Climate Feedback

Reconciling sea ice models with reality

ngeo467-f1.jpgThe Arctic has been losing summer sea ice fast. At the end of the melt season in September 2008 the ice extent was barely above its record low of 2007, even after a much cooler summer. Most climate models did not anticipate the pace of this ice disappearance and still can’t replicate it. And as shown in the figure above, the futures projected by models are all over the map.

But researchers are reporting this week that there may be a way around the problems with modelling sea ice. With a new method for using observations to constrain climate projections, they find that the summertime ice is going, going, and set to be gone well before 2100.

The study was published in Nature Geoscience (subscription) by Julien Boé and his colleagues at UCLA. They calculate that the models showing the fastest future ice loss are also the most realistic about the recent past – something you could guess at by looking at the graph above, where the actual ice trend is shown as a black line. So rather than simply averaging across models, their technique teases out the most likely future sea ice evolution using the differences between the models that show the 1979-2007 ice extent most accurately and those that perform less well at reproducing past trends (if you’ve a yen for statistics it’s worth checking out the paper itself – the method is explained in a neat graphic). They project, based on a moderate emissions scenario, that September sea ice will most likely disappear between 2066 and 2085. That is sooner than even the most pessimistic models suggest on their own (see graph again).

But another of their results suggests that the ice could be gone even before then. Arctic sea ice in September 2008 was probably thinner than scientists had ever seen; that is, even though the area of the ice was smaller in 2007, the volume was probably less in 2008, according to the National Snow and Ice Data Center. And when Boé et al. include the thin, vulnerable ice of 2008 in the period of past data used to filter the models, they see summer sea ice vanishing as soon as 2059-2078. It’s a rather uneasy feeling, watching that date advance as the most recent ice volume dwindles.

Anna Barnett

Image: The percentage of remaining sea-ice cover in September, as observed (black line) and as simulated (coloured lines) by 18 climate models. Reproduced from Boé et al. Nature Geoscience doi:10.1038/ngeo467 (2009).


  1. Report this comment

    Malcolm said:

    Over the past year I have read various scientific articles and scientific opionions that the Artic sea ice would be gone by






    2008 (by a NSIDC scientist, year now gone, but not the ice)

    This new study estimate is between 2066 and 2085.

    You see the problem, the scientific message on this matter is totally confusing.

Comments are closed.