Editor's Note (11/13/12): This article was edited after original publication in the print edition to include several corrections and clarifications.

The past three winters in parts of North America and Europe were unusual. First, during the winters of 2009–2011, the eastern seaboard of the U.S. and western and northern Europe endured a series of exceptionally cold and snowy storms—including the February 2010 “snowmageddon” storm in Washington, D.C., that shut down the federal government for nearly a week. Later that year, in October, the NOAA Climate Prediction Center (CPC) forecasted a mild 2010–2011 winter for the eastern U.S., based on a La Niña pattern of cooler than usual ocean temperatures in the eastern Pacific. But even with La Niña's moderating effects, very low temperatures and record snowfalls hit New York City and Philadelphia in January 2011, catching the CPC and other forecasters by surprise.

The winter of 2011–2012 brought even more surprises. The eastern U.S. had one of its mildest winters in history, while other parts of North America and Europe were less fortunate. In Alaska, the average January temperature across the state was a stunning 10 degrees Celsius below the month's long-term average. A single storm buried towns in southeastern Alaska in up to two meters of snow. At the same time, an extended outbreak of frigid weather descended on central and eastern Europe, bringing temperatures of −30 degrees C and snowdrifts that reached rooftops. By the time the deep freeze lifted in early February, more than 550 people had lost their lives.

How can we explain these outbreaks of severe weather, during a decade between 2002 and 2012 that was the warmest in the 160 years that instruments have tracked global temperatures? Scientists appear to have found an answer in a very unusual time and place: the recent, record-breaking losses of summer sea ice in the Arctic Ocean.

The Trigger: Record Ice loss

The arctic has changed considerably since my first trip above the Arctic Circle in April 1989. The most obvious change has been the diminishing extent of sea ice during the summer. Every winter the Arctic Ocean almost completely freezes over. The winter sea ice is composed of thick, multiyear ice that has accumulated over time and much thinner first-year ice that has frozen in parts of the ocean that had been open water the previous summer. Each September the summer melting reduces the extent of the sea ice to its annual minimum.

Back in 1989, the winter sea-ice extent was slightly more than 14 million square kilometers. About seven million of that was the thick, multiyear ice that persists through the summer. The situation today is different. Although the extent of winter sea ice in 2012 was close to that of 1989, only about half—slightly less than 3.5 million square kilometers—survived through this past September, a record summertime low.

The loss of Arctic sea ice during summer has not been gradual or linear. From the time ice measurements by satellites began in 1979 until 2000, losses in sea-ice extent were not especially obvious. From 2000 to 2006 the rate of decline accelerated, but it was not until a significant change occurred in 2007 that the world took notice. During that one year, the minimum summer sea-ice extent dropped by 26 percent, from about 5.8 million square kilometers in September 2006 to about 4.3 million in September 2007. This unprecedented reduction in multiyear ice caused scientists to reassess their projections for when the Arctic Ocean would experience its first ice-free summer. Based on data collected prior to 2007, the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report had projected that the first ice-free summer would most likely occur toward the end of the 21st century. Most studies now project that the event could happen decades sooner, between 2020 and 2040.

The changes in sea ice are part of the amplification of global warming that has been impacting the Arctic during recent decades. Although the rest of the world has observed a modest average temperature increase of about 0.8 degree C since the beginning of the 20th century, average temperatures in the Arctic have warmed by more than double that amount over the past 50 years. This rapid warming has altered Arctic weather patterns and melted vast areas of permafrost. Such modifications of the physical environment have disrupted critical habitats for the region's wildlife and threatened the long-term survival of many species. Similarly, the Arctic's native peoples, long renowned for their cultural adaptations to the region's cold and ice, are witnessing a significant disruption to their way of life and an increasing threat to their heritage.

Although these changes may seem remote to most of us living below the polar regions, the rest of the Northern Hemisphere is not immune to the effects of Arctic amplification and sea-ice loss. Midlatitude weather patterns are affected by Arctic climate, which raises a key question: Is global warming behind the recently observed outbreaks of severe winter weather, or do the outbreaks simply fit into the general pattern of the planet's natural climate oscillations?

Pressure in the atmosphere

Nature certainly made itself felt when I was growing up near Washington, D.C., in the 1960s, as I trudged through the snow to school with my friends during a decade of unusually harsh winters. Scientists now know that the sources of those cold and snowy winters were two natural climate oscillations referred to today as the Arctic Oscillation (AO) and the North Atlantic Oscillation (NAO), although they were unnamed at the time [see box on page 54]. These two climate oscillations arise from interactions between the atmosphere and ocean, exhibiting their most noticeable effects during the winter.

The strength of each oscillation is characterized by an index that quantifies anomalies—deviations from the long-term average—in the wintertime distribution of atmospheric pressure over a specific region. In the case of the AO index, the region is very large, encompassing most of the Northern Hemisphere, from the North Pole southward to the boundary of the tropics at 20° N (about the latitude of Cuba). The AO index can be positive or negative. Positive values correspond to lower than average pressures in the Arctic and higher than average pressures in the subtropics. During positive phases of the AO index, anomalously low pressures in the Arctic lead to a strengthening of the polar vortex, a persistent circulation of upper atmospheric winds flowing from west to east around the Arctic. A strengthened polar vortex tends to retain cold Arctic air masses north of the Arctic Circle.

In contrast, during negative phases of the AO index, anomalously weak low pressures in the Arctic weaken the polar vortex. It is less able to constrain cold Arctic air masses, allowing them to invade the middle latitudes to the south and deliver outbreaks of cold weather and increased snowfall. The U.S. eastern seaboard and northern Europe are especially vulnerable to these events during periods of strongly negative AO conditions.

The NAO index characterizes anomalies in the wintertime distribution of atmospheric pressure over a much smaller region, encompassing the Atlantic sector of the Northern Hemisphere between the subtropical high-pressure center near the Azores and the subarctic low-pressure center near Iceland. As is true for the AO index, the NAO index can also be positive or negative. Positive values correspond to higher than average atmospheric pressures near the subtropical high and lower than average pressures near the subarctic low. During positive NAO conditions, the enhanced pressure differences strengthen the westerly winds that blow year-round from west to east across the Northern Hemisphere's middle latitudes. The pressure differences also steer the fast-moving, circumglobal current of air known as the jet stream on a northeastward path from the eastern seaboard of North America toward northern Europe. Winter storms that cross the North Atlantic follow a similar track, delivering wetter and milder weather to northern Europe.

In contrast, during negative NAO conditions, reduced pressure differences weaken the westerly winds, and the jet stream leaving North America sweeps more sharply to the north, reaching Greenland before swinging back southward toward Europe. In this case, however, the storm track diverges from the jet stream, crossing the North Atlantic directly toward southern Europe and the Mediterranean, delivering wetter and milder weather to those areas. Northern Europe is left cold and dry.

Climate scientists disagree as to whether the AO and NAO should be treated as two distinct modes of natural climate variability. Some argue that the NAO is just a North Atlantic manifestation of the AO; others say that the dynamics of the two are different enough to warrant treating them separately. Although the two indices are highly correlated, their behaviors occasionally diverge in important ways, as they did last winter.

Stacking the Deck for Severe Winter Weather

As society's greenhouse gas emissions continue to alter the earth's climate system, whatever changes occur will be superimposed on the system's natural climate oscillationas. Discerning the human contributions to changes in climate is difficult and requires hypothesis testing. Recent research has provided new evidence strengthening the hypothesis that global warming and Arctic sea-ice loss are affecting our winters today by disrupting the normal rhythms of the AO and NAO.

Looking back to my childhood in the 1960s, we see that the AO and NAO indices were predominantly negative, leading to winters that were snowier and colder than average along the eastern seaboard of the U.S. There is no reason to suspect that this decade of inclement winter weather was anything more than the natural variability to be expected from the AO and NAO. In contrast, from the 1970s to 1990s the NAO index was predominantly positive, with only an occasional negative NAO winter. The resulting mild winters coincided with increased societal awareness of global climate change and led many scientists to hypothesize that increasing concentrations of greenhouse gases might be behind what appeared to be an unusually long run of predominantly positive NAO winters. Models referenced by the IPCC predicted that this trend would continue with the steady rise in greenhouse gases. Yet the predominance of strongly positive AO and NAO winters came to a close during the latter half of the 1990s.

Although the run of positive NAO winters ended, that change does not mean the hypothesized relation between increasing greenhouse gases and the NAO was incorrect. What was not anticipated at the time was the acceleration of Arctic amplification starting in the late 1990s. As the amplified effects of global warming played out above the Arctic Circle, climatic conditions entered into a period that Jim Overland of the National Oceanic and Atmospheric Administration and his colleagues have called the Arctic Warm Period. This period has been characterized by rapid losses in Arctic sea ice, the Greenland ice sheet, permafrost and continental glaciers. Central to each of those changes is a process referred to as ice-albedo feedback, in which an area's reflection of incoming solar radiation is diminished as its ice cover melts, exposing darker land or sea surfaces.

Scientists have been especially concerned with ice-albedo feedback in the Arctic Ocean. The loss of summer sea ice exposes more ocean water to incoming solar radiation. The absorption of this radiation leads to excess heating of surface waters, which results in two important feedbacks. First, a portion of the excess heat reinforces the summertime melting of sea ice. Second, the ocean gradually releases much of the remaining excess heat into the atmosphere during the fall, increasing atmospheric pressure and moisture in the Arctic while decreasing the temperature differences between the Arctic and middle latitudes.

An increase in the Arctic atmospheric pressure and a decrease in the temperature gradient favor the development of negative AO and NAO conditions during winter. This situation leads to a weakening of the polar vortex and jet stream. A weakened polar vortex is less able to constrain cold Arctic air masses, with their elevated moisture content, from spilling down into the middle latitudes and delivering severe outbreaks of cold weather and snowfall.

Furthermore, a weakened jet stream exhibits larger waves in its trajectory, ones that can get stalled in place, locking an affected region in a deep freeze. In combination, these altered atmospheric circulation patterns tend to stack the deck in favor of more frequent and persistent outbreaks of severe winter weather in North America and Europe.

Other factors can come into play, however. The El Niño/Southern Oscillation, another powerful climate oscillation centered in the Pacific Ocean, can strongly influence winter weather across the continental U.S. In the southeastern and middle Atlantic regions of the U.S., El Niño years bring wetter winter weather, while La Niña years bring drier winter weather. Together negative AO and NAO conditions during El Niño years can exacerbate the chance for cold, harsh winters along the eastern seaboard, which happened in 2009–2010. Negative AO and NAO conditions can also counteract the dry, mild winters expected during La Niña years. This was the situation during winter 2010–2011, when the low temperatures and record snowfalls in New York City and Philadelphia surprised forecasters who, based on the La Niña alone, were anticipating milder conditions.

The Winter Ahead

Although changes in the Arctic may have stacked the deck in favor of more frequent and persistent outbreaks of severe winter weather in the future, we can never be sure what hand will be dealt in any given year. After all, forecasting the weather always entails some level of uncertainty.

The 2011–2012 winter was a good example of the predictive challenges. The CPC forecasted relatively mild weather in the eastern U.S. because of a developing La Niña in the Pacific. The AO and NAO indices had started off positive during the early winter, but then negative AO conditions emerged during mid-January and persisted through early February, while the NAO stayed positive. Alaska and parts of central and eastern Europe were buffeted by deadly cold and heavy snowstorms, while weather in the eastern U.S. remained unseasonably mild. A La Niña associated high-pressure system in the eastern Pacific created a blocking pattern that steered the jet stream much farther north over North America than usual during the mid-winter's negative AO conditions, allowing warmth from the Gulf of Mexico to move up to the eastern U.S., leading to the region's fourth warmest winter on record. The jet stream's more northern path also brought relatively mild conditions to the North Atlantic and western Europe.

By early March the atmospheric high-pressure system in the eastern Pacific strengthened, further amplifying the unusual weather conditions and resulting in record high temperatures throughout the Midwest and eastern U.S. Despite the persistence of unseasonable warmth there, however, it should be noted that other parts of the Northern Hemisphere ended up with an unusually cold winter and early spring. In fact, the NCDC reported that the average global temperature for March 2012 was the coolest since 1999.

For the upcoming winter of 2012–2013, the cards appear to be especially stacked in favor of harsh weather outbreaks in North America and Europe. The record-setting Arctic sea-ice loss that was observed during this past summer should enhance the probability of cold Arctic air masses invading midlatitude regions. Although it is difficult to predict which midlatitude-regions will be most vulnerable, El Niño conditions developing in the Pacific during fall 2012 may increase the odds for a cold and punishing winter in the eastern U.S. The eastern seaboard could be especially vulnerable to the region's infamous nor'easter storms that bring bitter temperatures and deep snow. Although no one can say whether we will see a repeat of the unusually harsh nor'easters of winter 2009–2010, the summer and autumn buildup to the current winter bears a stronger resemblance to the conditions that unfolded during 2009 than to any other year since 2007, when the significant change in Arctic sea-ice loss occurred. As the next few months unfold, we will see what wild cards emerge in the hand we are dealt.