How Soon Will We See a Sea-Ice-Free Summer Arctic?

Current studies report a range of timing for a nearly sea-ice-free Arctic from 2012 to late 21st century. We examine the assumptions behind these estimates. When the IPCC AR4 Report was published in early 2007, the projected date for loss of summer sea ice cover for the Arctic was near the end of 21st century. Boe (UCLA) et al. also project an ice-free Arctic before the end of the century. The low sea ice extent and reduced amounts of old/thick sea ice observed in the summers of 2007, 2008 and 2009, below the envelope of the IPCC climate model projection ensemble results, support the hypothesis that the real world is on a fast trajectory for sea ice loss. Single ensemble members from the NCAR CCSM3 model suggest a timing of loss before mid-century. Using a subset of IPCC models which were selected based on observational constraints (mean and seasonality), and taking the sea ice extent from the summer of 2007/2008 as initial conditions, we have suggested a timing of 2037 for a nearly sea ice free Arctic, with a standard deviation of 10 years. Results from a regional scale model (Naval Postgraduate School) with high resolution ocean dynamics support an early sea ice free summer (2012) based on the role of ocean transport. Satellite observations also can support an early loss by 2013 (Zwally, NASA). However, data from ice mass buoys (Perovich) suggest that it will be difficult to remove sea ice from the vicinity of the North Pole in the immediate future. Given the evidence of the importance of ocean heat transport from summer/fall 2009 data, we now suggest a timing for sea ice loss before our previous 2037 estimate. Late century loss estimates tacitly assume that major individual events such as 2007 are random, occurring on an anthropogenic downward trend. In contrast, if sea ice changes from the 2007 event lead to a one way feedback, loss could occur before mid-century. Modification of ocean currents and heat storage could produce even earlier sea ice loss, as their impacts are not well included in IPCC type climate models. A recent NCAR model study shows that a thinner sea ice regime exhibits less predictability than a thicker sea ice regime (Holland), so that uncertainty in the timing of sea ice loss over the next decades will remain high.