Dynamics responsible for the AMV depends on region

There is intense debate about the relative roles of external forcing, stochastic atmospheric variability, and ocean dynamics (e.g., the AMOC) in forcing the Atlantic Multidecadal Variability (AMV). We suggest that the AMV cannot be determined solely by either the atmosphere or ocean; instead the variability arises from different dynamical processes depending on the region. Using gridded observations and CMIP5 models, we construct empirical models of SST variability in order to determine what portion of the AMV can be explained by dynamical modes of variability (i.e. the North Atlantic Oscillation, El Nino Southern Oscillation, etc.). To isolate processes that are important in different regions, we examine the degree to which these modes explain SST variability in the subpolar, subtropical, and tropical Atlantic. In our empirical model framework, we find that most SST variability in the tropical and subtropical North Atlantic can be explained as the response to dominant atmospheric modes of variability, but other processes (i.e. ocean dynamics) contribute to SST variability in the subpolar North Atlantic. While SST variance is underestimated in models compared to observations, the relative importance of different dynamical processes in different regions is broadly consistent between the models and observations.