Asymmetries between phases of Atlantic Multi-decadal Variability in the CMIP6 multi models

The Atlantic Multidecadal Variability (AMV) is a phenomenon in which North Atlantic Sea Surface Temperature Anomalies (SSTAs) occur almost simultaneously in the subpolar and tropical regions, imprinting their impact not only on neighboring countries but also on the global climate system.Due to its long lifespan, the natural variability associated with AMV seriously amplifies the uncertainty of future climate projections, as the exact mechanisms of the AMV remain unknown despite numerous previous studies.In this study, we investigate the asymmetry in two opposite phases of AMV in different models using preindustrial control experiments from 46 different models participating in the Coupled Model Intercomparison Project 6 (CMIP6). Overall, we find a well-fitted positive linear relationship for tropical Atlantic SSTAs with respect to subpolar SSTAs among 46 models. However, when investigating the model sensitivity between two opposite AMV phases in each model, we find that the strength and phase preference in terms of the tropical SSTA sensitivity to subpolar SSTA widely vary, resulting in AMV+ preferred groups, AMV- preferred groups, or symmetric AMV groups.Among the three groups, the characteristics of models in the AMV+ preferred group are found to be most distinctive. It is most notable with the AMV+ preferred models that the net surface heat flux in the subpolar Atlantic adds heat from the atmosphere into the ocean during the positive AMV phase due to a robust hemispheric reduction of the Westerlies and the Trades.In contrast, it is clearly indicated with the AMV+ preferred model during negative phases of AMV, or with all other model groups during both AMV phases, that subpolar SSTAs associated with AMV originate from the ocean, rather than the atmosphere.This contrast in subpolar A-O interaction found in the AMV+ preferred model can be partially explained as the result of competition between subpolar and tropical SST influences, involving surface ocean feedback in the Tropical Atlantic. As the AMV+ positive group shows a significantly larger weakening of the westerlies and trade winds during AMV+, the vertical cold advection due to Ekman divergence becomes significantly weaker during positive AMV, resulting in warm SSTAs. In addition to the Wind-Evaporation-SST feedback, this Wind-upwelling-SST feedback associated with equatorial convergence further intensifies SSTAs and the tropical positive feedback. Further investigation reveals that the reason for the asymmetric AMV+ preference is in the nonlinear feedback mechanism: positive SST anomalies strengthen the stratification to help local warming driven by anomalous downwelling, whereas negative SST anomalies weaken the stratification and hinder local cooling driven by anomalous upwelling.