Comparing Simulated and Observed AMOC Transport Estimates

Abstract:
Joint and consistent analysis of model simulations and observations of the Atlantic Meridional Overturning Circulation (AMOC) is important to advance our understanding of its trends, variability, and mechanisms as well as its latitudinal coherency. It is important that such model – observations comparisons use the same or as-close-as-practically-possible methods as applied in observations to provide apples-to-apples comparisons. Considering the Meridional Overturning Variability Experiment (MOVE) array at 16°N, the RAPID Array at 26.5°N, the South Atlantic MOC Basin-wide Array (SAMBA) at 34.5°S, and the Overturning in the Subpolar North Atlantic Program (OSNAP) array at 57°N, we revisit the underlying assumptions used to calculate their respective transports in comparison to transports from an eddying ocean hindcast simulation where transports obtained using the same observational methods can be directly compared against model truth. Our analysis reveals significant sensitivity of simulated and observed transports and their variability and trends to various reference level assumptions. Indeed, given i) the short observational records, ii) strong dependencies of transport estimates to their barotropic / compensating components, and iii) the presence of large interannual-to-decadal timescale variability, whether AMOC has been declining or not during the last couple of decades cannot be meaningfully determined. In general, simulated variability is weaker than in observations, particularly for the 9-site SAMBA estimate. We show reasons for these differences, considering contributions of transport components and the underlying transport assumptions associated with the observational methods. We also discuss complementarity of depth- and density-space AMOC depictions.