| Popis: |
Arctic sea ice loss in recent decades has been proposed to influence atmospheric circulation at lower latitudes, producing feedbacks that amplify ice loss via thermodynamic and mechanical forcing. One proposed teleconnection pathway arises from autumn Barents-Kara sea ice reduction and leads to a negative North Atlantic Oscillation (NAO) in winter. The existence of such a pathway could improve predictions of European weather on subseasonal to seasonal timescales. While autumn sea ice and the winter NAO are significantly correlated in satellite-era observations, this correlation appears to be absent in coupled climate models, calling into question the underlying mechanism. By subsampling long simulations to create satellite-length records, we find a small number of samples across a range of CMIP5 and CMIP6 models that reproduce the observed correlation. In these samples, we observe similar circulation signals (e.g., weakening of the stratospheric polar vortex) as in the observations, but there is no evidence for a driving role from sea ice changes via turbulent heat fluxes. Rather than sea ice, blocking of the atmospheric circulation by the Ural mountains appears to be the key precursor to the winter NAO signal. Overall, our findings reconcile differences between observations and models in representing this Arctic-midlatitude teleconnection, and highlight the important role of atmospheric internal variability in Arctic change. |
