The Winter 2011/12 Forecast Thread

[Catacol]

As far as forcasting lows and storms on certain dates is almost certainly more luck than

judgement unless we are in a zonal flow which we are now. This is when low pressures come

along one after the other so to say we will have a storm in two or three days say will be a

fairly accurate guess. As far as forcasting the weather by phases of the Moon etc is imo

total poppy cock.

The poppy cock lies in this statement above. At what point when the forecast was put together was a storm "in a few days" predicted? Do you mean "in a few weeks" and if so you are right - a prediction of 2 weeks of stormy weather was put together and it duly arrived 4 weeks after the prediction, followed by a prediction of several days of cold air - which also arrrived - and finally a remark about a possible storm early next week... and guess what? - that looks possible at this stage too, though maybe 24 hours later than predicted weeks and weeks ago.

Going back to last year Dec was accurate in my book, and if you actually look at the January charts from 2011 you will see accuracy there too. The transitional energy peak that was forecast for 17-20 Jan arrived on about the 18 and from there the forecast did go awry... but Roger himself hinted a couple of times that he was not totally happy with his Feb forecast and that he had not listened to his gut instincts about it: and these were comments made before Feb got close, not after. So 1/3 of last winter spot on, 1/3 close, 1/3 off target (but suspected so beforehand). I rate that as not bad... for poppy cock that is.

I mark academic scripts on a regular basis at work, and the kind of vague, baseless argument put together here by you CC is a head shaker at the best of times. If you want to suggest that someone else's forecast technique is off target you need to do a whole lot better than this. Some evidence would be a good start; you wont find any for this year yet that you can make use of, but come the end of Feb perhaps you could revisit this thread and put together something constructive.

If I am an reader of personality Roger will be the first to hold his hands up if the forecast goes wrong (unlike many others around in the media at present) but let's not forget that he is currently targetting 60 - 70% accuracy across the season so any criticism will need to be on the basis of a failure to meet this. And let's be fair: 60 - 70% accuracy across a season is damned good (if it comes off) and it makes me wonder why the likes of Maddon and Corbyn get a ton of publicity for putting out an awful lot of rubbish, and yet someone with a track record of reasonable accuracy gets ignored. Perhaps you need your own website Roger??

Edited December 23, 2011 by Catacol_Highlander

[leicsnow]

In Mark Vogan's latest YouTube vlog he completely admits he has gotten some things wrong. For that I admire him, unlike guys like James Madden who just think of cunning excuses and gradually twist their words.

He really is no publicity seeker - no James Madden/Piers Corbyn.

IMO he has good methods, and has gained a good knowledge of UK weather and climate but still has a lot to learn. I think we can expect better quality and more accurate forecasts from him in the future.

Edited December 23, 2011 by Tellow

[Roger J Smith]

Two points, the methodology being used does not refer to phases of the moon, it uses some data (but held to be about 20-30 per cent of the variance) for lunar position relative to other sources of energy in the model, but phase would be incidental in that.

Last winter's forecast of very cold conditions late November through December was issued a lot earlier than mid-November, that is a quote from a post perhaps but I can show that the forecast was in existence in October. It was not based on anything available on global models, I never consult them in making seasonal forecasts (usually because they are made 45-60 days before startup date).

Anyway, if somebody as slow to jump on bandwagons as JH thinks there is something to the methods after observing about seven years of my work here, then I think that says a lot, and certainly there is no claim being made by me or Fred about exceptional levels of accuracy. We're in an environment here where some players are making extravagant claims that are almost self-evidently inflated. So in fact I would reverse the paradigm and ask, do people really think that it makes sense to ignore this methodology at this point in time given the track record? And by the way, last winter, January was not a bust if you look at the details there were essentially good predictions of regime changes that took place on two occasions, not to the nearest day or with all the details exactly correct, but then we're expecting (reasonably) that a technique like this will merge from behind with medium-range accuracy, it would be unreasonable although technically possible to expect a statistical model like this to leapfrog medium-range model performance (with its known foibles, prime example the hurricane that never was last week) and mimic short-range performance, but I have to realize also that if I give specific dates then people will unconsciously compare to short-range rather than medium-range performance, whereas I tend to give the dates a medium-range sort of leeway on all counts. After all, the objective is to have a working model that is useful in the same general sense that medium-range modelling today is useful, and that has improved by a significant factor since I started forecasting in the 1970s (we never paid any attention to five-six day output then, it was basically FI style and in the bad old days even the 48h time scale on models could not really be trusted very much, I remember a storm in April 1979 in the eastern U.S. that went off on a 45 degree angle to the model runs and ended up giving southern Ontario a snowstorm instead of a warm spring day ... that sort of thing was almost a weekly occurrence in those days.)

I think where this is all heading is a sort of meet-up of conventional modelling gradually extending out beyond 5-7 days with reliability (my feeling is that 8-10 days is now coming into focus on global models as better than a shot in the dark) and a technique like mine which is essentially "black-box" and would only go off an average performance curve due to large external changes (something like a big singularity or a large shift in the grid being used). Whether that would happen because of my non-research efforts or some second generation of work when I'm done, or an independent person's work, remains to be seen, but I have to say, I don't think this will happen, I know it should happen.

The CFS is an independent effort to accomplish the same set of goals, but the white Christmas thread seems to establish which of the two is ahead in that race at this point in time. As perhaps stated earlier (can't remember) post 174 in that thread can be used to establish that, in context of similar time frames, my methodology was ahead on that one occasion at least. There is also a documented case of a high correlation map from back in December 2005 at 32 days using these methods. Some other experiments have seen my method behind other attempts although showing a positive correlation. The problem with those comparisons is that some of the other forecasts are hunchcasts taking high-frequency pattern matching, so while hats are always off to those who speculate correctly, there is no theory to compare. My forecasts are nowadays entirely theoretical, or closer than 90% to that, using data analysis -- if I have a "hunch" then I test it out here by a data analysis (for example, the hunch that extreme warmth in early October might prove to be a reliable signal of a cold winter, I tested that out, found it was not statistically very significant although the hunch was right side of random).

If I could just mention one thing here about past cold winters, I had a close look at all cases since daily data became available (1772) to see how very mild days around 22-31 Dec factored into future outcomes for January, and found (perhaps not surprisingly) that there is no negative indication from this, in fact there are an encouraging number of examples of very mild Christmas seasons being followed by severe cold often starting around 4-7 January. One particularly good one is 1819-20. I would say there's a subset of winters that turn cold around mid-December to 20 December and then stay very cold well into late January if not early or mid February, such as 1794-95 or 1962-63.These then become so-called "great" winters by definition. But there seems to be an equal cohort of flip-flop cases, and just a small sampling of what you might call near normal to very cold transitions. This may signal that if deep cold is not already established at this point in the winter, a fairly reliable scenario to get there is a pattern reversal. The daily CET on 30 Dec 1986 was just shy of 10 deg also.

I can promise that some detailed graphs on model factors will be made available in 2012. I have these ready now but I am waiting for the 2011 data to finish up so that I have exactly 240 years of data to make 40-year segments equal in weight. Of course at this point the changes to existing data sets will be almost negligible so I know what these data analyses will look like. And they are quite significant in appearance in some cases. The lunar cycle analysis has presented some interesting speculative outcomes to ponder. There appears to have been a slow change in how lunar declination works on the circulation and I could relate that to the slow warming trend over time throughout the 240 years, not to get into details here but it may show a subtle process at work in ocean-atmosphere coupling that performs a very slow hemispheric transfer (from southern to northern) of heat that may have peaked in the recent decades and now may be reversing. This is a very long-term and "background" process that nobody would notice in operational time scales. But very gradually, the declination signature has changed and I'll try to organize that presentation for some time in 2012 so that interested parties can see what I mean. But the modern data show that lunar declination is not a negligible factor, however it tends to be more significant in winter and summer when it aligns with the tidal maxima (full and new moons). It's on the order of 0.5 to 1.0 C deg over decadal time scales. But I use at least half a dozen other factors in my research model that have greater amplitudes, some approaching 1.5 C deg. This is why I keep saying, the lunar factors are not anywhere near being the full story of external energy modelling and any of the competition who use only lunar data are bound to suffer poor if not random results, they are leaving out what I would consider 70-80 per cent of available variance (and admittedly there are very likely other external factors not yet diagnosed and in my research model, so that could be more like 80-95 per cent).

But at the same time, and this is bound to be confusing on first glance, once you have the circulation prediction essentially right, then the lunar timing variables really "kick in" to make for a good end result on details. It is absolutely essential that you solve the pattern question before applying the lunar variables, otherwise, you don't know which subset of data to paint in. The subsets are a lot more "high amplitude" than the entire data set. I think this is why even sympathetic observers of my forecasts (and myself in validation procedures) find the forecasts to be a mixture of surprisingly accurate in detail and on occasion just way off the mark in general. Without the right pattern the short-wave modelling is useless. So that's the challenge of doing things this way. You'll have to admit, there is little if any reason to believe that you could get to a point of accurate details at 30-90 days if you don't have a driving mechanism on a day to day time scale. Large-scale processes do not have inbuilt theoretical reasons for events to be on day 37 and not day 38 if you follow my drift.

[Weather-history]

One particularly good one is 1819-20. I would say there's a subset of winters that turn cold around mid-December to 20 December and then stay very cold well into late January if not early or mid February, such as 1794-95 or 1962-63.These then become so-called "great" winters by definition. But there seems to be an equal cohort of flip-flop cases, and just a small sampling of what you might call near normal to very cold transitions. This may signal that if deep cold is not already established at this point in the winter, a fairly reliable scenario to get there is a pattern reversal. The daily CET on 30 Dec 1986 was just shy of 10 deg also.

Murphy's winter 1837-38 is a very good example of a very mild Christmas period quickly giving ways to a very cold period in the New Year.

[feb1991blizzard]

I think that recent years are the most relevant indicator if we are looking at forecasting forthcoming months. A mild December occuring since 1987 has led to the subsequent Jan and Feb being mild - sometimes exceptionally so - in 22 out of 24 cases. The 2 exceptions could ony be defined as ''average' months.

Some people are happy to quote never ending statistics but become uncomfortable when those stats support a theory that they wish to deny - not you Roger !

If feb 1994 is one of them as you said in an earlier post then i disagree, it was cold with a potent easterly around the 11th, -12 uppers in central southern England and snow in the IOW, heavy falls in the midlands, a pasting for thee pennines and a reload a week later if i remember rightly, however you are right with your prediction of a mild end to this december to be fair.

[Weather-history]

I think that recent years are the most relevant indicator if we are looking at forecasting forthcoming months. A mild December occuring since 1987 has led to the subsequent Jan and Feb being mild - sometimes exceptionally so - in 22 out of 24 cases. The 2 exceptions could ony be defined as ''average' months.

Some people are happy to quote never ending statistics but become uncomfortable when those stats support a theory that they wish to deny - not you Roger !

Support a theory that they wish to deny? How can you support a theory that doesn't exist? I am going to take the bait and let you choke on the facts...

Facts are

a) there has been no signfifcant pattern change. We still have an Icelandic low/Azores high as we did about 100 years. I will gladly post a map from 1913 showing those features. So where is the significant pattern change? If there is any change it is in the intensity of these features, thats all. The Icelandic low hasn't suddenly popped up after spending decades over Scandi . The increasing strength in the westerly flow theory goes back to the late 1930s by the way.

This is what you said on Ukweatherworld in early 2008

" Very interesting Kev, those deep purples over Scandi are now only reserved for Greenland - and Greece is mild. What a reversal in the m@dern era"

And two years later we had extensive northerly blocking, something you said was nigh on extinct. Then you said the cold stratosphere is responsible for these m+dern winters and when any sign of warming in the stratosphere, you dismiss it as not having any influence,. Have you ever heard the saying "trying to have your cake and eating it?"

And if there is a theory knocking about Ian, its called global warming........, no one is going around talking about Ian Brown's even larger teapot theory" in the general public except for yourself and a couple of hacks.

And this is what I said in response to that post in another thread.....

And often when it didn't exceed 5.0C, January or Feburay or indeed both was mild.

Such as Decembers 1989, 1991 (February 92 was mild), 1992, 1996 (February 1997 was mild), 2001, 2003, 2007. So what does that tell us?

January or February was very often mild in that period, whilst December was much less so. That is all that tells us

Edited December 24, 2011 by Mr_Data

[Liam J]

Not looking good at all for the cold fans amongst us, Good job it's Christmas and minds will be elsewhere over the next few days.

The current set up couldn't be any more different from last years remarkable spoils!

I know it's far to early to be writing the rest of Winter off just yet but, I wouldn't be surprised if this Winter turns out to be mild and above average, perhaps well above.

We are going to need a huge synoptic turn around of epic proportions soon within the outputs, people seem to be ramping on about strat warming but that has a lag effect of around 4-6weeks. Perhaps February will be our star Winter month with the cold express visiting the UK for a lengthy stay?

This could be a throw back to the naughty 90's Winters, Mother Nature is delivering a huge dose of spite after last December!

Never fear folks, still 2 months left to turn things round....

Edited December 24, 2011 by Liam J

[Backtrack]

Not looking good at all for the cold fans amongst us, Good job it's Christmas and minds will be elsewhere over the next few days.

The current set up couldn't be any more different from last years remarkable spoils!

I know it's far to early to be writing the rest of Winter off just yet but, I wouldn't be surprised if this Winter turns out to be mild and above average, perhaps well above.

We are going to need a huge synoptic turn around of epic proportions soon within the outputs, people seem to be ramping on about start warming but that has a lag effect of around 4-6weeks. Perhaps February will be our star Winter month with the cold express visiting the UK for a lengthy stay?

This could be throw back to the naughty 90's Winters, Mother Nature is delivering a huge dose of spite after last December!

Never fear folks, still 2 months left to turn things round....

March can sometimes be the snowiest Winter month, so make that 3 months to turn things around.

[Optimus Prime]

I'm just talking in CET terms for the months in question - not got the figure to hand but the return was something like 3.5C which I would class in the 'average' zone.

February 1994 was actually 3.2c. Certainly in the rather cold zone.

[feb1991blizzard]

I'm just talking in CET terms for the months in question - not got the figure to hand but the return was something like 3.5C which I would class in the 'average' zone.

3.2 Ian so a 0.6 negative anomaly, not bitter overall but then again feb 2009 was 4.1, so actually a mild month but i think that most sensible cold and snow fans on here would settle for a repetition of either of those months with the obvious caveat that not everwhere saw snow during those months. I think anyone looking for a sub zero or even close to zero monthly CET is delusional and going to be disappointed, december 10 was a very rare event, from my point of view i am hoping for a week or 2, or if we are really lucky 2 seperate spells of high latt blocking with uppers cold enought to drive widespread snow across the country, between now and the end of the winter, of course it may not happen and we could have a repeat of last year but my main point is, you dont have to have a bitter winter to see a severe snow / cold event.

[leicsnow]

Indeed, even though February 2009 CET was around average most would settle for a month like that again just for similar weather to the first half of that month.

I think a lot of people's dream winter would be like this:

December 2010

January 1987 (especially for those down south)

February 1986

Edited December 24, 2011 by Tellow

[feb1991blizzard]

Indeed, even though February 2009 CET was around average most would settle for a month like that again just for similar weather to the first half of that month.

I think a lot of people's dream winter would be like this:

December 2010

January 1987 (especially for those down south)

February 1986

I think i would swap 86 for 91 and although 2010 was exceptionally cold, i do remember my only proper thundersnow event was december 8th 1990 and although i lived in the midlands then, i think up here got plastered ridiculasly, although that was one small event in an otherwise zonal stormy month, Xmas day 1990 believe it or not i had inches of hail piling up. 87 the best january though and by some distance.

[Bristawl Si]

87 the best january though and by some distance.

Not down here in Bristol.

January '79 by some distance (and if we include 30/31 December '78 added onto that January) - by a mile!

[feb1991blizzard]

Not down here in Bristol.

January '79 by some distance (and if we include 30/31 December '78 added onto that January) - by a mile!

Sorry i was too young to remember that, 80s onwards for me and even early 80s i only vaguely remember, i have never been through the charts of 79, i will have a butchers at some point over Xmas.

[Bristawl Si]

Sorry i was too young to remember that, 80s onwards for me and even early 80s i only vaguely remember, i have never been through the charts of 79, i will have a butchers at some point over Xmas.

Yeah, take a look at some point - some classic blizzards driven by a potent easterly if i remember correctly.

As for this winter i don't hold out much hope down here for cold and snow. Maybe the odd incursion from the north/north west.

[Snowmad79]

What sort of tone is that ST79? The forecasts are made not using GFS, ECM etc...so what's your point?

BFTP

No tone pal, just a little criticism. 1st point was that your/RJS forecast last year wasn't as accurate as some people have made out and secondly your tendency to jump on any output and declare a cold blast is imminent even when mild is the outlook and lead people on without actually giving any information on why and how it will happen.

On the flip side of the last comment, I do admire the positive spin your comments add to the threads even if 99.9% of the time they are hope-casts lol

[BLAST FROM THE PAST]

No tone pal, just a little criticism. 1st point was that your/RJS forecast last year wasn't as accurate as some people have made out and secondly your tendency to jump on any output and declare a cold blast is imminent even when mild is the outlook and lead people on without actually giving any information on why and how it will happen.

On the flip side of the last comment, I do admire the positive spin your comments add to the threads even if 99.9% of the time they are hope-casts lol

I like this post too.....99.9% hopecasts...come on!!!!. Merry christmas to you...indeed to all as well.

BFTP

Kev, with regards to the Strat I said that a colder Strat was one of the factors in the progressive synoptics - not the only one. Similarly, I have been cautioning against a Strat warming suddenly bringing about in a reversal of pressure patterns that members want to see, as again there are other factors at work.

As for stable Northern blocking I will admit I was wrong in believing that we would never see it again.

Deserves highlighting and credit,

BFTP

Edited December 25, 2011 by BLAST FROM THE PAST

[syed2878]

happy christmas and a happy newyear every1. December 8th 1990 feb 1991 and 1987 was the most snowy wimters i remember and of course 2010. Well i hope newyer brings us a new weather patern wich i hope will have cold and lots of snow.

[Roger J Smith]

The bad news is that the maps for days 8-16 don't look that great. The good news is that they are maps for days 8-16.

Anyway, the critical phase is likely to be around then (7-10 January could be a turning point).

[mountain shadow]

The bad news is that the maps for days 8-16 don't look that great. The good news is that they are maps for days 8-16.

Anyway, the critical phase is likely to be around then (7-10 January could be a turning point).

Roger, is the 7-10 January when you expect the proper cold air to arrive or would that be later?

[Weather-history]

Deserves highlighting and credit,

BFTP

I think you are being far too generous because how can anyone seriously say that northerly blocking is virtually extinct based on only about 20 years of data which is nothing in climatic epoch terms? Come on!

And in any case, northerly blocking still occurred within the same time frame anyway! Look at February 1991, the winter of 1995-96, late December 1996 -early January 1997. To some limited extent during winter 2000-01, the back end of winter 2004-05.

[Methuselah]

One thing is true, I think: all those who forecast severe cold and mountains of snow during December were wrong?

[Ladyofthestorm]

Indeed Pete, I hope you all had a wonderful Christmas, if a little snowless, if the ongoing GFS output is to be believed, Nly blocking is looking a little unlikely now, I still think January will be the best chance of seeing snow in an otherwise average winter. Certainly stormy here!Sorry I havent been around my aunt died just before christmas and Ive been away working. So all go.

I wish one of those high pressures to the south of us would move westwards!

[La Bise]

Welcome back Lady, I do recall you sticking your neck out and predicting much average Atlantic weather this winter to howls of derision from the cold hopecasters and their new paradigm... :smilz38:

[BLAST FROM THE PAST]

The bad news is that the maps for days 8-16 don't look that great. The good news is that they are maps for days 8-16.

Anyway, the critical phase is likely to be around then (7-10 January could be a turning point).

Certainly looking at a very stormy period around this time, which might go on unil 12/13th I reckon but with an appearance far more like 00z than 06z, not that either are correct.

I think we will see the jet push south into New Year and storm track increasingly impacting on UK because of this. One concern I have, a niggle really, is that the storm phase remains too strong for blocking to control. Just a little niggle but it is likely to turn much colder either way IMO.

BFTP

Edited December 30, 2011 by BLAST FROM THE PAST