This time the Sea Ice Volume Series offers an interactive feature to explore sea ice reduction in 2012. Just hover your mouse over the months below to load the indicated map.

PIOMAS shows significant reduction in thickest ice category North of Greenland and the Archipelago in December. An overall thinning compared to December 2011 is also observable, but less surprising, because that's the trend.

While PIOMAS was not designed to make forward looking assumptions - it doesn't know anything about greenhouse gas emissions - it captures the development of sea ice retreat far better than any other GCM. Technical Note 91 from the Met Office explains this in detail:

Schweiger et al. (2011) compare PIOMAS ice volume estimates (see section 2.2 for a description) and ice volume estimates from the CCSM3 coupled climate model. They show that CCSM3 estimates of ice volume agree with PIOMAS over the period 1979-2006. The same conclusion is reached when HadGEM1 and HadGEM2-ES are compared against PIOMAS. However, since 2007 the PIOMAS volume estimates are lower than those from climate models.

Since 2007, the Arctic in the real world has been subjected to anomalous atmospheric forcing with years such as 2007 and 2011 where stronger than normal winds have displaced large volumes of ice (eg, Lindsay et al., 2009) and exposed larger areas of open water which can absorb more heat from the atmosphere. These Arctic circulation anomalies may have been driven in part by El Nino/La Nina (L’Heureux et al., 2008). It is therefore possible that the divergence between PIOMAS and the climate models since 2007 represents internal variability of the climate system. Further research and subsequent years’ observations will confirm whether recent years have been anomalous, or whether they are part of a trend that is underestimated by climate models.

More: Assessment of possibility and impact of rapid climate change in the Arctic.

A technicial remark: PIOMAS implements curvilinear coordinates, so to produce the maps above all data has been reprojected using this function matplotlib.mlab.griddata onto a 0.5°grid with a linear interpolation.