Green.view

Thinning on top

Predicting the Arctic’s summer fate is not so simple

MARCH is the maximum month. In March 1979, the first year for which satellite records are available, the sea ice in the Arctic Ocean covered 16.4m square kilometres (6.4m square miles). By the time summer was gone, the ice was down to about 7.2m square kilometres. Every March since, the ice has returned to an annual maximum, but never again has it been as large as the one observed that first year. Every summer, it shrinks back down, and the minimum has for the most part been getting smaller and smaller. In 2007 it was just 4.3m square kilometres. At first, polar scientists watched the planet’s icy pulse with academic interest and justifiable pride in their new observational capabilities. Now they monitor its every hiccup as if it were a patient on life support. But that does not mean they know how to interpret what they see.

This year they noted that the Arctic ice cap had a late-winter growth spurt, reaching its maximum on March 31st, the latest date ever recorded. By the beginning of June, the ice cover—as defined by the percentage of ocean covered by at least 15% floating ice—had dropped far below what is usual for the time of year. Both observations, by themselves, sound as if they should contain meaningful information about what to expect in the rest of 2010. They don’t.

Or at least, if they do, no one knows how to extract it. Every month between June and September, the environmental arctic change programme of America’s Arctic Research Consortium releases a “sea ice outlook” in which teams of researchers estimate how much ice will be left by September. The 2010 numbers are expected to become public in the next week or two.

In 2009 more than a dozen teams took on the challenge, and all of them failed to one degree or another. The group that came the closest to predicting the September sea-ice minimum (which was 5.36m square kilometres) was the one from the University of Washington, which had the figure chalked in as 5.2m square kilometres, overestimating the extent of the melting by an area five times that of Belgium. The University of Washington’s team made that prediction by “heuristic reasoning”—aka educated guesswork. The teams that relied more on modelling the processes involved, or on statistics from years gone by, did worse. All the teams overestimated the ice loss, some of them by more than 1m square kilometres.

Will it ever be possible to model Arctic sea-ice decline accurately? One bedevilling factor is that weather plays an outsize role in sea-ice conditions. A couple of high winds here, an abnormal cold snap there, and ice cover that had looked fragile can suddenly appear robust. Last summer the weather in August and September was dramatically different from that in June and July, when the initial predictions were made, which probably partly explains why everyone overestimated the extent of the melt. The record-breaking low in 2007 came in large part from a pattern of atmospheric highs that settled over the central Arctic, with winds pushing ice away from the Siberian coast and sending warmer temperatures coursing northward. Weather patterns such as this simply cannot be understood well enough months in advance to help inform sea-ice predictions.

Many other basic questions about ice also remain unanswered. Warmer temperatures mean more ponds of meltwater on the ice’s surface. That alters the ice’s reflectiveness, but the consequences of that alteration are not really understood. Thick ice that has survived multiple summers, and thinner, single-year ice respond differently from one another. An increase in the ratio of thin ice to thick ice seen recently has led Mark Serreze, head of the National Snow and Ice Data Centre in Boulder, Colorado, to suggest that the dynamics of the sea-ice system are changing in a fundamental way. If so, that might explain why past statistics are not yielding future predictions. But little comparative research on this has yet been done. And almost no one has a handle on how changing cloud conditions—expected as the Arctic warms and the proportion of ice and water shift—are affecting the rate of melting.

Acknowledging all the drawbacks, Marika Holland, a sea-ice modeller at the National Centre for Atmospheric Research in Boulder, says that ice predictions today are like those for ozone-hole severity or El Niño patterns a decade or two ago—of limited use, but with promise. Now, after years of refinement and recognition of their limitations, ozone and El Niño predictions have found their place in providing at least some helpful information to governments and other decision-makers. El Niño studies, for instance, can identify fairly reliably when the distinctive climate shifts caused by warm water sloshing east across the central Pacific are imminent, but cannot predict their intensity.

Perhaps in a decade scientists may be able to help shipping companies who want to know how early in the year they can expect to send freighters through a seasonally ice-free Arctic. Already, the prediction effort has produced a small-scale product for use by native hunters in Alaska: a regular “walrus outlook” projecting coastal sea-ice conditions for the upcoming week. And by focusing on what does and does not affect the ice from spring to summer, it may provide an insight into the bigger question of the ice’s long-term future—on the rate of the long-term decline that lies beneath the year-to-year noise. Few now disagree that, at some point this century, the Arctic will start to see ice-free summers. But estimates of when that will happen are as scattered, or more, than the yearly predictions. Some like 2050, some like 2030, some think it will be gone by 2020.

While waiting for elucidation on that, any would-be predictors who think this summer will be a hot one with melt-inducing winds should head over to Stoat, a blog run by William Connolley, who used to work for the British Antarctic Survey. Every year since 2007 he has been willing to take bets from all comers that the ice will not retreat past the 2007 minimum. If he keeps this up then it seems all but certain that he will be wrong one year; if you think this is the year, then it’s time for a bet. But best not pretend you are backing anything but a hunch.