Is the melting Arctic changing midlatitude weather?

In the winter, little solar radiation reaches the Arctic surface, and the region cools by the loss of IR energy to space. The loss is balanced by the transfer of energy from lower latitudes via atmospheric winds and ocean currents. On a rotating Earth, the north–south temperature difference produces cyclonic winds that flow primarily west to east across midlatitudes and the sub-Arctic. The counterclockwise wind pattern in the upper atmosphere surrounding the North Pole is traditionally called the polar vortex. (A twin Antarctic vortex flows clockwise in the Southern Hemisphere; see the article by Anne Douglass, Paul Newman, and Susan Solomon, Physics Today, July 2014, page 42.)

Weather tends to follow the direction of the mid-altitude winds. Figure 1 illustrates two configurations of the polar vortex; the winds follow contours of constant color. One configuration (left) from November 2013 had a relatively circular flow pattern around the North Pole, and the other (right), which emerged less than two months later, had a wavy flow pattern from numerous low-pressure centers. The jet stream (the white region) borders the polar vortex along its southernmost edge.

When the polar vortex is strong, driven by a large north–south pressure gradient, so are the westerly winds. But as the vortex weakens, wind directions can shift and meander north and south. A wavier jet stream thus has the potential to bring cold air to midlatitudes and warm air northward.

Because of the complexity of the atmospheric physics behind circulation patterns, though, it’s unclear exactly how a warming Arctic may give rise to a wavy jet stream. Although changes in north–south temperature differences can influence the jet stream, winds flowing strictly west to east are unstable on a rotating planet; instabilities cause north–south meanders in the wind pattern and eventually generate high- and low-pressure centers and severe weather, such as locally strong, cold winds and snow. When and where such instabilities form is unpredictable beyond a week or so in advance; meteorologists refer to atmospheric winds as chaotic, random, or internally variable. Indeed, the position of the jet stream over the Atlantic Ocean can naturally vary in latitude by up to 10° yearly.

In addition to that internal variability, forced variability in the wind pattern complicates the picture. Sea-surface temperature anomalies in the tropics, midlatitudes, and ice-free parts of the Arctic all influence atmospheric temperatures. Low- altitude, warm air increases the height of mid-altitude constant-pressure surfaces—the so-called geopotential, or gravity-adjusted, heights—and changes in the slope of those pressure surfaces influence the local winds. The greater the slope, the greater the wind speed. So long as the anomalies persist, they can be considered an external forcing on the atmosphere. And because the atmosphere can support waves, the external forcing need not remain local but can affect distant regions, an interaction known as teleconnection.

The energy associated with just the internal variability of midlatitude weather systems is larger than that associated with the external-forcing contribution from a warming Arctic. One must distinguish the effects of Arctic forcing on midlatitude atmospheric circulation not only from those of random events, but also from those driven by tropical and subtropical temperature anomalies. Detecting the Arctic influence is challenging because of its small signal-to-noise ratio. Compounding those difficulties is the relatively short period of time—roughly 15 years—over which major Arctic amplification has occurred.

Scientists are wrestling not only with how large and consistent an impact on climate to expect from Arctic warming but with whether it is even recognizable. This past year climatologists Elizabeth Barnes and James Screen broke the linkage issue down to three questions: Can Arctic warming influence the midlatitude jet stream? Has it significantly done so? Or will it do so in the future?⁶ Extensive simulations strongly suggest that near-surface Arctic warming and loss of sea ice can modify the midlatitude circulation. In most of those simulations, unsurprisingly, the responses were small compared with internal variability. And as Barnes and Screen point out, the mere plausibility of causal connection implies neither that Arctic warming has exerted a discernable influence in the past nor that it will do so in the future; other factors that control midlatitude weather may be too obscuring.

Model results from several studies by Screen and others this past year^7–9 show little support for future seasonal average midlatitude cooling trends or increases in the number of extreme weather events over the Northern Hemisphere. Other results suggest that if the Arctic does affect severe weather to the south, its influence is regional in extent and episodic, even within seasons.¹⁰ 

Fortunately, one can look for patterns in the variability itself using what’s known as the Arctic oscillation (AO) index to discern correlations with episodes of severe weather. The AO index is a measure of the dominant pattern of sea-level pressure variations north of about 55° N latitude. A positive AO corresponds to a strong zonal, or west–east, direction to the wind flow, exemplified by figure 1a, whereas a negative AO corresponds to a more north–south flow, characteristic of the wavy jet-stream pattern, exemplified in figure 1b, thought to produce the abnormally cold midlatitude winters.

Figure 2 shows the evolution in the AO index in Decembers for the past 65 years. Interestingly, the AO index reveals greater variability in wind patterns over the past decade: Both positive and negative AO values have increased in magnitude relative to earlier years in the time series.¹¹ Although the increased variability may be caused by internal atmospheric dynamics and has occurred over a period too short to yield definitive proof for or against linkages, it does yield information about current Arctic atmospheric conditions not emphasized in the studies mentioned above, which point to decreased atmospheric variability in a future Arctic. During the past five winters (2009–10 through 2013–14), the months of December and January have exhibited twice the expected number of negative AO events.¹¹ An increase in AO variability could be an early sign of destabilization in the Arctic jet stream. But is the increased wind variability related to recent thermodynamic changes in the Arctic?

Evidence behind a connection comes from case studies in north-central Asia and the eastern US.^10,12 In the first case, analyses of field observations over the past decade and of simulations suggest a direct linkage between the regional loss of sea ice in the Barents and Kara Seas north of Eurasia during the autumn months and cold, stormy conditions experienced in eastern Asia. The winters were colder during years with negative phases of the AO than during years with positive AO phases.

The chain of causality is complex. The atmospheric forcing starts with a loss of sea ice and relatively warm ocean temperatures. As the atmosphere cools below freezing in early winter, the thin sea ice readily transmits the ocean’s heat and moisture to the atmosphere. Locally warmer Arctic air reduces the north–south temperature gradients, which in turn reduce the strength of the westerly winds. Thanks to the slower wind, atmospheric pressures can build over Siberia, and cold air can collect on the east side of the Siberian high-pressure region. Smaller-scale storms can then sweep that cold air further eastward to China, the Koreas, and Japan. In 2014 Masato Mori and colleagues used a wind circulation model to show that the probability of severe winters in those countries has more than doubled because of the reduction in Barents–Kara sea ice.¹³ The association with cold weather in eastern Asia is to date the most consistently supported of the possible regional Arctic–midlatitude linkages.

The Arctic’s influence on North America is more speculative. It’s thought that instead of initiating midlatitude weather, the Arctic reinforces existing patterns—specifically a wavier jet stream. Internal instabilities in the atmosphere influence the wind pattern over the North Pacific Ocean and largely determine whether the downstream wind pattern over North America is primarily zonal or wavy. Nonetheless, consider recent years with large December AO index values of both signs since 2006. Figure 3a shows, for reference, a 30-year average of the December altitudes for a 700-hPa pressure level. The figure’s subsequent two panels, 3b and 3c, map the height deviations from that mean pattern for three Decembers (2006, 2011, and 2013) with large positive AO values and three Decembers (2009, 2010, and 2012) with large negative AO values, respectively.¹¹ 

In figure 3b, the negative height anomalies (blue) are centered near Greenland, a pattern that strengthens the north–south height gradient in positive AO years and thus strengthens the westerly winds over central and eastern North America. By contrast, in the negative AO years averaged in figure 3c, positive height anomalies (red and orange) are centered on Greenland. They counteract the influence of the low-height center in the region’s fundamental climatology (figure 3a) and divert the winds from a more zonal pattern into a more north–south one, carrying cold air south and warm air north.

The warm temperature anomalies over Baffin and Hudson Bays in figure 4 occurred during negative-AO Decembers relative to positive-AO Decembers and are consistent with the greater heights west of Greenland for the negative-AO cases in figure 3c. Those warmer temperatures increase the vertical distance between different constant-pressure surfaces in the atmosphere, which in turn modifies the wind patterns.

Climatologist Kevin Trenberth and colleagues recently argued that forced changes in thermodynamic variables, such as increases in temperature and moisture, from global warming tend to show sustained impacts on the environment, whereas changes in dynamic variables, such as wind circulation, appear to be mostly random and chaotic.¹⁴ The same argument applies to how Arctic warming may be linked to remote weather: Thermodynamic processes in the Arctic may reinforce and prolong, but not necessarily cause, an already wavy wind pattern. Thus, given two possible Northern Hemispheric patterns, zonal and wavy, one does not necessarily find, either in simulations or field observations, the same midlatitude response from similar Arctic forcing.^7,15 The difference implies that the connection between Arctic warming and midlatitude regions is state dependent, determined by whichever large-scale atmospheric wind pattern is dominant at any given time.

Two years ago, in an analysis of simulations in which sea-ice concentrations were reduced at a known rate, Screen and his colleagues concluded that if the ice loss was the only change in the system, it would take at least 50 years to distinguish the forced signal from internal variability in the large-scale wind pattern.¹⁶ But even rather well-accepted climate models can have trouble capturing some of the regional atmospheric dynamics associated with linkages. In a recent comparison of results from two models and from two data sets, which assimilate observations into the simulations, the models compared poorly with observations near 60° W longitude,¹⁷ a critical region for potential Arctic links to eastern North America.

As Barnes and Screen point out, the jury is still out on whether Arctic warming has tangible implications for weather at lower latitudes.⁶ Researchers continue striving to incorporate Arctic-specific physics into their models. For example, the effects of clouds and sea ice on climate are important but especially difficult to simulate. Better constraining the forecasting models greatly depends on increasing the number of observations of Arctic regions. Despite the current controversy over how the changes in the Arctic are manifest over the more populated parts of our planet, the predicted doubling of Arctic amplification over the next decades provides a strong incentive to understand how Arctic change interacts with a mostly chaotically varying jet stream. Extending weather forecasts from a week to a month or more would provide one way to anticipate and adapt to whatever is in store for us.

This work is supported by the NOAA Arctic Research Program of the Climate Program Office and by the US Office of Naval Research. I appreciate discussions with Jennifer Francis, Edward Hanna, Richard Hall, Seong-Joong Kim, James Screen, Ted Shepherd, Timo Vihma, and John Walsh.

James Overland is a research oceanographer at the National Oceanic and Atmospheric Administration’s Pacific Marine Environmental Laboratory in Seattle, Washington, and an adjunct professor of atmospheric sciences at the University of Washington.