Insights & News

Late 2021 Risk of Disruptive Cold Weather in Europe & Eastern USA

28th September 2021

Here, MetSwift presents an in-depth look ahead to Nov-Dec 2021, which expands upon this blog entry. Please feel free to switch to that channel if you’d prefer a lighter read!

The Northern Hemisphere’s autumn is barely a third done, yet already, the thoughts of some are turning to the cold season ahead. This is because it pays to size up severe winter weather risk and plan accordingly.

This year, more than most: I’m writing this blog in the wake of sharp rises in wholesale power and gas prices. This has put some energy companies at risk of collapse, including Bulb, the 6th largest in the UK.

Customers, faced with higher prices or being ‘left in limbo’ as their suppliers fold, will hope for a kind end to the year; one that demands less artificial heating than usual (i.e. has low heating degree days; HDD).

Are November and December going to ‘play nice’, or make matters worse? Let’s see what we can learn from history.

A Natural Leader: The North Atlantic Oscillation (NAO)

Analysis here will focus on the leading mode of variability for the North Atlantic and adjacent landmasses: The NAO. It varies between positive and negative states, which in northern Europe and the eastern USA are broadly associated with an increased frequency of mild and cold spells of weather, respectively.

Britannica provides a good illustration of it here. At MetSwift, we take it to the next level, by sub-dividing it into ‘East’ (ENAO) and ‘West’ (WNAO) variants.

Finer Details… West or East?

There is some significant variability within the overall negative NAO (NAO-) pattern. To demonstrate this, the following two maps both depict a strong NAO-, yet we can see some considerable differences.

Both feature high pressure across Iceland, but in the left-hand one (‘West NAO’; WNAO), that also extends westward to Canada, while in the other (‘East NAO’; ENAO), it instead extends eastward to Scandinavia. Meanwhile, the most likely location of low pressure reaches much less far to the west in the second case.

As the next two maps show, this has big implications for the weather over the eastern USA and to a lesser extent Europe:

In the eastern USA, a very negative WNAO brings widely below normal temperatures. The risk of ice and snow is increased substantially. In the Midwest and so the south of the Great Lakes, precipitation tends to be above average, too. In winter, that means some very hefty falls of snow.

By contrast, a very negative ENAO brings abnormally high temperatures. In some areas this tends to be coupled with drier than usual weather – it can even feel a bit like spring!

In Europe, both NAO variants bring widely colder than usual conditions. Exceptions are for south-eastern Europe in a WNAO, and northern Scandinavia during an ENAO.

NAO Variety Matters More in Europe When It’s Weaker

When the NAO is less negative, however, whether it’s in a West or East layout makes a bigger difference for Europe. This is because the high pressure doesn’t reach as far to the west or east of Iceland.

While even a weakly negative ENAO can still bring unusually cold weather to a wide swathe of the continent, a negative WNAO setup sees unusually mild and wet weather become increasingly extensive as the NAO is weakened. Initially in south-western Europe, then further north and east.

These details need to be kept in mind as we explore the NAO prospects for Nov-Dec 2021.

For those wondering, a positive NAO also has West and East variants, but the resulting conditions only have subtle differences for both the regions being looked at here. Essentially, it’s wettest and windiest in north-eastern USA during a WNAO+ but northern Europe during ENAO+.

Frequency Distributions

To obtain guidance on how the NAO is likely to behave, I have filtered daily NAO observations to ones on which three key driving forces of weather patterns (teleconnections) were behaving in the same way as they’re expected to be during the upcoming months.

  • The El Niño Southern Oscillation (ENSO): Variations in sea surface temperatures (SSTs) across the tropical Pacific from the Dateline eastward. Once every 1-3 years, these become strong and coherent enough to be classified as La Niña (low SSTs) or El Niño events (high SSTs).
  • Pacific Decadal Oscillation (PDO): Another SST cycle, but occurring in the northern Pacific Ocean, with predominantly positive and negative stages that usually last a decade or two.
  • Quasi-Biennial Oscillation (QBO): A cycle of westerly and easterly winds occurring in the stratosphere above the tropics.

That filtered data has then been compared with unfiltered data, using frequency distributions. This shows us what historically suggests the WNAO and ENAO are most and least likely to be doing.

Nov WNAO: Probably Not Up to Much

In the filtered Novembers, a much higher proportion of days had a near-neutral (zero) WNAO. When it’s like that, the weather is usually nothing out of the ordinary.

That negative skew to the distribution and raised frequency of extremely negative WNAO (-3.5) days are worth noting, though. Coverage of snowfall events could be a little above average, with mild and wet winter weather less prevalent than usual, especially in north-eastern USA.

Nov ENAO: Probably Weak One Way or the Other

The filtered distribution for the ENAO is also more focused around near-neutral values than usual. Not as much, though, so we see a wider range of most likely values, from weakly negative to weakly positive.

This suggests November will be dominated by ‘average’ conditions, especially in Europe. However, there is certainly some scope for notably mild or cold spells of weather (just not in a big way).

Dec WNAO: Positively Inclined…?

Now this is a more interesting filtered distribution. A weakly to moderately positive WNAO is suggested to occur most often December 2021. That implies a wetter, windier than usual month for the north-eastern USA.

However, a very negative WNAO (-3.0 to -2.5) also has a raised frequency compared to average. There could be a few more unusually cold, snowy days than usual in the eastern USA, perhaps Europe too.

Dec ENAO: Unusually Prone to Being Very Negative

Last but far from least, we come to December’s ENAO filtered frequency distribution. In contrast to November, a lowered frequency of ‘average’ weather is suggested. Not only that, but the combined frequency of negative (-2.0) to very negative (below -3.0) ENAO days is 9.3% above normal. Plus, the most above-normal probabilities are at the exceptional end of the spectrum.

This is a strong signal for some impressively cold, snowy spells of weather to feature across Europe in December 2021. Such would occur alongside unusually mild conditions in the eastern USA.

It’s possible that this signal is connected to something called a ‘sudden stratospheric warming’ (SSW).

SSWs and the Polar Vortex: Dominant Drivers of Winter Weather

As we’re already waist-deep in words at this point, I’ll avoid going into much detail on these here. Instead, here are some summary points:

  • The polar vortex occurs in the stratosphere, which is a layer of atmosphere sat on top of the one we live in.
  • It spins up in autumn and usually hits peak strength Jan-Feb.
  • When its strong, cold weather tends to be infrequent and fleeting in the mid-latitudes (e.g. northern USA and northern Europe).
  • When its weak, the opposite is true – but this alone doesn’t produce the biggest cold spells.
  • SSWs weaken the polar vortex via either displacing it from over the Arctic or splitting it apart.
  • The USA often sees extreme cold spells when the polar vortex is displaced. Europe, only occasionally.
  • Following a split of the polar vortex, Europe tends to see a prolonged cold spell kick off within one or two weeks. This is often followed by one in the USA another week or two later.

On these maps, the blue shading shows us where the polar vortex is. In the left-hand one, it’s strong and positioned over the Arctic. In the middle one, a SSW has weakened it and displaced it toward Europe, while in the right-hand one, a strong SSW has split the vortex in two – this one leads to the most dramatic weather.

These maps depict the coldest fortnights typically seen during the impact periods following displacement (top) and split (bottom) sudden stratospheric warming events. This has been achieved by identifying the lowest 14-day mean temperature for each event, taking a mean across all events, then subtracting the 1981-2010 long-term average from it.

A SSW By December?

For Nov-Dec 2021, some of the world’s largest supercomputers have been making some eyebrow-raising polar vortex predictions.

One of the most exceptional is from a model built by the European Centre for Medium-Range Weather Forecasting (ECMWF). This is one of the best-performing on Earth, so it’s worth paying note. In all its past forecasts since 2000, only one has predicted a weaker polar vortex.

That implies a raised likelihood of a SSW occurring by December. There are a few other supercomputers making similar predictions.

Does filtered historical frequency analysis support this prediction?

Not Majorly

I began the statistical analysis looking at only the biggest, rarest SSW events. Some call these ‘major’ SSWs. The climatological frequency of these in Dec is a mere 1%, while not one has been observed in Nov.

Filtering the historical December data in the same way I did for the NAO analysis, not one of the matching months featured a major SSW. So, that’s looking even less likely than usual for Dec 2021.

To include weaker SSWs, I had to use polar vortex strength as a proxy. In the filtered Novembers, 73% saw a weak or very weak polar vortex. For December, the proportion falls to 33%.

Overall, this suggests to me that a ‘minor’, displacement type SSW could well occur at some point in Nov-Dec 2021, most likely Nov. If we consider all minor SSWs, that suggests a big December cold spell is likely in the USA, with a small chance of one in Europe. Yet the NAO analysis suggests Europe has a higher chance. What gives?

The Elusive ‘Canadian Warming’?

The majority of observed polar vortex displacements involve it being pushed toward the North Atlantic by an ‘anti-vortex’ located above the Pacific. Even if it doesn’t displace far enough to register as a SSW, this shift in position increases the frequency of cold northerlies in the eastern USA, and of wet and windy weather in northern Europe.

The much rarer form of displacement sends the polar vortex toward northern Europe. On these occasions, the ‘anti-vortex’ is above the Canadian Arctic. Hence, some refer to these as ‘Canadian Warming’ (CW) events.

Two examples of CW events, in Nov 1980 and Jan 2020. The first led to a cold spell across the eastern USA in mid-late Nov then another in most of Europe during the first half of Dec 1980. The second was followed by very mild weather across the whole of both regions! It makes a huge difference how far the polar vortex is displaced away from the Arctic.

A displacement toward Europe causes the typical path of Atlantic storm systems to shift south. The largest displacements, which qualify as SSWs, move it as far south the Mediterranean. That turns the ENAO very negative and places much of Europe in a frigid run of weather, usually driest and coldest in the north, snowiest across the central belt. Meanwhile, severe impacts in the eastern USA tend to be limited to a big storm or two affecting the East Coast. These can feature heavy rain, freezing rain, or snow.

The ‘anti-vortex’ of a CW is rarely strong enough to cause a SSW once the polar vortex hits its seasonal peak in Jan-Feb. So, nearly all CWs leading to a ENAO- with substantial cold spells in Europe have occurred in Nov or Dec.

A Chilling Prospect

The long-range forecast models mentioned earlier seem to be suggesting that Nov 2021 could see one of these very rare SSW-level CWs. In this blog, we’ve seen appreciable support from statistical analysis of the NAO and polar vortex strength in filtered historical years.

In conclusion, there is a heightened risk (not guarantee) of a big cold spell hitting Europe by Dec this year. If a minor SSW occurs, probable duration of highly anomalous weather patterns is 3-6 weeks. One small comfort is that we’re not talking about a major SSW here – although I can’t rule that one out, sadly!

Regardless of duration, such an event would see soaring demand place strained energy companies under even more pressure. Certainly, one worth making contingency plans for, just in case.


James Peacock MSc
Head Meteorologist at MetSwift

Cover Photo, taken in Dec 2017 by Kris Hirst, is licensed under CC BY

Speak to one of our experts and discover how our analytic platform can revolutionise your weather & NatCat insight. Request a meeting, a chat & demo, or a free trial. Developed via consultation & collaboration.