North Atlantic Ocean Internal Decadal Variability: Role of the Mean State and Ocean‐Atmosphere Coupling

The origin of the decadal variability in the North Atlantic Ocean is investigated in a series of coupled and ocean‐only numerical experiments. Two versions of the IPSL‐CM5A model are considered, differing only by their atmospheric horizontal resolution (3.75° × 1.87° and 2.5° × 1.25°). When the ocean model is forced by the climatological surface fluxes from the low atmospheric resolution coupled model version, a 20‐year variability emerges, similar to the variability found in the coupled simulation. Such decadal variability is consistent with a large‐scale baroclinic instability of the mean flow in the west European basin. Increasing the atmospheric resolution leads to a more intense Icelandic low, which intensifies the western subpolar gyre, and warms the eastern North Atlantic subpolar gyre region. The mean state changes nearly vanish the associated internal oceanic variability under the corresponding climatological surface fluxes. Increasing the atmospheric resolution also produces a slightly weaker atmospheric stochastic forcing. Both the mean state and atmospheric variability changes are consistent with the decreasing amplitude of the variability in the coupled model. For both model versions, the amplitude of the internal oceanic variability is strongly enhanced in the presence of atmospheric stochastic forcing. Air‐sea coupling on the other hand has a moderate influence on the amplitude of the variability only in the low‐resolution model version, where the North Atlantic oceanic variability at 20 years increases by 23% due to coupling. The coupling effect is therefore modest and sensitive to the atmospheric horizontal resolution. Key Points The AMOC and AMV decadal variability can emerge from a self‐sustained oceanic mode based on coupled model results Increasing atmospheric resolution decreases the oceanic variability due to mean state changes Atmosphere‐ocean coupling causes a moderate increase of the AMOC variability