How random is our winter weather?

Climate scientists map out a hemisphere-wide web of connections

by Bob Henson

Although forecasters wrote the lines well in advance, the winter of 2002–03 strayed from its script across the United States. Extended outlooks, based on the presence of a weak to moderate El Niño, had called for wet weather along the nation’s southern tier and mild, dry conditions to the north and west.

In the Denver area’s heaviest winter storm in 90 years, some foothill locations got upwards of 200 centimeters (79 inches) of snow on 17–19 March. However, most of the intermountain West remains in a severe to extreme drought. (Photo by Carlye Calvin.)

Some of these trends played out—Michigan had its driest winter on record, for instance—but unforeseen events stood out in sharp relief. The mid-Atlantic endured one of its wettest, coldest winters in decades, with a parade of snowstorms marching up the East Coast. Instead of the expected north-to-south temperature contrasts, the nation was split from east to west, with record warmth across the Rockies and persistent cold east of the Mississippi.

These surprises point to the limits of the El Niño/Southern Oscillation (ENSO) as a long-range forecast tool. A flurry of research since the mid-1990s has explored another vast system, involving both the Atlantic and Arctic, that influences winter weather across the Northern Hemisphere. The full extent of this system and its physical mechanism are still matters of debate, but according to scientists at NCAR and elsewhere, it could serve as a useful tool for predicting weather swings within a given winter—and perhaps the long-term evolution of northern climate.

Bridging the Arctic and Atlantic

Two concepts, with considerable overlap, lie at the heart of this research.

  • The North Atlantic Oscillation (NAO) is a seesaw in atmospheric pressure that helps direct the flow of winter storms from eastern Canada to Europe. A positive NAO index means the contrast between high pressure over the Azores and low pressure in the far north Atlantic is stronger than normal. This usually drives mild Atlantic storms into the bulk of Europe but keeps the Mediterranean on the dry side (see graphic, pg. 5). A negative NAO index denotes a weakened pressure pattern that opens the door to cold, dry Arctic intrusions into northern Europe and wet, slow-moving systems across the south.
  • The Arctic Oscillation (AO) involves not just the Atlantic but the full band of winds that encircles the North Pole at about 55°N. When this vortex tightens (a positive AO index), it tends to lock Arctic air over the pole. When the vortex loosens (negative AO), frigid air masses can dive more easily into North America, Europe, and Asia. This pattern is also called the Northern Annular Mode, in parallel with its Antarctic counterpart, the Southern Annular Mode.

NCAR’s Clara Deser summarized the AO’s climate effects in February for the annual meeting of the American Association for the Advancement of Science. She noted that the Arctic vortex was weaker than usual early this winter, a switch from the strongly positive AOs noted since the 1980s. This might have helped lead to the bitter cold that swept the eastern U.S. and Europe—which seemed all the more bracing after the mildness of preceding winters.

Scientist Clara Deser and long-term visitor Christophe Cassou are among the NAO researchers in NCAR’s Climate and Global Dynamics Division. (Photo by Carlye Calvin.)

“The Arctic Oscillation has strengthened in recent decades, which is contributing to unusual warmth over the Northern Hemisphere land masses,” says Deser. Global climate models are on to this trend as well: as greenhouse gases increase, the models tend to produce a positive AO.

However, Deser stresses that the AO/NAO is an intrinsic aspect of climate. “You don’t need changes in greenhouse gases or sea-surface temperature or other external factors to find this pattern. It exists because of the nature of the atmosphere itself.”

Medieval discoveries

NCAR’s James Hurrell is lead editor of the most detailed book to date on the NAO (see photo). Hurrell says the cycle is “one of the oldest known world weather patterns.” Nearly a thousand years ago, Vikings recorded a classic aspect of the NAO: severe winters tend to strike Greenland in tandem with mild winters in Denmark, and vice versa


All about the NAO
Published by the American Geophysical Union last winter, "The North Atlantic Oscillation: Climatic Significance and Environmental Impact" is the most thorough treatment of the NAO to date. NCAR’s James Hurrell (above) edited the book with Yochanan Kushnir and Martin Visbeck (Columbia University) and Geir Ottersen (University of Oslo). The book’s twelve review papers cover the NAO’s history, predictability, ecosystem impacts, and other pertinent research. (Book cover art courtesy AGU; photo by Carlye Calvin.)

In the 1920s, pioneering climatologist Sir Gilbert Walker identified both the Southern Oscillation and the NAO. But unlike the oceanic metronome that gives ENSO its semi-regular beat, the NAO is a much more irregular cycle, less constrained by the ocean below it.

Researchers in the 1940s and 1950s, including Carl-Gustav Rossby and Edward Lorenz, focused attention on the west-to-east flow that girdles the middle and high latitudes. This provided a new way to think of the NAO: a variation in the strength of the westerlies across the Atlantic. Their work also paved the way for the current AO/NAO concepts, whose leading proponents include David Thompson (Colorado State University) and Michael Wallace (University of Washington).

Interest in the AO and NAO exploded in the late 1990s with the arrival of improved ocean-atmosphere models and more sophisticated statistical analyses. According to the Institute for Scientific Information, the number of papers that mention the NAO in either title or abstract leapt from 11 in 1996 to 179 in 2001.

What drives the cycles?

Scientists haven’t yet agreed on how the AO and NAO do or don’t differ. “In my view, the AO and NAO are different interpretations of the same phenomenon,” says Thompson. “I think the debate over this pattern attests to the absence of a unique theory for its existence in the first place.” In spite of the uncertainty, some potential forecast tools are starting to surface.

• Weather from the stratosphere. In 1999, Mark Baldwin and Tim Dunkerton (Northwest Research Associates) discovered that large changes in the stratosphere often precede similar changes in the Arctic Oscillation at the surface. Thompson and colleagues have since found that these stratospheric anomalies tend to precede extended cold snaps at ground level throughout much of the Northern Hemisphere. In related research, Thompson and Susan Solomon (NOAA) noted that similar downward-propagating stratospheric events are observed in the Southern Hemisphere, possibly initiated by the ozone depletion observed there since the 1980s.

Once a stratospheric climate signal hits the lower atmosphere—which can take a few days to several weeks—the effects may persist for a month or more. Even so, it’s not clear whether the stratosphere was responsible for this winter’s cold blasts, says Thompson. “This last winter there was some variability in the stratosphere, but nothing quite as dramatic as we’ve seen in other years,” he says.

Although many other factors influence our winter weather, Thompson is optimistic about using stratospheric data as a sign of weather changes to come in the 30- to 60-day time frame. “The evidence suggests that the usefulness of such forecasts may prove to be roughly comparable to the usefulness of forecasts based on ENSO,” he says.

• The Pacific influence. A peculiar sequence of high- and low-pressure centers straddling the midlatitudes was part of this winter’s weather picture. NCAR’s Grant Branstator was excited to see this pattern, because it supports a finding that emerged from his experiments with the NCAR Community Climate System Model. The idea is that energy in the northern midlatitudes gets channeled through the strong subtropical jet that roars across the globe around 30°N. One of the most prevalent modes of this channeling, says Branstator, is a pattern that looks remarkably like the AO/NAO.

Intrigued by this winter’s pattern, Branstator tried to reproduce it in the model by warming various parts of the tropical Pacific. The closest resemblance to the 2002–03 pattern occurred when the model’s ocean was warmer than average near the International Date Line. That’s close to where this year’s El Niño warming was centered.

“As these things go, this is a pretty good match,” Branstator says. He points out the likely presence of natural variation in the mix, but he’ll continue to examine the possible link between the tropical Pacific and the AO/NAO.

Further west, the role of the western tropical Pacific and Indian oceans in North Atlantic climate has been examined by modeling work by NCAR’s Hurrell and Martin Hoerling (NOAA). The two have found positive AO/NAO trends associated with rising sea-surface temperatures in both basins.

The other ocean

Thanks to El Niño, the tropical Pacific has long been a hot spot in climate modeling. Only recently has the tropical Atlantic been getting the attention it deserves, says Christophe Cassou, who is completing a two-year visit to NCAR’s Climate and Global Dynamics Division. To look at the AO/NAO system fruitfully, says Cassou, you have to analyze the short-term regime shifts that can get lost inside long-term averages.

Cassou will return this autumn to France’s European Center for Research and Advanced Training in Scientific Computation. He and colleagues at NCAR and CERFACS have been using a tool called cluster analysis to get at the distinct climate modes that predominate across the North Atlantic. Along with the positive and negative NAO regimes, they’ve found two other patterns at work over the last century, with some spatial overlap among the four. Only in the last 20 years has the positive NAO regime overwhelmed the other three patterns.

Ocean-atmosphere models are helping to relate these patterns to the Atlantic itself, says Cassou. But much work remains to be done—including work on the models themselves. “The basic climate features of both the tropical and extratropical Atlantic are poorly simulated in most of the models,” he says. Why? “Nobody knows precisely.”

With so many unanswered questions about Northern Hemisphere climate, Cassou’s reply may resonate for some time to come.


Also in this issue...

North America's ozone: a closer look

Super-sizing a community data trove

Chasing mesoscale monsters

Larry Winter: NCAR's new Deputy director

President’s Corner: University roles in the weather and climate services partnership

UCAR Community Calendar

Web Watch

Governance Update