A Tale of Two Winters

A review of winter just ended is “a tale of two winters” in more ways than one. With a quiet end to the season and signs of Spring now all around, the snow and freezing temperatures of December might seem a world away. But anyone who travelled over Christmas won’t forget so quickly, and UK airports in particular have had to learn some expensive lessons. December 2010 was the coldest in more than 100 years in the UK and it followed the cold winter of 2009/10. As we shall see, the two Decembers were very similar over the Atlantic and Europe, whereas the late winters were very different.

Early winter in both years was dominated by a band of cold anomalies around mid-latitudes, while the subtropics were warm and the Arctic very warm, continuing the longer term warming trend at high latitudes. The surface air temperature patterns for the two Decembers are remarkably similar over most of the northern hemisphere.  They differ most over western North America, consistent with the change from El Nino to La Nina conditions in the tropical Pacific between the two winters.

The large scale circulation anomalies that led to the surface impacts are evident in the jet stream winds and geopotential height in the upper troposphere, below. The Atlantic jet varies greatly but, on average, it extends northeast from the Gulf Stream region, bringing storms towards the UK and Iceland. In the last two Decembers the jet was split, with the strongest part oriented west-east and shifted south, sending storms into Iberia and the Mediterranean. The north Atlantic was dominated by a persistent ridge, with a weaker branch of the jet avecting warm air in depth towards Greenland and cold Artic air south over Europe. The ridge is evident in the height maps, where the climatological region of strong meridional gradient is split over the Atlantic.

There was also a persistent ridge in the north Pacific in both Decembers, but it shifted between the years, from the Gulf of Alaska in 2009 to the dateline in 2010. These patterns are typical of El Nino and La Nina. In an El Nino winter, tropical convection moves east towards the dateline, extending the upper level subtropical anticyclones and the north Pacific jet eastward over the Pacific – this was seen in December 2009, when the strong upper level winds extended beyond the dateline and the jet-exit ridge was located over the Canadian west coast. In a La Nina winter, tropical convection is confined over the Maritime Continent and northern Australia (this year giving the intense monsoon season that Nick Klingaman discussed in a January blog). The anticyclones are then confined to Asian longitudes and Pacific jet is short – all features of December 2010 where the jet-exit ridge was located over the mid-Pacific. These extratropical circulation changes are parts of Rossby wave trains arcing north and east from the tropical convection anomalies. They are sufficiently robust responses to El Nino and La Nina that they provide some predictability for north American winter weather.

The evolution of these circulation patterns through the two winters can be summarised in time series of the major modes of extra-tropical variability, shown below. The North Atlantic Oscillation (NAO) describes fluctuations in the latitude of the Atlantic jet, with NAO negative when the jet and storm-track are shifted equatorward. Recent research has shown that the onset of negative NAO occurs when the Rossby wave trough over eastern Canada breaks cyclonically, taking warm subtropical air north over Greenland. This “Greenland blocking” certainly characterises the last two winters. The Arctic Oscillation (AO) describes simultaneous jet shifts over the Atlantic and Pacific. Finally the Pacific North American pattern (PNA) describes Rossby waves propagating along arcing paths across the north Pacific.

The ridging and equatorward shift of the jets in both early winter periods appears as negative NAO and AO through both Decembers. But the remainder of the two winters were very different. Negative NAO and AO relented in early January in both winters, remaining positive this year but returning through the whole of February in 2009/10. This was a very stormy period in the subtropical Atlantic, causing flooding in Maderia and south-west Europe. In the Pacific the PNA pattern was mostly of opposite signs in the two winters, reflecting the different jet and ridge postions, due to the change from El Nino to La Nina.

The onset of negative NAO and AO in 2009 occurred in early December, when ridges developed into blocking anticyclones over both oceans. These split the pool of cold Arctic air in the troposphere (see the 250hPa height map for December 2009 above), creating a planetary wave pattern that propagated into the lower stratosphere, temporarily disturbing the polar stratospheric vortex. This year the blocking appeared not to disturb the stratosphere, and the stratospheric vortex remianed strong throughout the winter, as described by Andrew Charlton-Perez in an earlier blog.

The persistent negative NAO of the last two winters could have been chance, spontaneous variability of the Atlantic jet stream and storm-track. However, several factors were present in the more persistent 2009/10 winter that have historically correlated with negative NAO, and for which mechanisms have been hypothesised by which they can influence NAO. There isn’t space to pursue these in detail, but the “mind-map” shows the range of possible influences that have been investigated in recent years, including by members of the department.

Several of these factors are thought to affect the tropopsheric jets, including the phase of NAO, via the stratosphere. Simplified atmospheric models predict that cooling of the tropical lower stratosphere or warming of the polar lower stratosphere each tend to shift the jets and storm-tracks equatorward. A combination of low solar activity and the easterly phase of the Quasi-Biennial Oscillation (QBO) cools the tropical stratosphere, while a disturbed stratospheric polar vortex warms the polar stratosphere, as does drying of the lower stratosphere (there has been a drying trend since 2001). All these factors were present in winter 2009/10.

The influence of Atlantic SSTs on NAO is still uncertain, with the main atmosphere-ocean interaction thought to be atmospheric driving of a tripole pattern of SST anomalies by the NAO. This pattern in Spring does correlate with NAO the following winter, but the correlation could be due to interannual persistence of NAO. In Spring 2009 the SST pattern predicted positive NAO for winter 2009/10 according to this historical correlation. In contrast, dynamical forecasts for Winter 2009/10 from the preceding Autumn did predict a pattern similar to negative NAO in the west Atlantic, most likely part of the teleconnection pattern from the developing El Nino.

Composites of past El Nino and La Nina winters reveal that the teleconnections from the tropical Pacific to Europe change between early and late winter. Negative NAO, or at least anomalies that project onto it, are associated with La Nina in November and December but with El Nino in January and February. So this winter’s evolution did agree with past Pacific La Nina events. January-February 2010 were also consistent with past El Ninos, whereas December 2009 was not.

Much is still to be learnt about the factors influencing our winter weather in the UK and Europe. Most but not all of the factors discussed above are included in seasonal forecast models, so it is unclear whether the low skill of current European seasonal forecasts can be significantly improved. In order to understand past variability, and possibly attribute the behaviour in particular seasons, forecast re-runs (hindcasts) are needed in which individual factors are constrained. We can also use simpler models and clever diagnostics to investigate the anomalies and teleconnections.

You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.

Leave a Reply

XHTML: You can use these tags: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>