Three Atmospheric Patterns Dominate Decadal North Atlantic Overturning Variability

Atlantic Meridional Overturning Circulation (AMOC) variability originates from a large number of interacting processes with multiple time scales, with dominant processes dependent on both the latitude and timescale of interest. Here, we isolate the optimal atmospheric modes driving climate-relevant decadal AMOC variability using a novel approach combining dynamical and statistical attribution (dynamics-weighted principal component, or DPC analysis). We find that for both the subpolar (55 degrees N) and subtropical (25 degrees N) AMOC, the most effective independent mode of heat flux forcing closely resembles the North Atlantic Oscillation, and drives meridionally coherent AMOC anomalies through western boundary density anomalies. Conversely, established modes of wind stress variability possess limited quantitative similarity to the optimal wind stress patterns, which generate low-frequency AMOC fluctuations by rearranging the ocean buoyancy field. We demonstrate (by running a modified version of the ECCOv4r4 state estimate) that most AMOC variability on decadal time scales can be explained by the DPCs. The heat-redistributing north-south "overturning" circulation of the Atlantic Ocean is uniquely influential on climate, driven in part by chaotic atmospheric patterns of heat and wind that impart unpredictable variability into its strength. This study uses a novel modeling technique that combines the statistics of historical atmospheric data with the physics of the overturning circulation to uncover the distinct atmospheric patterns that most effectively stimulate this variability. We deduce that slow fluctuations in the overturning circulation are best forced by heat inputs that follow the shape of the well-known North Atlantic Oscillation pattern. However, we also uncover two new patterns of wind (one most effective nearer the equator, one nearer the Arctic) that lead to a strong, delayed response in the overturning circulation. Removing these atmospheric patterns from the a simulation of recent ocean history results in a substantial drop in the intensity of overturning fluctuations. The leading surface patterns forcing decadal AMOC variability are identified using a hybrid dynamical--statistical attribution technique The leading heat flux pattern is independent of AMOC latitude, drives up to 81% of AMOC variability, and is closely related to the NAO Leading wind stress patterns forcing AMOC are distinct from the NAO and differ for subtropical and subpolar AMOC