The National Center for Atmospheric Research | UCAR | UOP
photo Home Our Organization Our Research News Center Education Libraries Community Tools


Quarterly logo

The big thaw

New CCSM runs predict more trouble for Arctic sea ice, permafrost


by Bob Henson

Holland and Lawrence
Marika Holland and David Lawrence. (Photo by Carlye Calvin, UCAR.)

It's been a tough year for ice. The Arctic Ocean's perennial ice pack shrank last summer to a record low extent. At its yearly minimum in September, the ice covered an area 20% less than its 1978–2000 average, corresponding to a loss that's double the size of Texas. Meanwhile, recent warm summers have led to a patchwork of degraded permafrost across central Alaska. Some bike paths near Fairbanks resemble roller coasters, and unstable trees are creating "drunken forests" as they topple in slow motion.

If the Community Climate System Model is anywhere near correct, these effects could be dwarfed by what's to come over the next few decades. At the American Geophysical Union's annual meeting in December, two NCAR scientists presented results based on a new set of CCSM runs with interactive soil and sea ice. The runs show that Arctic sea ice may essentially disappear each summer by 2050, and that more than half of northern near-surface permafrost could thaw by then, with over 90% of it severely degraded by 2100.

While these numbers are expected to evolve as modelers refine their techniques, the new data point to impacts that may transcend those found elsewhere. For instance, the 2005 Arctic Climate Impact Assessment—using five models and a low-end emissions scenario—projects that only 10–20% of current permafrost areas will be degraded by 2100. Only one of the ACIA's five models indicates that more than half of September sea ice will be gone by the year 2050.

Permafrost in peril

A report by NCAR's David Lawrence, published last December in Geophysical Research Letters, is the first to examine the state of permafrost in a global model that includes interactions among the atmosphere, ocean, land, and sea ice, as well as a soil model that depicts freezing and thawing. "People have used models to study permafrost before, but not within a fully interactive climate system model," says Lawrence, whose coauthor is Andrew Slater of the University of Colorado's National Snow and Ice Data Center.

About a quarter of the Northern Hemisphere's land contains permafrost, which is defined as soil that remains below 32°F (0°C) for at least two years. The soil model in the CCSM allowed Lawrence and Slater to track the interplay between the permafrost's active layer—the thin topmost layer, extending anywhere from a few centimeters to several meters deep, that melts and refreezes each year—and the deeper layers that remain frozen unless a long-term thaw works its way into the soil. The active layer across many parts of the Arctic has been expanding downward, responding to long-term warming. As it does so, it's thawing out soil layers that in many cases have been frozen for 2,000 years or more.

The CCSM soil model includes 10 layers spanning a range from the surface down to 3.43 meters (11.25 feet). The CCSM runs for the 21st century are based on high and low scenarios of greenhouse-gas emissions from the Intergovernmental Panel on Climate Change (IPCC). In the high-emission run, the area with permafrost in any soil layer shrinks from 4 million to just over 1 million square miles by 2050 and to about 400,000 square miles by 2100. Even in the lower-emission scenario, which assumes major advances in conservation and alternative energy, the permafrost area decreases by more than 50% to about 1.5 million square miles by 2100.

This sinkhole near Fairbanks, Alaska, formed due to the melting of a large ice pocket within permafrost that is gradually thawing as temperatures warm. (Photo courtesy Vladimir Romanovsky, Geophysical Institute, University of Alaska Fairbanks.)

The new permafrost analysis triggered a lively discussion at the AGU meeting. Some permafrost experts suggest that substantial amounts of heat could be transferred to deeper soil layers not depicted in the CCSM; these and other factors, they claim, might help put the brakes on permafrost degradation in the layers studied. The biggest concern expressed by Vladimir Romanovsky (University of Alaska Fairbanks) is the CCSM's limited soil depth and the model's inability to exchange heat between its modeled soil and deeper layers. "This makes the upper layer much more dynamic and biases it toward the warmer side," he says.

Based on field work, Romanovsky argues that some permafrost harbors up to twice as much ice wedged in the soil as simulated in the CCSM. The energy required to melt that ice would further slow the permafrost thaw rate, he believes. Also, the recent loss in permafrost has been focused near areas where the soil is disturbed by fire, agriculture, or construction. A more uniform degradation, as projected by the CCSM, has yet to materialize, although Romanovsky expects this may occur in the next decade. "Qualitatively, this model produces good results," says Romanovsky, "but we need to look more carefully at it and compare it with what we measure."

Lawrence acknowledges these uncertainties, some of which he says he and Slater recognized early on. He adds, "The critical question, from my perspective, is not precisely how much permafrost will degrade, but how that degradation will interact with climate and what feedbacks will result." With input from Romanovsky and others, Lawrence is now carrying out sensitivity tests to evaluate the impact of a deeper soil column in the model as well as the insulating effects of vegetation, including shrub expansion.

Eventually, Lawrence hopes to use the CCSM to evaluate the possibility that thawing soil could release major amounts of greenhouse gas. Estimates vary widely, but anywhere from 200 to 800 petagrams of carbon may be locked in permafrost, compared to current fossil-fuel emissions of around 7 petagrams per year. Getting a quantitative handle on this issue will require a soil model with interactive carbon that can determine how much of the decomposition is aerobic, which leads to carbon dioxide emission, or anaerobic, which leads to methane emission. The decomposition mode, in turn, depends on local hydrologic conditions, says Lawrence. "Simulations of the complex interaction between the hydrologic and carbon cycles in permafrost zones will be challenging and will take some time to develop," he says.

Summers without sea ice

The same CCSM runs that throw permafrost in jeopardy also spell trouble for the Arctic's mantle of sea ice. Averaging around 3 meters (10 feet) deep, the ice thickens and expands each winter and thins and contracts each summer, with a core of year-round ice atop the waters surrounding the North Pole. At the AGU meeting, NCAR's Marika Holland presented new data on the potential evolution of Arctic sea ice over the next century. The simulations, covered in more detail in a paper now in review, suggest that year-round ice may be little more than a memory by the late 21st century.

Holland discovered an abrupt transition that occurs in the CCSM runs early in the 21st century. During this interval, a rapid infusion of warm water from the North Atlantic melts large quantities of Arctic ice from below. The influx of warm water triggers a cascade of positive feedbacks, and the summer sea-ice minimum plummets within a decade from about 80% of modern-day coverage to about 20%. By the 2040s, the model average shows only fragments of September ice left along the far north coasts of Canada and Greenland. Otherwise, the Arctic is virtually ice-free each September through the year 2099.

Sea-ice 1979
Sea-ice 2005
These images show the extent of sea-ice coverage in September 1979 (left) and 2005 (right). (Images courtesy NASA.)

Each of the seven CCSM ensemble members depicts this rapid transition in the September ice cover. However, the timing and length of the abrupt transitions differ among the members, says Holland, with the onset occurring as early as 2015. Holland points to the unsettling rapidity of changes across the Arctic over the last few years as a sign of the region's potential for quick transformation. Many of the Arctic's largest glaciers are accelerating seaward while melting at their leading edges, producing a net retreat. Last summer—July in particular—brought record warmth across the region. "These changes are pervasive throughout the Arctic system," says Holland.

This isn't the only period of dramatic warming in the Arctic's observational record, Holland notes. The 1920s and 1930s saw warm anomalies comparable to recent ones, albeit from a cooler baseline. The latest temperature rise is consistent with simulations of anthropogenic influence, says Holland. However, she adds, "we really need a longer time series to put the historical changes into perspective."

Bruno Tremblay (Lamont-Doherty Earth Observatory) is teaming with Holland and other colleagues to put the CCSM through additional paces. "We're trying to find out what causes the pulses of Atlantic inflow into the Arctic, and what the criteria are for these rapid events to occur," says Tremblay. "They seem to be related to the ice thickness prior to the events in some simulations, but we haven't been able to pin them down in a nice, consistent way. " He says the model has some factors that may cause an unrealistically rapid ice retreat—such as a slight warm bias in oceanic heat flux—but some factors that work in the other direction, such as a layer of warm Atlantic water deeper than what observations show. Melt ponds that form atop warming ice and hasten its decline are also absent from the model, though they may soon be added.

Tremblay spoke on his recent work at a February joint meeting of CCSM's working groups on polar climate and climate variability. As part of the community outreach integral to the model as a whole, each of the CCSM's ten working groups meets once a year or more to review progress and brainstorm possible upgrades. The land-model group plans to examine Arctic feedbacks at its March meeting.

Creation of the CCSM's sea-ice model was led by Cecilia Bitz (University of Washington) and Elizabeth Hunke and William Lipscomb (Los Alamos National Laboratory). The end result stacks up well against its peers, says Tremblay, who is reviewing various Arctic simulations for the Intergovernmental Panel on Climate Change. "The CCSM has the best sea-ice representation in a global climate model," says Tremblay. "From what I'm seeing, it behaves a lot more like the real system than any of the other models." He adds that the CCSM isn't perfect, but "for the polar climate, this is as good as it gets."


On the Web

Sea-ice decline (NSIDC)

Community Climate System Model

CCSM Polar Climate Working Group



Also in this issue...

Untitled Document