And now there's more bad news: the early models didn't include warming-induced processes that could boost temperatures further, such as methane release from melting permafrost and growth of new vegetation that lowers albedo. As if that wasn't enough, it looks like a couple of biases in the models mean that they may be underpredicting climate warming.
Currently models indicate that cloud cover over the Arctic in summer will increase as the planet warms. That would have the effect of reflecting back some of the sun's rays and dampening warming, said Kay. But in a forthcoming paper Kay has correlated observations of sea ice for 2006 and 2007 with the actual and modelled cloud responses. The model projected an increase in low cloud cover in all the areas where sea ice was lost. But in reality cloud cover only went up in a portion of the region, so the dampening effect on climate change would be less.
In the winter, meanwhile, temperature inversions - when the air nearest the Earth becomes colder than the air above - are common in the Arctic. These temperature inversions have a cooling effect. Alex Hall of the University of California Los Angeles has found that climate models that use stronger temperature inversions tend to predict less warming for the Arctic Ocean. But recent observations in the field indicate that a lot of models are using temperature inversions that are too strong - the measured values were at or below the bottom end of the range used by climate modellers.