SnowBallz

Actually, that isn't true. Members were discussing the models, which were - at the time - arriving at solutions which proposed cold conditions arriving. The debate was very much focused - quite rightly - on whether we'd get the cold first, and the debate for snow potential would be discussed thereafter. On countless occassions, I witnessed the sage posters wisely advising to refrain from talking about snow, as - at that range - it would be futile. So, to suggest otherwise, is quite disparaging to the intelligence and experience of a number of notable members on here. You were being cautious. Great. So was everyone else, though. Of course there were the fair share of rampers and nay-sayers, but - in the main - I felt the general 'mood' to be one of wise caution, with frequent reliance on experienced members to make sense of what was clearly a very volatile weather pattern. I feel that there is a clear acceptance that the British Isles, by nature of it's geographical position, doesn't lend itself often to extreme cold, so it would be foolish to count ones' chickens before they'd been hatched. I simply haven't seen the majority doing that. I don't think it's so much that mild posters get trampled on. I think that's a bit dramatic. As I mention above, I think there is a clear shared understanding that, the majority of the time, the British Isles will experience mild, atlantic, fairly mundane and non-eventful weather. So, that becomes the baseline. Interest quite understandably grows when there is a deviation from that baseline, either in forecast or indeed if it verifies. And it isn't always cold. If major storms are foreseen, or a heatwave of some sort - interest grows. That, surely, is understandable? We are in the winter season now, and as such the 'extreme' for this season is any spell of severe cold weather. That's the deviation from the baseline, and that's what will determine interest. You can't force people to have an interest in every aspect of the weather, or every season - it's each very much to their own, and I think it should be left that way. As for learning: I think we're all trying to learn, are we not? None of us are experts though, and I think one puts themselves in a vulnerable position if they begin to believe that they are. I think it's better to take the positives out of situations and what pleases me, is the fact that this current period - whether it verified or not - actually made people want to join the forum and enter the discussion. That's got to be a good thing

I don't know why, but you seem to feel a bit victimised? What I recall, is that the charts in pretty much all outputs were strongly pointing in a very cold direction. These charts were posted, along with accompanying analysis and/or narrative. Sure there was some hyperbole, but - considering such charts are incredibly rare - I should expect that to be understandable. I don't think it adds much to the discussion to - in the face of all evidence - take a stance, almost for the sake of it, to disagree with the output. Moreover, to disagree and not to even give a substantive reason; comes across as a little bit 'devils advocate' and/or court jester. Neither adds anything to the discussion, merely irks people and rubs them up the wrong way. Are there 'rampers' on here? Absolutely, 100%. But there are also enough inherent pessimists too, to counter-balance their enthusiasm. If we're trying to be scientific about this, merely saying 'oh, well I don't think it's going to happen' without giving any tangible reason as to why, well that just doesn't really cut it. Likewise, saying 'this is going to happen!' without backing such statements up, is equally as useless. My advice would be, to try not to take it all so seriously. If people want to enjoy themselves imagining things that might never happen, then let them - and you can wallow in the satisfaction of them being wrong, if that's your tipple. If you strongly disagree with the output - that's fine, but just back-up your scepticism with something more than just a hunch or natural pessimism. Plenty more important things to be angry about, so enjoy the weather - it makes a fool of everyone!

He is being serious, because his expectations are tempered in the knowledge that - in low-res - not only does the margin of error exponentially increase, but also the GFS has a predetermined bias to explode low pressure systems. Therefore, it's wise and measured of him to take a view out to the boundaries of a range where we can have more confidence. Beyond that, as they say, is FI - and anything can (and often does) happen in FI For me, this run continues the theme and - in some respects - is more plausible, as it lacks the extreme nature (severe cold/implausibly mild) of some outputs over the last 36hrs; it's a sort of 'middle of the road' option. That should bring comfort to those who are looking for something cold and snowy in the mid-term. Looking like a cracking week coming up.

And Ian, that explains - to me anyway - why the narrative attached also appears to change so frequently. We both know it's within a probabilistic context, therefore intentionally vague and noncommittal, so there's always room for manoeuvre. I just think that, to be fair to other models and other forecasting tools, I don't find the EC32 to be all that exceptional. As I said, the CFS was advertising the strong possibility of a block, and indeed the reverse zonality solution, way back in early-mid autumn. However, as you know, the CFS moderates and fluctuates, as does the EC32 - neither are consistent; indeed, isn't it true - in a mathematical sense - that flopping implies greater confidence, where a stagnant proposition belies mathematical apathy? It's a common misconception, that consistency or recurrence is ergo a positive outcome. I do agree with you though; there's more than enough in the short-term to get excited about, and we can leave the MR & LR until such time as it becomes relevant!!

To be honest, I've no confidence at all in the EC32; its output changes as frequently as any of the MR models, and I've seen nothing from it which suggests it should be given the respect that Matt Hugo obviously does. Matt must surely recognise, that his summary of said model continually changes, therefore what exactly is that telling us? I also find it a tad disingenuous for him to suggest '10 out of 10 to the EC32 for predicting the Easterly'. Really? That being the EC32 which flirted with just about every single possible output? Throw enough darts, and one is always likely to hit the board. I cannot believe Matt doesn't notice this, as he's typing his summary out. So the latest, from fabled EC32, is a moderation. This is in contrast to the extreme it was advertising not-so-long-ago, which is consistent with the recent amendments to the MR models. It's with facts like this in mind which lead me to question why any exceptional credit is given to the EC32. Moreover, the CFS advertised the strong Easterly chances for December at least 60 days ago; does that then make it better, for instance, than the EC32? Well, no, because that model too flip-flopped. It's because that's what mathematical models do! We saw extremes advertised around 48hrs ago; real "epic" predictions. But what we have seen since, is a moderation - a "downgrade" - in those predictions, both in intensity of temperature and associated longevity. Yes, it will still be cold - that is obvious - but not as cold, and not for as long. So, it is reasonable to state that the outlook - in the mid-term - has been proportionally downgraded. We mustn't dwell on the intra-runs too much though - it IS all about trends. I would advise taking a wider, more encompassing view of the outputs, and base your thoughts - and indeed your expectations - on those. I tend to operate my analysis on a rolling 48-72hr window, which allows me to iron out any of the obvious peaks and troughs, or "outliers" as we know them.

Can a gentlewoman help? Here ya go (NAO): http://www.cpc.ncep....index_ensm.html Edit: And the (AO) is there: http://www.cpc.ncep.noaa.gov/products/precip/CWlink/daily_ao_index/ao_index_ensm.shtml

I see it as a bit macro ~ micro. I'm not so much bothered about the variance in short-range, because - to me anyway - it's so predictable that it doesn't really hold much interest for me. What holds greater sway and interest for me, are background signals and teleconnections because it is those which set conditions to be conducive or not; cause and effect. As we enter January, it would appear that we're looking to be in an excellent position, teleconnectively. That's where we want to be. Of course nothing is certain, but - to coin an analogy - if you want to build a house, then you first need to lay some foundations, and then have your tradesman on-hand. My understanding of Stewart's posting, is that the teleconnections are proposing strong foundations locking into place. We all know that that doesn't guarantee anything. But it does work in our favour. Winter is 3mths, and we've only just entered the beginning of it, so there's plenty of excitement to come I'm sure. I'm extremely excited about this consistent signal which is appearing for the strong build of the Scandi high, and also the Siberian one. Shortwaves, as we know, can cause endless frustration, but re-amplification of background conditions look set to be 'the norm'. Very interesting to see where we are in 10 days time; whether this Easterly will develop as we all so wish, or whether we'll have to wait a little bit further. The SR/MR models have been broadly consistent across 72hrs, in showing that this is a strong signal and the LR models are also resonating a similar, if not always identical, theme. So, we have some agreement here. Remember, it isn't always about chasing the runs, one after the other; however exciting that can be - if that's your thing - I don't think it'll tell you a great deal about 'confidence', specifically whether throwing a party or shredding your nerves are justified. I just think we learn more through assessing the wider picture, over a number of days and a multitude of runs. I think we can all agree though that - for once - we're starting a winter right on the front foot in terms of potential, and that's got to raise the excitement levels all around. Let's hope we maintain this momentum though the season, and - with any luck - we'll have some amazing stories and pictures to share at the end of it

Coincides with the next movement of the 60-day tropical wave cycle. Stewart touched on the effect of the timing of this, bearing relevance to whether January would be just below average, or well below average. The inference I'm inclined to take - mindful of the context - is that the timing would be conducive to the latter. Or, he might just be talking about post-Christmas blues

It's because all algorithms have an in-built mathematical bias. Some models will exhibit noticeable tendencies, and it's because those eventual outcomes have been placed there, by design. The 'design' is based on long-term probability, which - as we know - is a reliable reference. It's all part of the rules and thresholds which have to be within models, in order to control the obvious random nature of chaos theory. In short, if you had no general principles - say for the Northern Hemisphere as a region - then compounded chaos theory could, quite plausibly, imagine the earth spinning on an entirely different axis! So with this in-mind, I feel it's often wise to resist placing so much reliance on the low-res output of, in particular, the GFS. We have seen, indeed we all recognise, it's propensity to 'blow up' low pressure systems - therefore it's reasonable to infer that there is inbuilt design for this. Hope that helps to give you an understanding of why you see this patterns/trends/favourable evolutions in some models, especially in low-res?

Well, yes and no. Yes it's cold as in it's hardly tropical outside, but to be less obtuse, it's the memorable, "event" cold which I think most model observers - in the winter - both look for, and get excited about. I think it's also fair comment to state that most like cold to be interspersed with periods of snow - disruptive if possible, and there's not a great deal of chance in the current synoptic. So yes, if you look for no colder than the current, fairly mundane fair - then excellent, and you'll be in your element. But for those who seek excitement in the extremes, then the current conditions don't quite ring that bell. Each very much to their own

To a degree - and I accept it takes a leap - there is a flavour of consistency in the fact that the output isn't consistent. You can imply from that, that there is significant disagreement owing to a determining factor. We all know that factor rests on the exit of energy off the Eastern seaboard, and specifically its transit thereafter; different models favouring different evolutions, which then propagate further divergence throughout the rest of the run. So being consistent isn't always about having a clean evolution or transition to an agreed phase. The over-arching consistency which we've all witnessed in the operationals is one of - in some instances - black/white disagreement. Some like it to be more absolute than that, but it's not to say that you cannot extract information from variable outputs - it's just a bit more subtle, intangible but yes - accepted - also tenuous. I have to say, reading this mornings' musings on here didn't make for fun; statements like "...may this be a lesson to you..." are just so unnecessary, not to mention premature. At the end of the day, everyone is trying to learn - and everyone is at different stages of building their knowledge up - so, unless you know it all - and I've yet to read anyone on here who does - then statements like that are disappointing to read. Very interesting to see Stewart's observations, and I implore you all to view the outputs to come with those underlying thoughts in-mind. Also, quite sagacious of John to refrain from committing to a forecast until such time as more confidence is available, lending to a stronger foundation. These are your pointmen in complex situations like this.

I think some of you have forgotten - or indeed you may not know - but the BOM model is built on the UKMO base platform, ie: it's essentially the same model, with slight variation to account for SH and, specifically, the Australasian region. The UKMO sells its forecasting services to foreign meteorology agencies - one of its main revenue generators. So, if you want to extrapolate forward beyond the 144h range of the NH UKMO, take a view from the BOM. However, caveat that with the understanding that each model does have regional tweaking.

In a nutshell, bang on target Tom. And yes, once cold is established, there is a virtuous cycle which can prolong it and often you need either a lot of energy or a high latitude change to displace it. Good post Tom, you're picked things up quicky.

Well, we all take different and varying degrees of interest from the models, meteorology and mathematics. My interest is really focused on using regression analysis to determine confidence in a signal. Put simply, it's about interpreting the influence of feedback within a cycle. My analysis, and therein doubts, around the ECM 32 isn't the fact that it changes - you're stating the obvious there - it's the degree and pace of change which is significant, hence why both John and myself are asking the right question, ie: has this, what may appear significant, paternal change been a recent one, and how established is it as a new trend? That's not "trying to sound clever" it's about using intelligence to determine confidence within a mid-term forecast. Strong divergence - which is what the ECM 32 relative to its previous solution implies - suggests low confidence in the model. I'm trying to be as educational as possible, because I think it'll save quite a few people from sleepless nights and early-morning headaches. Models do, and will invariably, chop and change - but instead of reacting to every single peak and trough, it's maybe worth challenging objectively whether there is enough confidence in a solution which offers a widely different proposal. You are new on this forum, so won't have experienced the swings and roundabouts of the winter season, or the educational, scientific interest which Stewart (GP) and I take. I don't think either of us are so much interested about 'the devil in the detail' which gets churned out, but more the over-arching signals which - ultimately - always govern what then subsequently filters down into low-level output. Focus on those, and the rest takes care of itself. My personal view - which I've held for 10 days or so - is/was, that cold is absolutely guaranteed to be with us in the next 7-10 days. That view hasn't been influenced or altered by low-level outputs, but more the strong signals which were present some 30 days ago. Those signals have remained consistent, moreoever increasing with confidence and strength. As we enter the more reliable timeframe, you're going to be exceptional outputs from the intra-day models - and we've all seen that over the last 48hrs. Of course there will be bumps - but, as I'm trying to stress to you, good analysis is about filtering the outliers out from the prevailing trend and - whichever way you wish to cut this cake - this cake is very much cold, and with exceptional potential for a lengthy period of it too. Again, some solutions will suggest longer/shorter - but, again, it's about noticing the underlying trend. This mornings' models do not surprise me in the least - as they are on-cue with the underlying trend, and serve to highlight outlier solutions which popped up 24hrs ago. Those are the ones which can be easily dismissed and no sleep lost over. Expect further 'upgrading' as we transit through the next 48hrs and the pattern falls further and further within the reliable timeframe. I hope you've found that helpful, and happy model watching

Hmm. To be honest, to see such a swing in what is widely regarded as a respectable MT model, is alarming. I don't think that's positive at all, more that it implies it to be unreliable. Whatever way you interpret it, that's a huge divergence in probabilistic determination. Don't get me wrong, I'm as keen as anyone to 'get the cold' and see it in the models - but, as a mathematician and scholar of algorithms, such variance worries me. John asks the right question: is this switch sudden, subtle change or is it flip-flopping? There's more to be learnt about the probability of model forecasts by how they react to changing variables over time, not-so-much the intra-runs. I think sometimes it's really worth taking a step back and, instead of jumping all over the latest output, assess with an objective mindset whether what you're seeing gives you confidence, or a lack of? For me, flip-flopping - like this - predicates poor confidence in the underlying deterministic calculations. I recall writing more at length on a similar theme last winter, re: short, mid and long range modelling. Not for wanting to regurgitate, but the ECM 32 is not filling me with much confidence at all.

Quite right. The consistent scatter in low-res should really be enough of a hint to put people off taking the output with any degree of credibility. Mathematically, the variation is laughable. Over the last week or so, we've seen particularly large variance in the long-range, low-res output. To me, as a mathematician, that signifies that there is significant margins of error involved, either in the start data - unlikely - or the final prognosis, more likely. Therein, I would not be so bold as to make such confident claims either way, re conditions in the 10-14 day window. The background signals are more important, and avoiding the intra-day variance, you can see the gentle evolution towards what Stewart has been alluding to. For me, that's a lot more important, than seeing pretty little pictures in "FI" which constantly change, not only by day, but each 6hr'ly run. I don't think their low-res output adds or brings a single thing to the table. What we need is broad consistency across a range of wider signals, and Stewart fantastically describes the favourability in terms of these indices in his video. I've yet to see any contribution to this forum as thoroughly informative and genuinely educating as that one single video, and I'd implore most of you take the time to watch it.

Excellent presentation Stewart

In short, not even worth consideration. Think about it: even the most reliable weather model is just about reaching subtle confidence with forecasts out to days 7-10. That in itself represents a massive step forward. There is significant, massive variance in the monthly 30-dayer ECMWF forecast - so much so that I rarely give that much credence either, so why anyone would even begin to take a 3mth forecast, 3mths in advance - it's laughable. Yes we have patterns, but patterns change - and it's the rate of change/decay which can often be crucial. Put simply, there is absolutely no model on the planet which can accurately forecast out to this range these subtle nuances. I'm sure that - in time (and it will take a VERY long time) confidence will extend to such ambitious ranges, but - until then - I would treat these 'forecasts' with an absolute truck load of salt. And I have neither a preference for either cold nor mild, by the way - I just know mathematical nonsense when I see it; and believe me, 3mth 'forecasts' are the work of Comedy Central.

The UK Met Office is currently presenting its business case for a £42m upgrade to its supercomputer in 2015.

I think it's important - at this point - to ensure that the term 'risk' is not ignored; while the Met Office may have released 'exciting' outlooks - it's just a risk of it, ie: it's NOT nailed-on. I think it's an important point to make, as it sounds a 50/50 call to me. Personally, I wouldn't wanna be hedging my bets on a 50/50 shot, because - rightly so - the Met Office have also outlined what the quite boring and tiresome - yet also plausible - outcome could be. Factors conducive to the colder scenario though are favourable forecasted indices (AO & NAO) which lends credence to establishment of the blocking pattern. As we know though, NOTHING is certain in this game. My interest will rest on how muh variance occurs in the ECMWF runs over the next 48hrs or so, as I think this is going to be absolutely crucial. I don't take any notice of the GFS as I believe it to be a mickey mouse model. I believe ECMWF will track the patterning better, and you'd have to infer that the Met Office's view has been influenced by the 32 ECM output. Therein, how that output filters down to the 'normal' output will be very interesting to view. But I think it could go either way. Anyone trying to suggest they know - at this stage - is fooling only themselves. It would be nice to have a first real blast o the season though...

Nick, I always find your NOAA extracts interesting viewing; what is their view today, do you know? I know they were largely dismissive of the GFS Op's modelling of the energy exiting, favouring the ECMWF's interpretation. I wonder if this has changed at all, mindful of ECMWF's slight tempering of its own output?

I think you may be misinterpreting the root cause of the frustration; it's not necessarily what the models are showing - it's how certain members on here are, for whatever reason, giving intentionally misleading interpretations of them. That is can be seriously annoying, because the discussion then becomes needlessly sidetracked, as their spurious nonsense needs to be corrected. It's only one or two who are doing it, but that's all it takes. Taking a holistic view of the model thread - ignoring those outlined above - I can't see where there is this perceived rage at the models? It's commonly accepted that there is a lot of disagreement in the models, so I don't think it comes as that unexpected when, for example, GFS goes one way and ECMWF goes another? That's just a continuation of the theme of the last 48hrs, ie: nothing new at all. End of the day, none of us know for certain what is going to happen - and I think the fact that multi-million pound computer models can't even work it out, is evidence of that. I see this current situation as a mathematical crossroads; one path leads to cold, the other to fairly mild/average - and, to me anyway, it seems a 50/50 call (going by model output of late) Therein, I cannot fathom how anyone could speak with such authority one way or t'other, over the outcome of a 50/50 call? Seems laughable. Of course, as we enter into higher resolution timeframes, the parameters which will influence that 50/50 call are going to become more fixed. It is then, and only thereafter, that I feel any firmer views of 'the longer term output' can be voiced with any credibility; otherwise it's just a guess, based on a guess, based on a hunch which is probably based on bias. Hardly scientific.

As you said: there are trade-offs. For some reason, the Americans like to run the GFS four times a day. That's more often than any other model of its ilk. I don't see how, running a large model that often, can have the depth of data fed into it - because, put simply, it would take too long otherwise. I wouldn't be surprised if many of the variables entered are merely averages, which - by definition - introduces a large error skew, even before the model has even run. Why do they run it so often? No idea: possibly the whole 'everything is bigger in USA' mentality. Or, it could be that their forecasters simply like to have a larger spread of possible outcomes - that's another option. The problem with that though, is that there is going to be a lot of redundancy and, at the end of it, large portions will require human interpretation. To a degree, the idea of using mathematical modelling is to reduce the need for so much human re-analysis (it's not efficient enough otherwise) As we know, ECMWF is run twice a day and its verifications stats consistently affirm its credibility. That is irrefutable fact. Conversely, on the same verification stats, the GFS Op consistently underperforms. With that being so, I have absolutely no idea why NCEP still use it; perhaps some parts of the model are greater than the others - that's plausible too, especially if 'the good bits' are material enough to justify running with it.

I wouldn't be so quick to write anything off; ECMWF High-res det is rather interesting wouldn't you all say...

Yep, it definitely does this - was very evident the last couple of winters, especially the last. That's mainly why I tend to completely ignore it; I'll view its output, but I give it no credibility whatsoever. My MRF tool of choice is, and always will be, the ECMWF. For me, that model consistently verifies to a degree which deserves respect. I think it has shown that it has a better handle on the 'local' atmospheric dynamics far better than GFS does.That's not to say that it's always right - no model is. But form is a long term average, and the ECMWF outperforms every other model; hedge your bets on it, and you won't be too wrong, too often.