Autumn And Winter

[parmenides3]

As I posted on the sea ice thread (sorry to duplicate), Alaska ice desk tends to view 1986 and 2001 as good pattern matches for the Summer. Whilst this is far distant and not necessarily a signal for us in the UK, I would be interested if anyone can tell me what those two Winters were like, as I have the memory of a goldfish.

Also, how close do you think those Summers were to the one we have just had?

P

[summer blizzard]

The winter of 1987 was below average and featured a blow average January or February, the winter of 2002 featured a below average December but was above average, this is further 'fuel to the fire' to support at least one below average winter month.

[Dawlish]

Hi Wilson,

I'm absolutely certain we'll have periods of snow this winter, I'm certain we'll have periods of dry, mild weather and some periods of stormy weather too; however, the prediction of this based on present knowledge of teleconnections (such an unnecessarily technical word for high pressure here, low pressure there) and especially the prediction of a particular North Atlantic pressure regime during the winter is just not good enough, yet, to have even reasonable confidence.

I've got to ask anyone involved in predicting these pressure regimes; what actual confidence do we have in the predictions being correct and upon what do you base that confidence?

There's a lot being made of teleconnections and research in the field may produce dividends, sooner, rather than later, but the prediction of these patterns, several months in advance, is so poor at present. I don't deny the science may prove to be the holy grail of LRF, but to express confidence that a particular pattern will occur has to be rooted in how many times that particular analysis has proved correct in the past. What statistical confidence have we got in the outcome being what we are saying it will be?

Why, exactly, does the forecast "look promising"? Is it, instead, that the forecast simply looks nice for snow, ignoring whether the forecast itself is of any use whatsoever, because the same technique has proved inaccurate in the past. When you actually consider what I've said there, the whole thing collapses as a real forecast and stays as a hopecast.

I'll keep asking these questions, as I know the success of using teleconnections is not high enough to trust, from any major forecasting agency, so any forecast, by an individual, on here, must be taken with an enormous pinch of salt and the individuals making the "forecasts" really must qualify their confidence with realism. Maybe "this could happen" rather than "I'm feeling increasingly confident", or "It looks promising at this stage". That is no denegration of the individual's knowledge. Many of the forecasts are presented in such a good, scientific, way as to be wholly believable, if you don't question the accuracy of the underlying techniques. It is just a fact.

There is no accurate way of predicting what this coming winter will be like.

If an organisation gets it right 9 times out of 10 I'll start taking notice of the methods they are using, even though it is still a really small dataset. Get it right 16 times out of 20 and they'll be approaching a low level of significance. Keep that at an 80% correct level up for 30 years and it really will have a usefulness. Any higher than 80% and you've just about cracked it.

If anyone gets this winter correct, we should say "well done", smile and then forget it, because, statistically, it won't mean doody.

Paul

Edited September 5, 2006 by Dawlish

[parmenides3]

Do you think that looking at past comparable seasonal patterns is a better way of hopecasting, or are we all p@ss@ng in the wind?

P

[Dawlish]

Do you think that looking at past comparable seasonal patterns is a better way of hopecasting, or are we all p@ss@ng in the wind?

P

Hi P3,

Nobody is ever wetting the onshore wind, in producing these forecasts. It is all useful, because, out of it, something will eventually come that is really useful... but it will take time and my point is that we aren't there yet, which is why we must qualify our forecasts on here.

I wish I could refer to past articles in "Weather" where researchers have attempted to do just that; but the mice, in the garage, ate through the plastic box I'd put them in and chewed them for nesting material (as well as peeing all over the rest), but I don't know of a single study that has shown that using past seasonal conditions can predict future conditions with any useful level of accuracy. I could be wrong and I'd appreciate a link, from anyone, to any researchers that have managed this.

When you think of it, if any had been proven to be statistically significant, they would be being used by organisations today to make money. After all, businesses paid the Met Office a lot of money for advance warning of that coldest winter in 10 years. They aren't paying anyone for matching forecasts, which implies, to me, that there is no way of doing this.

Paul

[parmenides3]

Hi P3,

Nobody is ever wetting the onshore wind, in producing these forecasts. It is all useful, because, out of it, something will eventually come that is really useful... but it will take time and my point is that we aren't there yet, which is why we must qualify our forecasts on here.

I wish I could refer to past articles in "Weather" where researchers have attempted to do just that; but the mice, in the garage, ate through the plastic box I'd put them in and chewed them for nesting material (as well as peeing all over the rest), but I don't know of a single study that has shown that using past seasonal conditions can predict future conditions with any useful level of accuracy. I could be wrong and I'd appreciate a link, from anyone, to any researchers that have managed this.

When you think of it, if any had been proven to be statistically significant, they would be being used by organisations today to make money. After all, businesses paid the Met Office a lot of money for advance warning of that coldest winter in 10 years. They aren't paying anyone for matching forecasts, which implies, to me, that there is no way of doing this.

Paul

Thanks, Paul. I suppose you could argue, though, that pattern-matching is a part of the climate modelling process, so in that sense there would be a connection, but it would be more related to trends than straight 'matches', and would undoubtedly not be any more reliable than other methods. After all, we do have a good idea of how the NAO drives the weather, though without knowing what drives the NAO, it would be hard to look much beyond a week or two. Perhaps this might be the next 'breakthrough?

P

Edited September 5, 2006 by parmenides3

[VillagePlank]

Thanks, Paul. I suppose you could argue, though, that pattern-matching is a part of the climate modelling process, so in that sense there would be a connection, but it would be more related to trends than straight 'matches', and would undoubtedly not be any more reliable than other methods. After all, we do have a good idea of how the NAO drives the weather, though without knowing what drives the NAO, it would be hard to look much beyond a week or two. Perhaps this might be the next 'breakthrough?

P

I think that Dawlish has a very good point. The NAO, for sure, is simply a description of where the H, and L. That after all is the weather. I'll come back in a few days for my more accurate thoughts on teleconnections based on complexity, emergence, and non-linearity; they do not make pretty reading for anyone who likes LRF, and support almost exactly what Dawlish was saying.

My previous post was not a prediction, it was simply a pointer of what to look for given synoptic trends. That was it nothing more, and nothing less. I apologise for not making that more clear.

[Turnedoutniceagain]

http://www.meto.gov.uk/weather/seasonal/wi...06_7/index.html

A date for your diaries

[johnholmes]

re this

I wish I could refer to past articles in "Weather" where researchers have attempted to do just that; but the mice, in the garage, ate through the plastic box I'd put them in and chewed them for nesting material (as well as peeing all over the rest), but I don't know of a single study that has shown that using past seasonal conditions can predict future conditions with any useful level of accuracy. I could be wrong and I'd appreciate a link, from anyone, to any researchers that have managed this.

I am of the belief that there has never been any 'real' connection between any preceding month, wetter, drier, warmer, colder, etc etc, for any subsequent month. Not enough to set a rule in being. Yes there are occasions when there does appear to be a link but also those when it does not appear.

There was never any empirical rule within the Met Office which would allow any possible prediction to be possible. That must be obvious as that approach is never used.

That is quite different to trying to link various what have become known as 'teleconnections' for attempting to do any LRF.

John

Edited September 5, 2006 by johnholmes

[The Enforcer]

The current CET compared to the same period from 1997 is:

1997: 11C

1998: 10.9C

1999: 11C

2000: 10.7C

2001: 10.2C

2002: 11.1C

2003: 11.2C

2004: 10.9C

2005: 10.8C

2006: 10.7C

But when you compare with the long-term it doesn't make pretty reading at all:

Look how minute these drops are in comparison with the massive increases that got us into this mess in the first place. If it starts falling by comensurate amounts (which is unlikely to apply to this year) then perhaps we can start to entertain the realistic return of winter proper.

Plus CET is really a sideshow, as clearly many warm spring/summer/autumn months in a year can inflate the figure, but not prevent the key winter months from being snowy, which is what is really important to most of us.

[Damien]

Thanks for your reply Dawlish.

Much as I enjoy reading your posts, there is a great deal of hopecasting in them; you tend to filter what you don't wish to see and exaggerate some things that you hope will prove that your wishes may be realised.

I do like your posts Damien, don't get me wrong, but there is so much wishing for particular event to occur, that I feel it is clouding your judgement!

Interesting. Can you quote examples?

Maybe all of the many times when I've mentioned that "the NCEP forecast won't come off" or "this forecast is implausible" (as in the case of ECPC's shockingly cold output at the weekend) or "I don't trust the Martin Mallett forecast" would be a good starting point?

(Actually "you tend to filter what you don't wish to see and exaggerate some things that you hope will prove that your wishes may be realised" can be said to be true when those exceptionally "red" global IRI charts were published. I pointed out in the grand scheme of things a minute cold anomaly (which was actually more just a moderately below average anomaly if anything ) over Western Kazakhstan, and said that "it could possibly bring us some easterly winds" being their a cold(/slightly cold) anomaly of some sorts over that region. That was just making the most of a bad situation to be honest with you - I will admit that.)

I'll, bet somewhere, deep down, you still believe that we had a spell of Easterlies forecast in mid September.

Of course - the more easterlies the better; and this can bode well for the winter ahead.

(What's that I hear you say, European-type easterlies can give us warmer weather in September? Well, yes - but we can all benefit from that: the warm-lovers, and the people who are looking out for easterly synoptics as a sign of what may be to come a little further down the line. )

I wouldn't mind betting that, had I not asked where they had gone, you would refer to that non-forecast, during the winter, especially if the winter is colder than average, to back up your wish that a pattern change is under way.

Yes I would, because, as I have said before, the models are picking it up - even though it may only be the models that pick it up. However, at the end of the day, if March 1, 20?? comes and we have a had a mild winter, with the easterlies only remaining in FI, as much of last winter was, in a situation that is, as the coming winter will be if this does occur, once again mild or chilly but snowless like 2005/06 was, then we can start to ask questions about the potency of the models amid those of our future climate.

As you say, "There is a big difference between a long-range gfs chart and a forecast". Your own words put the point over very well.

Now if those Easterlies were to reappear and stick, that's another matter. I'd then be interested.....and it could happen, though the odds are against it, as they are with any random chart event making a reappearance and becoming reality.

But does that include the warmer than average September that you mention below, based on the latest runs?:

There is no statistical evidence that a cooler and wetter September should lead to a ceratin type of winter....and this September is unlikely, as it stands today, to produce that.

(But I've got to admit that the former is historically true. The long-awaited cold and snowy spell of late February 2005 first came on the charts in about mid-January, and only came to fruition around late February of that same year.)

PS I don't quite get your reference to posts in the Arctic ice-watch section?

Sorry, I meant the "Carinthian's Latest Arctic Reports" thread. There's always some interesting discussion on there and it's good to have your posts and carinthian's back.

PPS I've never posted on TWO and I say nothing different on UKww to what I say on here.

Okay - but I was just saying that would you post that all the people who comment on the long-range and FI/far FI forecast models are "hopecasting" as to realistically commenting on the, albeit perhaps sometimes far-fetched and slightly unrealistic, model output?

P.S.: NCEP is now showing September to be much milder too. November and December are typical "mild-cold" battles with the east and cold victorious in January but the Atlantic reclaiming it's still warm throne in February.

I'll have something significant-ish on Friday, fingers crossed.

Edited September 5, 2006 by Damien

[ribster]

Well, I'm Hoping for some snow, ice (the odd icicle too!) and some generally cold weather this Winter. I'd like to think I can continue to have these hopes, without the self-appointed weather police jumping all over me!

Edited September 5, 2006 by ribster

[fuzion]

yeah i hope this winter will be a great one for snow allover the uk

cus last year here was a let down but we have to have hope

[Damien]

But when you compare with the long-term it doesn't make pretty reading at all:

Enforcer, I agree with most of what you say, but the exceptionally cold CETs/winters of the 1890s and earlier were caused by massive amounts of floating ice in the northern North Atlantic, in our case going down towards Britain. In fact Octobers quite often brought snow to the South-East and London in those days. Comparing the modern day October/winter to such periods is ludicrous; unless, as you say, something starts to drastically change and soon. But it doesn't look likely does it - certainly not in our little part of the world.

By the way, do you have any data on this for Europe, in particular the regions:

-Central Europe,

-Eastern Europe,

-The Balkans.

TY.

Turnedoutniceagain - thank you for the link. Why do they do this to us?

Hopefully we will get a "leak" like last year, when the Scottish newspaper reported on a MetO statement given to gas unions about the coming winter which did not correlate the "mild" dynamic August run. I was the first to spot that BTW, and what happened next?

[The Enforcer]

Enforcer, I agree with most of what you say, but the exceptionally cold CETs/winters of the 1890s and earlier were caused by massive amounts of floating ice in the northern North Atlantic, in our case going down towards Britain. In fact Octobers quite often brought snow to the South-East and London in those days. Comparing the modern day October/winter to such periods is ludicrous; unless, as you say, something starts to drastically change and soon. But it doesn't look likely does it - certainly not in our little part of the world.

By the way, do you have any data on this for Europe, in particular the regions:

-Central Europe,

-Eastern Europe,

-The Balkans.

TY.

You don't need to go that far back - look at the sudden change at the end of the 1980's.

I lifted that from the metoffice - so no info on Europe. Besides, it is what happens here that counts. Europe can often be much colder than here in winter.

Edited September 5, 2006 by The Enforcer

[Damien]

You don't need to go that far back - look at the sudden change at the end of the 1980's.

Good point - certainly an irony once the establishment started going on about global warming around the late 1980s did our winters suddenly turn a lot milder - and Communism was defeated in the Cold War and we all united like good capitalists to fight global warming/the death of the British winter.

I miss snow and socialism dearly. (*Plug over.*)

Besides, it is what happens here that counts. Europe can often be much colder than here in winter.

Yes, to our enjoyment in between December-February/(November-March?) it doesn't mean much (except to annoy us all once again, LOL); but ask yourself what it means in term of the wider global warming/climate change debate?

The same goes for North America and other continents - to see some figures on these areas would be nice.

Edited September 5, 2006 by Damien

[Dawlish]

My previous post was not a prediction, it was simply a pointer of what to look for given synoptic trends. That was it nothing more, and nothing less. I apologise for not making that more clear.

I know, I realised that and I avoided any direct criticism of what you'd said. I look forward to your post as I see chicken and egg all the time in teleconnections. Does the weather drive the teleconnections, or do the teleconnections drive the weather? Perhaps it doesn't matter and the outcome is all, but to rely on teleconnections to predict the weather might be the real innacuracy and it may create such complexity that it prevents teleconnections being the holy grail of LRF. Perhaps computer modelling will take this far further than human interpretation ever could?

Paul

[Cecil Hogmanay Melchett]

Damien,i to love the snow and your'e posts are always informative.HOWEVER,why do you constantly make references to communsim/capatlism and the cold war?And could you please xplain your 'i miss socialism'remark,because if your refering to communsim in the eastern block then i can assure you many of my own family were persecuted under the communist regime as were millions of others in satellite states across eastern europe.

(If ive got you wrong then please accept my apologies).

[Stratos Ferric]

Good point - certainly an irony once the establishment started going on about global warming around the late 1980s did our winters suddenly turn a lot milder - and Communism was defeated in the Cold War and we all united like good capitalists to fight global warming/the death of the British winter.

I miss snow and socialism dearly. (*Plug over.*)

Yes, to our enjoyment in between December-February/(November-March?) it doesn't mean much (except to annoy us all once again, LOL); but ask yourself what it means in term of the wider global warming/climate change debate?

The same goes for North America and other continents - to see some figures on these areas would be nice.

Damien,

There's some good plots and data on this site...

http://carto.eu.org/article2509.html

I think it's fair to say, if you go and view other plots, that the temperature has been rising recently right around the globe with only a few localised exceptions. The UK does seem to be warming slightly more than continental Europe, but with a more maritime climate this can easily be rationalised; the thermal capacity of the oceans is a lot greater than that of land, hence changes in the surface envelope will be stored more readily in the oceans than the land. There is, however, a mounting pile of evidence suggesting to corroborate the warming - see the posts made recently in, I think, the enviroment pages regarding the lengthening growing season / shortening winter in Europe as a whole.

Re the LRF debate - I'm completely with Dawlish on this. His argument is one that we trot out on here every year as the first of the inevitible rash of hopeful projections for a cold / cool winter appear.

If you were to retro analyse forecasts on here over the past 3 years against outturn you would see I think a significant bias to cold forecasts that are seldom matched in reality. This tends to repeat on other sites too. Further more, as Dawlish suggests, if anyone had found the holy grail for seasonal forecasting they'd be cashing in big time by now and we'd all know about it. They aren't and they haven't. That said, I do await with interest the UKMO's seasonal outline; whilst it is very (and sensibly) general, if they score another inner / bull's eye again this year they will become a body to watch when it comes to LRFs.

[Scorcher]

Comparing the modern day October/winter to such periods is ludicrous; unless, as you say, something starts to drastically change and soon. But it doesn't look likely does it - certainly not in our little part of the world.

Why would you want October to be colder than it is? It's an autumnal month and not a winter month. I think a warm first half and a cooler second half would be fitting for October, but there certainly should be no snow. Cold Octobers are depressing for most of the population as there are 4 or 5 months of constant gloom and cold which follow it, I really can't understand people wanting cold conditions before late November/early December.

[Roger J Smith]

In these discussions, we keep reading statements that suggest that there are no reliable or accurate long-range forecasts available "yet."

Is it not more correct to state that the orthodox science of meteorology has not yet identified any reliable or accurate long-range forecast systems, although they tend to give their own a very good review?

The truth of the matter is that dozens of people issue long-range forecasts for every month and every season in the UK, and in North America too for that matter. Then, for the most part, people forget all about them and don't do any kind of a review of them after the fact.

This problem is exacerbated by the fact that there is no set format for issuing and thereby validating LRFs. One person gives an anecdotal account without numbers, another talks in terms of a seasonal average, another gives some monthly anomaly predictions, and yet another offers up weekly predictions.

When people say categorically that nobody has demonstrated much skill, what is that based upon? Circular reasoning -- "I know these LRFs can't be very accurate, so I'm not going to waste my time reading them over at the end of the season, I'll just assume they are wrong or random."

I call that the Canadian substitute for the scientific method, since unfortunately I have had that applied to my work when I know for a fact that a seasonal forecast has been obviously non-random. This has been going on for twenty-five years in my case, to the extent that I keep checking the calendar to see if it's 2006 or 1609.

I think I'd be luckier if it were 1609 (I know, for one thing, there would be more snow in the winter).

We have some summer forecasts "pinned" on this site -- who's saying all of these are junk? Mine wasn't, so put that in your computer and resolve it.

[Dawlish]

In these discussions, we keep reading statements that suggest that there are no reliable or accurate long-range forecasts available "yet."

Is it not more correct to state that the orthodox science of meteorology has not yet identified any reliable or accurate long-range forecast systems, although they tend to give their own a very good review?

The truth of the matter is that dozens of people issue long-range forecasts for every month and every season in the UK, and in North America too for that matter. Then, for the most part, people forget all about them and don't do any kind of a review of them after the fact.

This problem is exacerbated by the fact that there is no set format for issuing and thereby validating LRFs. One person gives an anecdotal account without numbers, another talks in terms of a seasonal average, another gives some monthly anomaly predictions, and yet another offers up weekly predictions.

When people say categorically that nobody has demonstrated much skill, what is that based upon? Circular reasoning -- "I know these LRFs can't be very accurate, so I'm not going to waste my time reading them over at the end of the season, I'll just assume they are wrong or random."

I call that the Canadian substitute for the scientific method, since unfortunately I have had that applied to my work when I know for a fact that a seasonal forecast has been obviously non-random. This has been going on for twenty-five years in my case, to the extent that I keep checking the calendar to see if it's 2006 or 1609.

I think I'd be luckier if it were 1609 (I know, for one thing, there would be more snow in the winter).

We have some summer forecasts "pinned" on this site -- who's saying all of these are junk? Mine wasn't, so put that in your computer and resolve it.

Yup, he says, giving a pat on the back and smiling. Never stop looking at the sea and wetting the onshore wind Roger!

May I refer you to 2 paragraphs of an earlier post of mine:

"If an organisation gets it right 9 times out of 10 I'll start taking notice of the methods they are using, even though it is still a really small dataset. Get it right 16 times out of 20 and they'll be approaching a low level of significance. Keep that at an 80% correct level up for 30 years and it really will have a usefulness. Any higher than 80% and you've just about cracked it.

If anyone gets this winter correct, we should say "well done", smile and then forget it, because, statistically, it won't mean doody."

Paul

[VillagePlank]

re this

I wish I could refer to past articles in "Weather" where researchers have attempted to do just that; but the mice, in the garage, ate through the plastic box I'd put them in and chewed them for nesting material (as well as peeing all over the rest), but I don't know of a single study that has shown that using past seasonal conditions can predict future conditions with any useful level of accuracy. I could be wrong and I'd appreciate a link, from anyone, to any researchers that have managed this.

I am of the belief that there has never been any 'real' connection between any preceding month, wetter, drier, warmer, colder, etc etc, for any subsequent month. Not enough to set a rule in being. Yes there are occasions when there does appear to be a link but also those when it does not appear.

There was never any empirical rule within the Met Office which would allow any possible prediction to be possible. That must be obvious as that approach is never used.

That is quite different to trying to link various what have become known as 'teleconnections' for attempting to do any LRF.

John

Very true in the context of this debate, John.

However, there are long term patterns for all to see - the most obvious being that winter months invariably have a colder CET than summer months. It may seem obvious, but it's a perfectly, and demonstrable example of a pattern prediction that holds true virtually all of the time. I could claim that in 10 years time, the Dec CET for 2016 will still post lower than it's subsequent June value.

Whether we like it or not we can postulate predictions that are true for most of the time.

The devil, as they say, is in the details, and this, I presume, is where teleconnections attempts to join the party.

Is it not more correct to state that the orthodox science of meteorology has not yet identified any reliable or accurate long-range forecast systems, although they tend to give their own a very good review?

I agree with this in its entirety. The methodology, the approach, for long-term forecasting by definition is wrong. If this were not the case then we would not have the case of diminishing returns . . .I argued this point here Edited September 6, 2006 by Wilson

[Glacier Point]

One aspect of the current modelling I find quite encouraging is the persistence of height anomalies over Greenland and a strong slow moving or even stationary Icelandic low helping to draw down some of that anomalously cold Arctic Air into the Greenland High.

Arctic surface temp anomalies August

With the re-emergence of the (SSTA driven) summer pattern into September and the east-Atlantic trough-west European ridge, this will help to maintain the longitudinal pattern of Greenland High - Icelandic Low in turn perpetuating cycles of cold air development over Greenland.

The 06Z GFS demonstrates this quite nicely:

cold pool building.......

http://www.wetterzentrale.de/pics/Rtavn002.html

http://www.wetterzentrale.de/pics/Rtavn1202.html

http://www.wetterzentrale.de/pics/Rtavn1802.html

dispersal.....

http://www.wetterzentrale.de/pics/Rtavn2282.html

re-building......

http://www.wetterzentrale.de/pics/Rtavn2762.html

http://www.wetterzentrale.de/pics/Rtavn3482.html

This run has been used to illustrate the theorectical process and is no guarantee of what might happen but it is a trend worth looking for as we head towards October. Notice even at t120 we have -20 850 values. Repeat that cycle several times and a sizeable cold pool could have developed by November.

GP

Edited September 6, 2006 by Glacier Point

[Snowyowl9]

Looking good, this time last year wasn`t nowhere near as cold as that up over greenland.

B)

S9