 days supply,
the Met Office were forecasting most likely a mild winter, so why lay in any
more?

The Winter forecast now up on the Met Office site says 45% chance cold - but
this revised forecast was issued when the current (or is it now "recent"?)
cold spell had already started. Will this be the one they use to verify, or
will it be the original? With all respect to the Met Office, anyone can get
a forecast right if they wait for the weather to start happening before
issuing the prediction which counts.

If seasonal forecasts are to be any use, then in my opinion they must be
issued far enough in advance that the information is useful for planning
purposes *and this is the version which verifies*. By all means update them
later on, but now we are just having a discussion so these cannot count for
meaningful statistics.

Until these seasonal forecasts can be shown to have significant skill, they
should be clearly labelled "Experimental" or - like some machines in the old
"penny arcades" - "For amusement only - no prizes". Of course it is
possible that the variability of our local climate and weather is such that
reliable seasonal forecasts cannot be issued with technology likely to be
available any time soon. If this is the case, let's hold our hands up and
be honest about it.

I have no problem with issuing experimental seasonal forecasts for
comparison purposes while developing the necessary technology - but if this
is the case they should be in a separate section of the website and not put
up alongside the short term forecasts which are clearly operational and for
public use.
--
- Yokel -

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

Until these seasonal forecasts can be shown to have significant
skill, they
should be clearly labelled "Experimental" or - like some machines in
the old
"penny arcades" - "For amusement only - no prizes". Of course it is
possible that the variability of our local climate and weather is
such that
reliable seasonal forecasts cannot be issued with technology likely
to be
available any time soon. If this is the case, let's hold our hands
up and
be honest about it.



.... for those of us who were around in the 1960s, all this is a bit
"Déjà vu" ! Good grief, that was half-a-century ago :-)

Martin.


--
Martin Rowley
West Moors, East Dorset (UK): 17m (56ft) amsl
Lat: 50.82N Long: 01.88W
NGR: SU 082 023

"Martin Rowley" wrote in message
news |
| Until these seasonal forecasts can be shown to have significant
| skill, they
| should be clearly labelled "Experimental" or - like some machines in
| the old
| "penny arcades" - "For amusement only - no prizes". Of course it is
| possible that the variability of our local climate and weather is
| such that
| reliable seasonal forecasts cannot be issued with technology likely
| to be
| available any time soon. If this is the case, let's hold our hands
| up and
| be honest about it.
|
|
|
| ... for those of us who were around in the 1960s, all this is a bit
| "Déjà vu" ! Good grief, that was half-a-century ago :-)
|

Those forecasts were produced by the "analogue" method. Basically, they
looked for similar months (analogues) to the one which has just gone and
worked on the basis that similar weather would follow. So if you had 4
Decembers like the one just gone and the 4 Januarys that followed were
similar, you would use the weather in those Januarys as the basis for your
January forecast. If the months following the analogues had different types
of weather, you issued a forecast on the lines of "dry spells, some rain in
places" and hoped no-one noticed.

As you doubtless recall, this method was not a winner then. If climate
change is real, it is even less likely to be a winner now as going back to
the past will not be comparing like-with-like any more.
--
- Yokel -

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

On Jan 16, 5:26*pm, Natsman wrote:
On 16 Jan, 15:29, Dawlish wrote:





On Jan 16, 2:19*pm, "Colin Youngs" wrote:

http://news.bbc.co.uk/2/hi/science/nature/8462890.stm

Colin Youngs
Brussels

Just read it Colin; you beat me to posting it. *))

This is a very balanced piece. I agree with most of it, though at
times, it mixes seasonal and annual forecasting, which are two
actually two very different things. In terms of seasonal forecasting,
I believe they should either abandon it, or explain the experimental
nature of the forecasts and the likelihood that they may not prove
correct, far better than they presently do.

Even with the MetO's difficulties in forecasting seasonal weather
correctly, they are as good as, or better than eveyone else, except
using hindcasting odds. If anyone feels they are not, then present us
with the longer-term forecast accuracy of the person, or organisation,
you think is better. That's all you have to do. At least the MetO are
prepared to discuss their track record and don't hide it, whilst
basing their "expertise" on a few remembered successful forecasts and
forgetting the rest. They all count. They really do.

The real difficulty is that, IMO, *no-one* can forecast seasonal
weather with confidence which is backed by outcome forecast accuracy
stats. If there is; show us, don't just bleat about the MetO not being
able to do it when no-one else can. It's an area at the edge of
possibility. It's not something from which MetO accuracy should be
expected.

This from Biased BBC:

"...I asked, last week, how long it would be before the intrepid Roger
Harrabin came up with a defence of the Met office, after his Yorkshire-
based colleague, Paul Hudson, dared to suggest that Accuweather's Joe
*******i (among others) was more accurate with his weather-forecasting
than the Met and its £170m global warming lying machine (aka a
supercomputer). Well, it's taken him all week. And if you can
understand his back-flipping, contortionist - nay, fantastical -
reasoning, you deserve a prize. As I see it, our friend Mr Harrabin
believes that when the Met Office is wrong, they are actually right,
because they are nearly right; and that in any case, it doesn't
matter, because it's getting much hotter, and their supercomputer can
see that, whereas the day-to-day incidences of freezing etc, don't
really count because they are part of the 'frying tonight' overall
trend - and on that, of course, the Met Office is always right. As for
those who doubt any of this, well, according to Mr Harrabin, he
doesn't give a damn, because they don't count, and of course, they
can't count (unlike the Met). Something like that. Me? I'll stick with
Mr *******i. His writing style might not be the most elegant, but his
message is crystal clear and honest. The Met Office are warmist
crooks..."

Fair comment, I think.

CK- Hide quoted text -

- Show quoted text -

Foercasting prowess *has* to be backed by outcomes. Joe Bastadi got
all 4 of his USA winter forecasts wrong between 2004/5 and 2008/9. The
MetO didn't get all 4 of our winter forecasts wrong during this time.
So who issued the better forecasts? Is a forecast good because it's
message is "crystal clear and honest", or is a forecast good because
it ends up correct?

I don't think the MetO's seasonal forecasts are anything like good
enough to trust, but the evidence is that nobody else's is good enough
to trust.

If Joe *******i's (or anyone's) forecasts, are better, all you have to
do is to present longer-term evidence that they are and not rely on
his most recent one - which is only half way towards outcome and
already contains significant errors.

On Saturday 16 Jan 2010 23:15, Yokel scribbled:


"Martin Rowley" wrote in message
news |
| Until these seasonal forecasts can be shown to have significant
| skill, they
| should be clearly labelled "Experimental" or - like some machines in
| the old
| "penny arcades" - "For amusement only - no prizes". Of course it is
| possible that the variability of our local climate and weather is
| such that
| reliable seasonal forecasts cannot be issued with technology likely
| to be
| available any time soon. If this is the case, let's hold our hands
| up and
| be honest about it.
|
|
|
| ... for those of us who were around in the 1960s, all this is a bit
| "Déjà vu" ! Good grief, that was half-a-century ago :-)
|

Those forecasts were produced by the "analogue" method. Basically, they
looked for similar months (analogues) to the one which has just gone and
worked on the basis that similar weather would follow. So if you had 4
Decembers like the one just gone and the 4 Januarys that followed were
similar, you would use the weather in those Januarys as the basis for your
January forecast. If the months following the analogues had different
types of weather, you issued a forecast on the lines of "dry spells, some
rain in places" and hoped no-one noticed.

There was a lot more to it than that. That was the simple version that was
either supplied to the media or the only bit they could understand from a
more detailed briefing.

Before the monthly forecasts were issued, there were several meetings, each
dealing independently with an aspect of the atmospheric and surface
conditions. These were then brought together with a final meeting. I used to
take part in the one which dealt with surface conditions, ice, snow, SST
anomalies, etc.


As you doubtless recall, this method was not a winner then. If climate
change is real, it is even less likely to be a winner now as going back to
the past will not be comparing like-with-like any more.

Agreed. Even what I thought was the most reliable contribution to the
forecast, the SST anomalies, seems less reliable now, perhaps due to GW,
though it's worked really well this season.

--
Graham P Davis, Bracknell, Berks., UK. E-mail: newsman not newsboy
"I wear the cheese. It does not wear me."