The NAO was in its strong positive polarity from the 1960s to the mid 1990s and during this time sea-ice concentrations decreased in the Barents Sea and increased in the Labrador Sea. When we asked the question in Atmosperic Global Climate Model (AGCM) simulations, is there a feedback from this spatial pattern of change in sea ice back onto the NAO (or atmospheric circulation), we found a clear negative feedback in the equilibrium winter response. We have recently examined the transient response to this sea-ice forcing to determine what processes control the evolution to a negative NAO. We found that the initial modest circulation response from the change in surface fluxes allows a changed configuration of Rossby wave breaking and it is the latter effect that leads to the more prominent and larger scale (equilibrium) response of a negative NAO. Thus internal (or natural) variability indirectly sets the stage for the prominent response to a changed sea ice distribution. It is the interaction of the short time-scale internal variability with the forced initial response that sets the stage for the evolution of the amplified large-scale change.
We are now entering an unchartered era in sea-ice variability. In addition to the NAO related mode of sea-ice variability (the Labrador-Barents Sea dipole), rapid anthropogenic sea-ice loss, even in winter, is an even more prominent mode of variability. Sea-ice observations are beginning to show this effect, but the clearest signature may be seen in climate model projections. Interestingly, the climate model projection show the NAO related dipole of variability as the second leading mode, the overall sea-ice decline is the first leading mode.
In this research we seek to identify, understand and quantify the dynamical feedback processes between 1)the atmospheric circulation, 2)sea-ice concentrations and 3)the oceanic heat flux, from observations and a hierarchy of numerical models with the ultimate goal of facilitating prediction of North Atlantic climate on interannual to decadal timescales. The models range from linear stochastic equations linking the NAO index and sea-ice concentration, to AGCMs, to coupled (atmosphere, sea ice, ocean) climate models with a simplified ocean that allow for easier identification of processes, to output from fully coupled state of the art climate models. With coupled reanalysis products on the horizon, the research is timely and holds great potential in the quest for decadal prediction of climate.