The Atlantic meridional overturning circulation (AMOC)—the zonal integral of the oceanic meridional flow—has become a touchstone of climate change discussions. Using a two-decades long least-squares fit of an ocean/ice general circulation model to almost all of the global data sets now available (the ECCO Project), it is possible to estimate AMOC mean values and variability. A number of simple conclusions emerge:
(1) Over the last 19 years, the AMOC is indistinguishable from a stationary Gaussian random process, at almost all latitudes.
(2) Its variations are correlated with latitude across the tropics and subtropics, but are uncorrelated (with 19 years of record) between the subpolar and subtropical gyres.
(3) Observed extreme events in its magnitude are consistent with easily computed (Rice) statistics and the Gaussian stationary assumption.
(4) Explicitly computed predictability exists at some latitudes for a few years. Whether the estimated skill level is of any use is less clear.
(5) The Gaussian nature of the fluctuations is consistent with expectations derived from the central limit theorem, and the long list of reasons why one would expect the integrated transports to vary (including wind and buoyancy-exchange fluctuations, internal instabilities, boundary current variations, disturbances emanating from other oceans).
(6) The AMOC is a very limited test of model skill, as it integrates through a host of distinguishable dynamic regimes …