Multiyear predictability of Northern Hemisphere surface air temperature in the Kiel Climate Model

The multiyear predictability of Northern Hemisphere surface air temperature (SAT) is examined in a multi-millennial control integration of the Kiel Climate Model, a coupled ocean–atmosphere–sea ice general circulation model. A statistical method maximizing average predictability time (APT) is used to identify the most predictable SAT patterns in the model. The two leading APT modes are much localized and the physics are discussed that give rise to the enhanced predictability of SAT in these limited regions. Multiyear SAT predictability exists near the sea ice margin in the North Atlantic and mid-latitude North Pacific sector. Enhanced predictability in the North Atlantic is linked to the Atlantic Multidecadal Oscillation and to the sea ice changes. In the North Pacific, the most predictable SAT pattern is characterized by a zonal band in the western and central mid-latitude Pacific. This pattern is linked to the Pacific Decadal Oscillation, which drives sea surface temperature anomalies. The temperature anomalies subduct into deeper ocean layers and re-emerge at the sea surface during the following winters, providing multiyear memory. Results obtained from the Coupled Model Intercomparison Project Phase 5 ensemble yield similar APT modes. Overall, the results stress the importance of ocean dynamics in enhancing predictability in the atmosphere.