Jump to content
  • entries
  • comments
  • views

(Preliminary) Winter Forecast 2014/15 - Return to Cold?



[size=3]Before I get started on this I must at least explain last winter's forecasting failure on my part – I did indeed get the mild, wet start to winter right but the second half was so epically wrong that it more than offsets the relative successes of my previous few attempts. In my defence, the analogue charts weren’t that far off for late winter and in fact the cold trough over the US was correctly placed, but on this side of the Atlantic heights were never quite high enough west of Scandinavia to divert an incredibly strong jet far enough south for cold air to back west, though it was cold enough for over 9m of snow to accumulate on the top of Cairngorm at one point. Anyway, a few lessons from last winter:[/size]
[size=3]1) [/size][size=3]Sometimes strong wave 1 activity (as predicted by many before winter kicked off) can’t stop the Atlantic winning out, particularly when all the wave 1 activity did was to compress the vortex into NE Canada creating a turbo charged jet[/size]
[size=3]2) [/size][size=3]The OPI (read on to find out more about what this is) is good, and its creators did well at pointing to the overall hemispheric pattern pretty well, but the jury is still out on its ability to predict the AO – last winter came in neutral in spite of having the second highest OPI value in the series. It may well be that this was a one off, and given its remarkable hindcast correlation I am willing to give them the benefit of the doubt that this was a particular exception, but this winter will be a key tester for the OPI[/size]
[size=3]3) [/size][size=3]Analogue charts can be deceptive – trying to work out how our little corner of NW Europe benefits from low heights generally to the south and higher heights to the NE is a difficult task. January’s height anomaly was superficially similar to that of January 2013 but the outcome was completely different for most of western Europe.[/size]
[size=3]So, with those things in mind, what does 2014/15 have in store for us? I’m going to go through the factors I’ve taken into consideration (this may end up being even more lengthy than last year’s) and finally arrive at a forecast for the winter ahead broken down by month, so if you don’t fancy reading through the technical bits then scroll down to the bottom and get the ‘will it snow in my backyard on Christmas Day’ rundown (top tip – I don’t know the answer to that question and nor does anyone else, but this should at least give you a rough idea of what to expect).[/size]
While last two winters have seen generally neutral conditions prevailing over the ENSO region, the expectation for this winter based on current conditions and the modelling consensus is that we’re likely to see a weak El Nino i.e. 0.5-1, developing over the course of the winter:
[attachment=228579:El Nino forecast.gif]
In addition, the Quasi Biennial Oscillation, which fluctuates roughly every two years between positive(westerly) and negative(easterly) is strongly negative:
What’s the significance of this to our winter? Well the current research suggests that the stratospheric vortex over the winter months is generally warmer during El Nino phases than La Nina, in particular when combined with a negative/Easterly QBO, and that Sudden Stratospheric Warmings (I’ll get into that a bit more later) are more likely during higher amplitude ENSO events than when ENSO is neutral, which means that the likelihood of cold air being displaced from the Arctic to mid latitudes is greater. To illustrate, and this will directly feed into my overall forecast, I’ve put together a list of analogue years for the expected –QBO and weak to moderate (0.5-1.5 MEI) El Nino winters, along with 1976/77, which was very marginally under the El Nino threshold for the DJF period. This is the height anomaly:
[attachment=228581:ENSO QBO anomaly.png]
As you can see, without even taking into account the more ‘experimental’ or recently developed indices, the signal for this winter is for strong blocking over Greenland and the Arctic with low heights across Europe, a classic –AO/-NAO pattern. Of the 7 winters, all 7 had overall negative Arctic Oscillations and 5 had negative NAOs, with the mean temperature anomaly for Scotland 1C below average and 16 of the 21 months were below the 61-90 average.
[size=5][b]October Pattern Index[/b][/size]
Last winter the October Pattern Index had just been developed and, while I used it in my forecast, it was still very much an experimental index. It remains in an experimental stage, with a paper on it due to be published after this winter, but at least we have a reasonable understanding of how it’s calculated. Last winter’s results were mixed – its creators nailed the specifics across the pond of an Alaskan Ridge and the vortex dropping into Eastern Canada but as a result of this the value of the AO was significantly lower than the OPI, although still positive overall. However, as a general measure of the difficulty of shifting the vortex it worked out well – the jet just wouldn’t let up, even when heights started to build to the northeast in January as anticipated, and given the significance of the correlation between the OPI and the AO in hindcast it would seem foolish to ignore this completely. Fortunately, it doesn’t particularly disagree with the prevailing consensus. If you want to read more on the OPI I’ve attached links to both the thread on it, with the original explanation along with some fine analysis on it, but essentially it views October as the critical month in vortex development and extrapolates that a strong, compact vortex in October leads to a strong, compact vortex, and +AO, over winter, and the same for a weak vortex. The OPI for this winter is [b]-2.08[/b], the second lowest value in a series going back to 1976/77. The lowest value ever? -3.4 in October 2009, just before the coldest winter since 78/79 in England, and the coldest since 1962/63 in Scotland, so as you’d expect very negative values correlate pretty well with the most brutal winters. Taking winters with OPI values below -1.5, and to extend the timeseries a bit I’ve also taken into account winters with an AO below -1.5, which essentially assumes that the AO/OPI correlation holds going back further than 76/77, gives this result:
[attachment=228580:OPI analogues.png]
Given the winters we’re looking at this isn’t a surprise but it is excitingly, or if you hate cold worryingly, similar to the ENSO/QBO analogue chart. The average AO for these winters is -1.8, with the highest AO value at -0.9 in 86/87, while 9 of the 10 featured negative NAOs, with 5 of the 6 lowest NAO values since 1951 being found in the list of winters.
It should also be noted, however, that [font=calibri][size=2]the Italians’ even more experimental new toy, the Indice de Zonalita Emisferico, disagrees somewhat with the 'consensus' so far, going for a very negative AO for December but positive January into February and coming out neutral overall. [/size][/font]
[size=5][b]Snow Advance Index[/b][/size]
Another tool, devised by Judah Cohen, is the Snow Advance Index (SAI) which is a measure of the increase in snow extent from the 1[sup]st[/sup] to the 31[sup]st[/sup] October. This has also been shown to correlate strongly with the AO, with positive SAIs generally leading to –AO winters and vice versa. This October, it appears we’ve had a very substantial gain in snow cover over the last month - October finished with a snow extent similar to 2009/10 in spite of a big fall in the last few days, with the SAI value up there with the largest values of Octobers 1976 and 2009.

A brief overview of these – in spite of being in what appears to be a long term –ve Pacific Decadal Oscillation phase, the PDO this winter will be positive, which for the US means high pressure around the Pacific NW with colder conditions for the east and southeast with the jet tracking further south than normal. This, combined with solar activity being still reasonably high (although far lower than at this stage in most cycles), leads to the likelihood of a very strong jet once again this winter. However, a +PDO is not all bad – the last consistently +PDO winter was 2009/10, where the jet was displaced far to the south of the UK by persistent height anomalies over Greenland:
[attachment=228589:09-10 pattern.gif]
As for the Atlantic, we’re once again looking at a pattern of anomalous warmth for high and low latitudes with colder temperatures in the mid Atlantic, which may aid a -ve NAO, although it doesn't look like much of a factor either way.
SSTs are taken into consideration, with analogues (using an incredibly sophisticated system of looking at others' analogues for SSTs and having a look to see if it's similar to our current pattern) used to weight the sample to generate our composite charts.

[size=5][b]Solar Activity[/b][/size]
After a monster sun spot flared up last month, causing a spike which went against the general downward trend as we head towards the next solar minimum, we're back down to a more normal solar flux of 120 and sunspot number of 71 at the time of writing:
This means we're straddling the borderline between high and low sunspot activity. Generally speaking, colder winters are associated with solar minima, and in particular the chances of a Sudden Stratospheric Warming are greater with low sunspot activity combined with a -QBO (or high activity combined with a +QBO), so if you're looking for cold you'll be looking for solar activity to continue to fall over the next few months. This is also factored into my analogues, with analogue years with sunspot numbers between 110 and 40 being weighted more strongly.
For those looking for a detailed explanation of why the stratosphere is so important to winter weather, look no further than the strat thread:
[url="https://forum.netweather.tv/topic/81567-stratosphere-temperature-watch-20142015/"]https://forum.netweather.tv/topic/81567-stratosphere-temperature-watch-20142015/[/url] I certainly don't claim to have as much expertise in this area as the likes of Ed, so much of this is based on rather crude statistical analyses using the FU Berlin strat site (and probably doesn't take enough account of early strat warmings earlier than 2009/10 which aren't listed as Canadian Warmings).
From my list of ‘strong’ December composites i.e. with 2 or more variables matching the expected pattern, 7 out of 12 winters, or 14 out of 22 winters when taking into account those that were double or triple weighted, featured either a Canadian Warming or bottom up wave 2 wavebreaking over Greenland (2009/10). This is the key to early winter – winters featuring this warming saw an average December temperature for Scotland of 1.5C (1.4C below average) and a CET of 3.7C(0.9C below average), but without it saw mean temperatures for Scotland of 3.1C and a CET of 5.2C. While in either case the chances of a below average January and February remain fairly high, it does has a major impact – analogue winters without a CW saw mean temperatures of 2.4C for Scotland for the winter as a whole, as opposed to 1.1C with. Basically, the November/early December strat, despite the favourable early signals, really does need to play ball and help us out with some lower-mid strat vortex disruption early doors. Without it, there still exists the potential for a very cold January and cold February (the sample size of winters with such favourable conditions are very low, and these tend to manifest later on in the winter), but our chances of getting a below average winter, particularly in Scotland, are vastly better if winter gets off to a decent start.
So how does this winter look? Well, rather interesting to say the least. The vortex has struggled to get particularly organised at any level, and it has generally been found towards the Eurasian rather than the Canadian side of the pole (very much unlike what we saw last winter). Moreover, there are already signs of vortex disruption, both from wave 1 activity propagating downwards stretching and compacting the vortex and also, potentially, from wave 2 activity from the troposphere. Today's ECM ensemble mean for D10 highlights this well [attachment=228593:EDH101-240.gif] This, combined with the other factors (which to be fair the analogue years already factor in, suggests that, to date, we're more likely to be in the first category of winters, one with a disrupted November vortex, than the second, where the vortex strengthens through early winter.
As for later on in the winter, the analogues, and most of the research papers on the strat forcings I've read, suggest an SSW mid winter is more likely than last winter, and if not formally an SSW at least something similar to it.

[size=5][b]Long range models[/b][/size]
I will mention these as this is yet another interesting test of how ‘good’ they are. However, I would very much advise anyone who takes these too literally to have a read of this: [url="http://www.esrl.noaa.gov/psd/events/2014/arctic-predictions-science/presentations/tue/arctic-wkshp-051314-cohen.pdf"]http://www.esrl.noaa.gov/psd/events/2014/arctic-predictions-science/presentations/tue/arctic-wkshp-051314-cohen.pdf[/url]
A compelling case for why the long range models are particularly bad at forecasting winters. Essentially, the (correct) modelling of warmer Arctic temperatures, particularly around the Taymyr Peninsula (expect to read a lot more about that in the coming months), is modelled as bringing milder weather to mid latitudes, when in reality there’s no mechanism for this and in fact what happens is colder air spills out into the Eurasian continent and across into Europe, with the same thing happening over the other side of the pole. They also appear to struggle with the stratospheric modelling, with some of the models not even extending as far up as the top of the stratosphere. These factors appear to have been responsible for the general misses by the long range models of 2009/10 and December 2010, so it's not clear that, if we are heading towards a cold winter this time, the models would be able to pick up on it.
Anyway, the signal from this is somewhat mixed, although more models are going for a milder winter than a cold one, with a generally +ve NAO and AO signalled by most, going against most of the factors listed above. I'll link to the two most commonly cited long range model based forecasts, the Met Office long range outlook the CFS [url="http://www.cpc.ncep.noaa.gov/products/CFSv2/CFSv2_body.html"]http://www.cpc.ncep.noaa.gov/products/CFSv2/CFSv2_body.html[/url] for reference.

[b][size=5]Background Factors Summary[/size][/b]
[size=5][size=4]Before we get onto the actual forecast I'm just going to sum up the signals for the coming winter:[/size][/size][list]
[*]-ve QBO and El Nino points towards a more blocked winter than average for the Northern Hemisphere and hence greater chance of a colder than average winter
[*]very -ve OPI points towards strongly negative AO and also a more blocked winter than normal, though this is still in its infancy
[*]very +ve SAI again points towards a negative AO and more blocked winter than usual
[*]SSTs generally point to a more active jet than normal, but also somewhat towards increased high latitude blocking
[*]So far the stratospheric, and tropospheric, vortex has struggled to become established and this looks likely to continue through November, potentially leading to a cold end of November/start of December
[*]Long range models still don't really buy any of the above, going for a milder than average winter with low pressure to the northwest, similar to last winter
[b]Composite chart methodology[/b]
Using the criteria listed above I've compiled a list of composite winters. However, given the apparent divergence between December and the rest of winter the factors assessed for the December composites are somewhat different. As with last winter I've decided to use a broader composite matching than most, meaning that some months are double or even triple counted if they are very similar. For December, the composite months are as follows:
1958(x2), 1965, 1968 (x2), 1969, 1976(x3), 1979, 1984, 1986, 1993(x3), 2002(x2),2009 (x3), 2012 (x2)
For these months, the mean AO is -1, the mean OPI (for those months where an OPI value is available) is -1.6, the mean winter Nino3.4 value is 0.6 with the mean QBO at -10.1.
For January/February, the composite months are as follows:
2013, 2010 (x3), 1994, 1987(x2), 1985(x2), 1977(x3), 1970, 1969 (x3), 1966, 1959, 1952
For these winters, the mean OPI is -1.9, the mean winter AO is -1.7, the mean Nino3.4 value is -0.5 and the mean QBO is -10.1.

[b][size=5]A quick look at November[/size][/b]
I've already given a brief overview of how November is shaping up for the vortex in the strat section, but a slightly more detailed look is perhaps called for given that, in Scotland at least, the second half of November has the potential to behave very much like a winter month, and also because, if the OPI/SAI is right about this winter, November should look similar to other -ve SAI/-ve OPI Novembers.
So what are we looking for? The Cohen high SAI SLP anomaly for November doesn't seem to be publicly available, but the characteristics of it, by the account of those who have access to it, is very similar to this, the height chart for the six highest SAI values that I crudely calculated using the Rutgers monthly data: [attachment=228619:sai.png] Troughing to the northwest stretching to the UK with mean heights over Scandinavia/Western Russia (driving Wave 1 activity in the strat). The height anomaly for the 7 sub -1.5 OPI Novembers looks similar[attachment=228621:opi nov.png] What about this year? Well, this is the current 8-14 day height anomaly from the NOAA [img]http://www.cpc.ncep.noaa.gov/products/predictions/814day/814day.03.gif[/img]

Once again, reassuringly similar to the analogues above. The pattern itself isn't, immediately at least, a particularly interesting one for the British Isles - generally wet with alternating cooler and somewhat milder periods when heights manage to extend a bit further west or cool and dry when low pressure slackens off. But there is major potential down the line - the omnipresence of the scandi high combined with a vortex which, unlike last year, seems determined to stay as far from Greenland as it possibly can, leads to the possibility that by month's end we're likely to see an attempt at an easterly. While November 2010 was a ridiculously extreme example, I still think that there's a chance for us to see some widespread snowfall before the start of winter proper, and at the very least before Christmas. Overall though, I'd expect temperatures to be close to average given the mild start, with a CET around 6-7C and a Scotland temperature average around 5C, with precipitation well above average.
[size=6][b]Winter Forecast 2014/15[/b][/size]
[size=3][size=4]Caveats: as usual, the more specific I am the lower confidence it has. [/size][/size]
[size=4]My analogue charts for the month are as follows. Heights:[/size]
[attachment=228638:dec comp.png]
Higher heights centred around Novaya Zemlya/Svalbard stretching across to Greenland and to the pole, giving at least a fighting chance for colder air to come at us from the east. Low heights around the UK suggests a very unsettled period, although whether that results in a snowy month or a very wet and windy one, or one where both feature in roughly equal measures, is hard to say.
Temperature: [attachment=228639:Dec temp.png]
You'll notice that, I think largely as a result of such a large sample size, the values aren't particularly large, but that the signal is generally for cold around our latitude, with the height of the cold sitting around Scandinavia/western Russia stretching across to Scotland, but with slightly above average temperatures for southern Europe.This is consistent with a cool, unsettled theme, with troughing too far north to deliver to southern Europe but potentially far enough south to deliver for our neck of the woods.
Precipitation: [attachment=228640:dec precip.png]
Again, consistent with higher heights over Scandi, troughing further south, lots of precipitation for the south/southwest of England.
[size=4]I'm expecting a month which is largely unsettled but, potentially, very cold, if we continue to see major vortex disruption in the stratosphere over the course of November. The month may well start with a cold spell from the east but I'm expecting most of early-mid month to be dominated by low pressure systems bringing a lot of wind and rain. Depending on how they're orientated and how far west high pressure can make it, we could well see some transient snowfalls, particularly further north and east, but the general pattern is for above average precipitation and average to somewhat below average temperatures, with occasional southerly winds bringing in the odd milder spell (and drier for the south of England). However, from mid month onwards pressure across the Greenland-Iceland-Scandinavia corridor is likely to rise, bringing in colder conditions. This transition is likely to take the form of a brief settled spell, with high pressure ridging up from the south to the Scandi high and eventually retrogressing to Greenland with a scandi truogh, to bring in a significant cold spell from the east and eventually north-northeast to end the month. During this period, snowfall would be expected to be above average, with any drier periods bringing some deep frosts, and if the timing is right we could well end up with a widespread white christmas. HOWEVER, and this is a big caveat, this outcome is dependent on things going the right way over the course of November - if the vortex does manage to recover by the end of this month a more likely outcome is an unsettled month without any prolonged deep cold, although with the potential for snowfalls from the northwest to Scotland or more widely further south when low pressure dives into the continent and with temperatures generally on the cool side. I'd put it at about 60/40 in favour of the colder variant at the moment(although of course it isn't in reality a binary choice), but I'll update towards the end of the month to let you know which way it's heading.[/size]
[size=4]In the colder variant we're looking at Scottish mean temperatures around 1.5C and a CET around 3C, with precipitation above average, moreso the further south you are, and perhaps below average for the northwest of Scotland, and a good amount of snow about. If the vortex isn't significantly weakened we're more likely to see a Scottish mean temperature of 2.5C, much closer to the average, with a CET 4-4.5C, with precipitation well above average everywhere and snowfall near average for most, but with the wind a major feature. Either way, not a very mild month with the prospect of at least some snow for most.[/size]

Strangely, I think January is likely to pan out colder than average no matter how the early part of winter goes. If it's the colder setup in December, we'll simply see a continuation, perhaps even a deepening, of the cold, with the jet undercutting to give snowfall potential. If not, the analogue years would hint at a fairly abrupt switch to cold via a Sudden Stratospheric Warming.
Composite heights: [attachment=228641:January 15 height anom.png] Stonking positive height anomaly over southern Greenland, modest low height anomaly stretching from the eastern seaboard, strongest in NE France, -ve AO, with vortex remnants over central Russia - for snow lovers this is about as perfect a chart as you could wish for for January, pointing to a month dominated by winds somewhere from North to southeast.
Temperatures: [attachment=228642:January 2015 brrrrr.png] Bloody cold - a much more robust negative anomaly, once again centred around the Baltic extending right across Europe, and a clear >1C below average.
Precipitation: [attachment=228643:Jan precip.png] Below average across the Northwest Highlands and more generally for western Britain, closer to average elsewhere.
Stratosphere:[attachment=228644:strat temp jan 15.png] temperatures much above average, any sign of a stratospheric vortex is over towards the Pacific.
Basically a colder than average month, with an increased risk of snowfall. Biggest potential probably around mid month, but frankly difficult to pick a less cold period, other than there is obviously likely to be one at some point. 1/2 of the analogue months give sub 1C Scotland mean temperatures, and only 2 are above average - 1969 and 2013, and for the CET zone 60% came in sub 3C (a modern benchmark for a particularly cold month), with this time Januaries 1994 and 1969 coming in as the only above average months. In 1969 we saw for long periods what I think the main danger for January/February not delivering major cold is - a west based -ve NAO, where the pattern is too far west, resulting in the UK getting stuck in a mild, wet oasis among a sea of mid latitude cold. Of course in both winters this frustrating period was made up for with severe cold a bit further down the line, with February 1969 an exceptionally cold month, moreso in Scotland where proper Arctic northerlies swept the country, and March 2013 being colder than any of a not too mild winter 2012/13 and seeing all precipitation in Edinburgh from about the 5th March through to the start of April falling as snow. My CET punt would be 2C, although with the potential for it to end up quite a bit lower, particularly if we are starting from a cold point in late December, with the Scotland temperature average a spectacular 0.5-1C.
In many ways February has a very similar look about it to January - only 1, February, 1959, in the analogue list comes in above the 81-10 Scotland average and only 3 do for the CET zone. However, a look at the height anomaly gives a subtle clue as to why February may not be shaping up to be quite as cold as January:[attachment=228664:feb 2015 height anom.png]
The low height anomaly over the Atlantic has strengthened and moved further north, pointing to a more active jetstream, with the potential for mild incursions from the southwest along with possible battleground snowfalls. The temperature anomaly still suggests that below average temperatures are likely to persist but the 0C anomaly line is significantly further north than in January [attachment=228665:Feb 2015 temp.png] Precipitation for southern and central England looks to be above average once again but potentially below average again for the NW Highlands: [attachment=228666:feb precip.png]
The stratospheric temperature perhaps gives an indication of what might be causing the jet to power up - a suggestion of the vortex drifting towards Western Greenland for the first time [attachment=228667:feb strat.png]
Generally, this month looks to be characterised by a north/south split - Scotland seeing temperatures remaining well below average with snowfall inland from a variety of setups, with a mix of northerlies and battleground setups, while south of the border, and moreso for the Midlands southwards, it will feel cold but likely end up with a wet and windy month dominated by storm systems. The boundary line is fluid, and means that we could either end up with a month which trends towards milder weather away from the Highlands, or in fact the boundary could shift a bit further south and leave more of us with the cold, but this looks the most likely solution to me. Predicted temperatures will reflect this shift, with a CET around 3.5 but a Scottish mean temperature of 1-1.5C.

I won't do a forecast for March in full but all I'll say is that it looks pretty cold as well at this stage, and if you're looking for your snow fix and it hasn't panned out by February there's still potential for March to deliver something interesting.

So, in summary, a return to cold winters is expected after last winters mild 'blip'. Strong signs for a negative AO which will at the very least push the jet far enough south to make things interesting and with the potential for some really classic spells of cold and snow, particularly but not exclusively for January, and moreso for Scotland. Expect another good ski season, although one more likely to see useable snow (and probably not 9m this time) and the potential for big snowfalls almost anywhere, but at the same time there's always likely to be a fair chance of a breakdown, and the further south you go the more likely you are to end up with a lot of rain. The mean winter CET on my projections would be somewhere around 2.7-3.5C, so cold but perhaps not extremely so(somewhere between 1985 and 2008/09, most likely on a par with 95/96 or 10/11), with the Scottish mean temperature around 0.8-1.8C, which would put it somewhere between 1976/77 and 95/96, potentially in the top 10, or even the top 5, coldest winters since 1950.
I'm planning to update this by the end of the month when the early winter prospects are, hopefully, a bit clearer, maybe even a video update if I can work out how to do this adequately, but for now I hope you've enjoyed reading this.


Recommended Comments

Thanks for taking your time out to provide this in depth thought/analysis. From here on in it's now up to the weather to do it's piece. Great stuff.

Link to comment

Well presented, in-depth forecast. Really good and much easier reading than I thought, you explain things really clearly. Good luck with results and I hope you are correct!

Link to comment

Really interesting and well-explained forecast which I enjoyed reading; and of course I really hope you are right:)

Link to comment

Really enjoyed reading your well reasoned thoughts. Sounds like a very interesting winter coming up. I agree with you, that last year was a blip, and that winters are to continue colder for some time to come. Heres hoping anyway and if you are close to being correct then we will finaly see some snow! Good luck to you :)

Link to comment
  • Create New...