Category Archives: SSW

10 years later: the January 2009 SSW

January 24th, 2009. This was the ‘central date’ (defined as the day on which the 10 hPa 60N zonal-mean zonal wind reverses from westerly to easterly) of a remarkable, record-breaking major Sudden Stratospheric Warming event, and there are several reasons why this event is worth a revisit 10 years later.

The Jan 2009 event set a large number of significant records in the stratosphere – and it still holds almost all of these to this day, despite strong competition from warming events in 2016 and 2018.

The aspect of the event that I always recall is the monumental (you might say…stratospheric) deceleration of the 10 hPa 60N zonal wind, which you can see in Figure 1. In early January 2009, the stratospheric polar vortex was date-record strong, with westerly zonal-mean winds ~70 m/s. Only a few days later, they were all-time record-weak, with zonal-mean easterlies of ~30 m/s. Only the final stratospheric warming (FSW) event of March 2016 comes close to rivalling 2009’s easterlies – the event is in the clear in terms of SSWs that are ‘major mid-winter warmings’. The mean deceleration rate between the peak (Jan 8th) and trough (Jan 28th) was an astonishing 10 m/s/day!


Figure 1: 10 hPa 60N zonal-mean zonal winds for 2008-09 from MERRA-2 (via NASA Ozonewatch).

Associated with this rapid deceleration of the vortex was a rapid warming. This is the most wonderful example of why we call them “sudden warmings”. On January 12th, the mean 60-90N (polar-cap) 10 hPa temperature was 202K. On January 23rd, it was 253K – a rise of 51K in 11 days! The peak of 253K remains a satellite-era record for this region of the stratosphere, as you can see in Figure 2.


Figure 2: 10 hPa 60-90N average temperatures for 2008-09, according to MERRA-2 data.

Of course, to produce such a huge warming, you need a massive heat flux (which also indicates huge amounts of wave activity propagating into the stratosphere, through the Eliassen-Palm relation). The 45-75N heat flux at 10 hPa was another metric which hit outrageously high values that have never been matched. The peak of 564 K m/s on Jan 19th is 6.5 times larger than the daily-mean climatology (and 2.5 times larger than the 90th percentile!), which is shown in Figure 3. 


Figure 3: 45-75N 10 hPa eddy heat flux ([v*T*]) for 2008-09, according to MERRA-2 data.

This all came together to produce a textbook wave-2 vortex split. Figure 4 shows this fantastic evolution (thanks to Patrick Martineau for the excellent graphics) – you can really see the cold, strong vortex that existed beforehand, and the strong heat flux from the Atlantic sector. I think you could call this “catastrophic vortex failure”. It’s also worth comparing the location of the daughter vortices in this event versus what’s been happening so far in 2019.


Figure 4: Animation of 10 hPa geopotential height (left) and temperature (right) during the Jan 2009 SSW. Animation created by Patrick Martinaeau (

This SSW was also a Polar-night Jet Oscillation (PJO) event (Hitchcock et al. 2013), so had the associated impressive appearance on time-pressure plots of polar-cap geopotential height (Figure 5) – with a long-lasting signal in the lower-stratosphere. According to Karpechko et al. (2017)’s Table 1, 100% of the days 8-52 after the central date had a negative 150 hPa NAM (one of only 5 times that has occurred since 1979). I always think Jan 2009 looks like a side-on view of a foot stamping down on the troposphere! You can also see that the event affected the troposphere until late March – over 2 months after the initial warming.


Figure 5: 65-90N standardized geopotential height anomalies for JFM 2009 (credit: NOAA CPC).

This moves the discussion nicely onto the downward propagation of the event. It wasn’t as strong as some SSWs – only 69% of the days 8-52 after the event had a negative 1000 hPa NAM index, and therefore ranks nearer the bottom end of Karpechko et al.’s “downward-propagating” SSWs (dSSW) (and that’s not surprising looking at Figure 5). However, it was nevertheless a dSSW, and had various impacts across the N Hemisphere. I’m somewhat limited in my analysis by the current US government shutdown, so I’ll focus on the British Isles.

Figure 6 (a snippet from some work I’ve been doing on SSWs and easterly outbreaks in the UK) shows the British Isles were influenced by easterlies 7-13 days after the SSW – which was the only case of mean easterlies in the 45 days following the event. Using this metric… the 2009 SSW isn’t special at all, although does pass my semi-arbitrary threshold of 5 consecutive easterly days for a true “outbreak”. However, I still do find it incredible that such a huge, hemispheric phenomenon as a major SSW, involving massive planetary waves and propagation from 30-50 km above our heads, can have a detectable response in such a small area as the British Isles. That’s one of those “isn’t the atmosphere amazing?!” moments.


Figure 6: Average 850 hPa zonal winds across the British Isles in the 45 days following the January 2009 SSW, according to JRA-55 reanalysis.

This easterly spell brought with it colder-than-normal temperatures and snow. The Met Office’s monthly summary for Feb 2009 notes that “it was very cold during the first part of the month with snowfalls in many areas. This was the most widespread snowfall as a whole since February 1991”. Figure 7 shows the Met Office surface analysis for 18Z Feb 1st, with a negative NAO pattern and an easterly flow over NW Europe evident.


Figure 7: 18Z Met Office surface pressure and frontal analysis for Feb 1st 2009.

Finally, the January 2009 SSW will always be special to me on a personal level, as it was the first time I had heard of sudden stratospheric warming and its influence on the tropospheric weather patterns. A schematic posted by the Met Office in a press release (announcing an increased likelihood of cold weather in the next few weeks), showed wind reversals propagating down from the stratosphere to the troposphere and eventually the surface. This fascinated me, and it started a journey which, 10 years later, finds me doing a stratosphere-related PhD project.

met office ssw

Figure 8: Met Office diagram showing the downward propagation of zonal wind reversals associated with a major SSW.


February 2009 – Met Office:

Hitchcock, P., T. G. Shepherd, and G. L. Manney, 2013: Statistical Characterization of Arctic Polar-Night Jet Oscillation Events. J. Climate., 26, 2096-2116,

Karpechko, A. Y., P. Hitchcock, D. H. W. Peters, and A. Schneidereit, 2017: Predictability of downward propagation of major sudden stratospheric warmings. Quart. J. Roy. Meteor. Soc., 143, 1459-1470,

NASA MERRA-2 Annual Meteorological Statistics:

NOAA CPC Stratosphere-Troposphere Monitoring:

Polar vortex animation during Stratospheric Sudden Warming [Patrick Martineau]:

Wetter3 UKMO surface chart archive:

Not all SSWs were created equal

Non-downward propagating SSWs? 

Major stratospheric sudden warming events (SSWs) attract widespread attention because they are now known to have significant impacts on the tropospheric circulation (e.g. Baldwin and Dunkerton 2001, hereafter BD01). Anomalies in the stratospheric circulation (often expressed as the Northern Annual Mode (NAM) index, or polar cap geopotential height anomalies) propagate downwards through the stratosphere into the troposphere, rather like “dripping paint” (such as BD01 Fig. 2). A major SSW is associated with the development of a negative NAM in the stratosphere; the “typical” response is the development of a negative NAM (or the associated NAO/AO) in the troposphere ~10-14 days after the central date of the SSW (when the 10 hPa 60N zonal-mean zonal wind becomes easterly) which can persist for several months.

However, not all SSWs were created equal – and some SSWs do not strongly couple to the tropospheric circulation. A recent study by Karphechko et al. (2017) classified major SSWs as “downward propagating” (dSSW) or otherwise (nSSW) based on the 1000 hPa NAM index following the event, and found 43% were nSSW – i.e., not followed by a strong and persistently negative surface NAM. This is not a small fraction of SSWs, and the atmospheric evolution following the two types was found to be significantly different. 

Our perception of SSWs in recent years has been highly influenced by a relatively unusual clustering of vortex-split, downward-propagating events (Jan 2009, Feb 2010, Jan 2013 and Feb 2018) which all had similar tropospheric impacts (all 4 of those events were followed by an outbreak of snow/cold in the UK, for example). The most recent nSSW occurred in Feb 2008. Thus, the announcement of a major SSW – particularly on social media – has become synonymous with a specific weather pattern.

In the nSSW cases considered by Karpechko et al., the composite (their Fig. 1c) actually shows intermittently positive NAM in the troposphere following the SSW – with the sign of the NAM opposing between the lower stratosphere and the troposphere for ~40 days following the central date. This is very different to the picture of dripping -NAM anomalies into the troposphere that BD01 made famous (which is consistent with Karpechko et al.’s dSSW).

Composites of all major SSWs are influenced by the higher frequency of dSSW and the stronger circulation anomalies induced, but this work suggests we need to be wary of these stratospheric events which don’t strongly influence what happens beneath. However, forecast models often struggle to predict the downward propagation – so forecasting these events is troublesome. It also presents a communication problem, which current forecasts (see below!) suggest we may be about to run into: a major SSW could mean a significant reversal of the normal tropospheric circulation (with the potential for “Beast from the East”-type events in the UK), or it could mean very little (e.g. January 2002 following the non-downward propagating Dec 2001 SSW). Predicting these differences, and understanding the mechanisms involved, is an area of active research – and something I hope to address in my PhD work.

Do current forecasts suggest nSSW or dSSW?

As I write this, we’re in a tentative stage – the main stratospheric heat flux event has occurred, and the 60N zonal-mean zonal wind has reversed to easterlies in the upper stratosphere. However, at 10 hPa we’re still decelerating – with the event expected to become ‘major’ around Jan 1 (Fig. 1 & 2) if current forecasts are correct (inter-model agreement has substantially increased now the upper-stratospheric reversal is in the observations).  The event looks very likely to be first driven by a wave-1 displacement of the vortex towards Eurasia, with an increasing likelihoodo of a vortex split (wave-2) to then occur, with the dominant daughetr vortex over Eurasia and a smaller vortex over N America (interestingly, this is opposite to what happened in Feb 2018). However, agreement on the split evolution remains lower than the displacement.


Figure 1: Forecasts of the 10 hPa 60N zonal-mean zonal wind from 00Z December 27th. There is a good agreement between the GFS and its ensemble of a major SSW occurring around Jan 1st.


Figure 2: ECMWF operational forecast from 12Z December 26th for 12Z January 1st showing a major SSW. Source: 

So, predicting the tropospheric impacts is a challenge when the stratospheric forecasts don’t agree! The spread in the GEFS forecasts beyond 10 days is very large – with some members showing a quick return to stratospheric westerlies whilst others flirt with record-strong easterlies. There’s even some indication of bifurcation in the ensemble at longer ranges (perhaps relating to whether or not a split occurs), which may render the ensemble mean of less use.

Despite the uncertainty, one aspect that has been relatively persistent is the absence of a signal for downward propagation in the deterministic GFS (Fig. 3) and the longer-range models such as CFSv2 (Fig. 4). Comparing Fig. 3 here with the nSSW composite in the Karpechko paper is striking – there are many similarities, including the weak -NAM before the main event and the ~day 10 tropospheric +NAM development. On its own, this screams nSSW – but of course is just a single deterministic forecast from one model.


Figure 3: GFS NAM analyses and forecasts from 00Z December 26th. Source: Zac Lawrence’s website ( 

The CFSv2 initially trended strongly towards a -NAO for January 2019 as the SSW signal grew – but this has since decayed and transitioned more towards an Atlantic ridge pattern (Fig. 4). The model clearly picked up on a major SSW occurring – but, like all forecast systems this time, has struggled to predict the type of SSW. There is currently no indication (Fig. 5) from the CFSv2 forecasts of a widespread hemispheric cold outbreak (a “warm Arctic-cold continents” pattern).


Figure 4: CFSv2 forecasts from Dec 1 – Dec 27 for January 2019 700 hPa geopotential height anomalies. Note the initial trend away from a +NAO towards a strong -NAO, before trending towards an “Atlantic ridge” pattern.


Figure 5: CFSv2 2m tempertaure anomaly forecast for January 2019 from an ensemble of forecasts launched between Dec 16-25. Base period 1999-2010. Source: 

My advice would be not to hold your breath for a “Beast from the East 2019 Edition”. But as predictability typically increases once a major SSW has occurred, we should gain a much better picture in the first few days of 2019.

Takeaway message: the impacts of SSWs are more complex than whether it is a displacement or a split, and the mere reversal of the 10 hPa 60N zonal wind doesn’t mean you’ll be shovelling snow 2 weeks later.


Baldwin, M. P., and T. Dunkerton, 2001: Stratospheric Harbingers of Anomalous Weather Regimes. Science, 294, 581-584,

Karpechko, A. Y., P. Hitchcock, D. H. W. Peters, and A. Schneidereit, 2017: Predictability of downward propagation of major sudden stratospheric warmings. Quart. J. Roy. Meteor. Soc., 143, 1459-1470,