Predicting multi-year North Atlantic Ocean variability

24 We assess the skill of retrospective multi-year forecasts of North Atlantic ocean 25 characteristics obtained with ocean-atmosphere-sea ice models that are initialized with 26 estimates from the observed ocean state. We show that these multi-model forecasts can 27 skilfully predict surface and subsurface ocean variability with lead times of 2 to 9 years. We 28 focus on assessment of forecasts of major well-observed oceanic phenomena that are thought 29 to be related to the Atlantic meridional overturning circulation. Variability in the North 30 Atlantic subpolar gyre, in particular that associated with the Atlantic Multidecadal 31 Oscillation, is skilfully predicted 2-9 years ahead. The fresh water content and heat content in 32 major convection areas such as the Labrador Sea are predictable as well, although individual 33 events are not captured. The skill of these predictions is higher than that of uninitialized 34 coupled model simulations and damped persistence. However, except for heat content in the 35 subpolar gyre, differences between damped persistence and the initialized predictions are not 36 significant. Since atmospheric variability is not predictable on multi-year time scales, 37 initialization of the ocean and oceanic processes likely provide skill. Assessment of 38 relationships of patterns of variability and ocean heat content and fresh water content shows 39 differences among models indicating that model improvement can lead to further 40 improvements of the predictions. The results imply there is scope for skilful predictions of 41 the Atlantic meridional overturning circulation. 42 43