Observed and simulated variability of the Atlantic Meridional Overturning Circulation at 41°N

Abstract

The Atlantic Meridional Overturning Circulation (AMOC) at 41°N from a global 1/16° eddying simulation is compared with ARGO-based transport estimates over the 2004–2013 period. Three different methods for calculating the modelled meridional transports are used. The first method (MOCmod) is simply based on simulated velocity fields. The second method (MOCob) is based on the same hydrostatic and geostrophic relationships applied to ARGO observations, and the third (MOCob2) relies on the same assumptions, but does not use a reference depth of known motion. MOCmod and MOCob2 methods correctly reproduce the time-mean AMOC strength, while the MOCob result is ~7% weaker.The comparison of the three overturning calculations demonstrates that ignoring transports near the western boundary (ARGO floats are restricted to ocean regions deeper than 2000m) leads to the seasonal cycles of the non-Ekman component of the AMOC from the model and observation to be out of phase. Due to a lack of ARGO data and the consequent use of extrapolation/average processes near the western boundary, uncertainties exist in the definition of density field near the western boundary, which can enlarge discrepancy between modelled and observed variability.Furthermore, the meridional covariability of the modelled AMOC at 26.5°N and 41°N is analysed and compared to the covariability of the Rapid Climate Change programme and the ARGO-based time series. Similar to other model comparisons, the model output shows covariability between the two latitudes at some frequency bands, while the phasing differs for the observed data.