Science Briefs

Photo of meteorological instrument being set

SHEBA meteorological instrumentation being set up on the Arctic sea ice. (Photo credit: NOAA/ESRL/Daniel Wolfe)

Satellite photo of Beaufort Sea

MODIS Terra image from July 25, 2006, showing the ice-laden Beaufort Sea region of the Arctic Ocean north of Alaska and northwestern Canada, where the SHEBA experiment was conducted from October 1997-October 1998. (Image credit: NASA Visible Earth)

Line plots of SHEBA temperature observations

SHEBA observations of the evolution of temperature over the course of winter within the atmosphere (red), at the snow surface (black), at the top of the sea ice (green), and at the ocean surface beneath the sea ice (blue). The air and snow surface temperatures warm when storm clouds pass over the site, acting as a blanket that traps heat, and then they gradually cool as heat radiates to space after skies clear. Although the sea ice layer is more massive than the atmosphere, the clouds persist long enough in each storm episode to affect heat conduction through the snow and noticeably warm the sea ice while they are present. (Image credit: NASA/GISS)

Clouds — An Unwelcome Blanket for Arctic Sea Ice?

The climate of the Arctic appears to be extremely sensitive to human influences, but it is relatively poorly understood because it is poorly observed and involves interactions among the ocean, the overlying sea ice, snow resting on the sea ice, and the atmosphere above. Climate models predict significant future Arctic sea ice decline as the planet warms, but the models disagree over the rate at which this will occur, and the rate of sea ice retreat observed during the satellite era is actually faster than the model predictions, suggesting that the models do not adequately simulate the physical processes that regulate sea ice thickness and extent.

One of the best Arctic datasets we have was acquired over a decade ago during the Surface Heat Budget of the Arctic (SHEBA) field experiment, which deployed a variety of instruments on the ice north of Alaska for almost a year. We revisited the SHEBA dataset, focusing first on the Arctic winter when there is no sunlight and the ice is too cold to melt, to understand how winter weather "prepares" the sea ice for the spring melt season.

Most previous studies of the Arctic atmosphere focused on the monthly average climate and how it varied over the seasons. We found, however, that the Arctic atmosphere and its effect on the surface are very different depending on whether a storm is passing and skies are overcast or whether high pressure is in place and skies are mostly clear. Overcast skies act as a blanket that traps heat and warms the sea ice and snow, while clear skies allow heat to be radiated away to space, cooling the snow and sea ice. These effects are exaggerated in the Arctic because the cold air is too dry for water vapor molecules to trap much heat, and because the clear and cloudy episodes last longer than in other parts of the world, giving the snow and sea ice more chance to react.

The Arctic mostly oscillates between these two states in winter, spending little time in between. Thus, unlike the "bell" curve that describes grades on a test in a big class, where many students get grades close to the average grade, the average winter climate of the Arctic almost never occurs — it is almost always much colder or much warmer than the average, either losing heat or gaining heat.

This has several implications for predictions of future sea ice decline. In winter, the clear and cloudy states occur just often enough that sea ice temperature fluctuates between warm and cold but has no systematic upward or downward trend. In spring, however, cloudy conditions begin to dominate, causing temperatures to warm on average and move the ice closer to its melting temperature, even before the newly risen Sun is strong enough to matter. Thus, more persistent clouds as spring approaches may cause the sea ice to first reach its melting temperature at an earlier date, and more frequent Arctic clouds in a warmer climate might accelerate sea ice decline.

Thus, models that simulate the seasonal cycle of Arctic cloudiness incorrectly may also predict the wrong time of onset and duration of sea ice melting, perhaps explaining some of the spread in model predictions of the future and their overall underestimate of sea ice decline in recent decades. Also, to date the models have been evaluated almost exclusively by asking whether they can reproduce the average conditions in the Arctic. If they do so by simulating near-average conditions most of the time, they are getting the right answer for the wrong reason, and their predictions of future sea ice decline should be discounted.


Stramler, K., A.D. Del Genio, and W.B. Rossow, 2011: Synoptically driven Arctic winter states. J. Climate, 24, 1747-1762, doi:10.1175/2010JCLI3817.1.