Coupled modes of Barents sea ice and atmospheric anomalies in autumn and winter

Recent studies report linkages of late autumn Barents-Kara sea ice and winter atmospheric circulation anomalies but it is unclear what the zero-lag feedbacks and higher coupled modes do. Here, we revisit the problem using Maximum Covariance Analysis (MCA), examining not only the first modes but also higher modes that contribute significantly to the squared covariance fraction (SCF). Besides sea ice concentration, SLP, and geopotential height, we also analyse heat advection and surface temperature to elucidate the climate impact on Europe. In reanalysis/observational data, the first two atmospheric modes at zero lag advect cold/warm air southward/northward over the Arctic-Atlantic sector, and are associated with positive/negative sea ice anomalies. Additionally, the atmospheric modes at zero lag can cancel or reinforce, depending on the regions, the lagged atmospheric responses to sea ice reported previously. Applying the same analysis to coupled model simulations, we find that models can reproduce the first MCA modes, but they are far too dominant (SCF > 60%) compared to those in observational data (SCF ~ 40%). This suggests that coupled modes in models are not as diverse as in the real world, and that the lagged coupled modes are underestimated by models. By considering all the leading modes, we can have more complete understanding of the lagged and zero-lag feedbacks. Additionally, examining MCA beyond the first modes can be used to assess coupled modes simulated by models.