Séminaire
Understanding the preferential timescales of variability in the North Atlantic, usually associated with the Atlantic meridional overturning circulation (AMOC), is essential for decadal prediction. However, the wide variety of mechanisms proposed from the analysis of climate simulations, potentially dependent on the models themselves, has stimulated the debate of which processes take place in reality. One mechanism receiving increasing attention, identified both in idealized models and observations, is a westward propagation of subsurface buoyancy anomalies that impact the AMOC through a basin-scale intensification of the zonal density gradient, enhancing the northward transport via thermal wind balance. In this study, we revisit a control simulation from the IPSL-CM5A-LR coupled model, characterized by a strong AMOC periodicity at 20 years, previously explained by an upper ocean-atmosphere-sea ice coupled mode driving convection activity south of Iceland. Our study shows that this mechanism interacts constructively with the basin-wide propagation in the subsurface. This constructive feedback may explain why bi-decadal variability is so intense in this coupled model as compared to others.
Additionally, we also follow a perfect model approach to assess the performance of several nudging techniques (restoring towards different surface variables) to initialize the bi-decadal variability in the model. A specific focus is made on the representation of an extreme positive AMOC peak in the original control simulation used as "surrogate reality". All standard nudging approaches considered succeed in reproducing the timing of the extreme peak, but underestimate its amplitude. A careful analysis of the AMOC precursors previously identified reveals that this underestimation comes from a deficit in the formation of the dense water masses in the main regions of convection. This issue is largely corrected in an improved nudged simulation that uses a varying restoring term, proportional to the mixed layer depth. This development improves the restoring of surface temperature and salinity in the regions of convection, and eventually the representation of AMOC variability, preventing unphysical restoring fluxes elsewhere. This is therefore a promising nudging strategy that applied to the real world can help to better constrain the recent AMOC variability over the last few decades.