A Spring Predictability Barrier for Arctic Sea Ice

The dramatic changes to sea ice in the Arctic has created significant interest surrounding its predictability on seasonal-to-interannual timescales. Of particular interest to many stakeholders is the prediction of the summer sea ice extent minimum. Some studies suggest that forecasts of this minimum are potentially skillful beyond 12 months in advance, while other studies have found a predicability barrier in the spring season. This barrier suggests that there may be a sharp limit on skillful predictions of the summer sea ice extent minimum, as prediction skill for lead times beyond 3-5 months fall of dramatically. Here, we explore this emerging idea by examining an ensemble of general circulation models (GCMs) from phase 5 of the Coupled Model Intercomparison Project (CMIP5). Using the preindustrial control runs of 37 different GCMs, we show that this spring predicability does indeed exist across the CMIP5 ensemble and is most pronounced in the Kara, Laptev, East Siberian, Chukchi, and Beaufort Seas. In these regions, correlation values between sea ice volume and September sea ice area drop off significantly in the months preceding June and May. Using these results, we discuss implications for operational predictions of the summer sea ice extent minimum.