Resumen
The Atlantic Meridional Mode (AMM) is the dominant mode of interannual climate variability in the tropical Atlantic, maintained primarily by the positive wind?evaporation?sea surface temperature (SST) feedback in which the wind anomalies lead the SST anomalies by ~2 months. A previous study revealed that climate models from Coupled Model Intercomparison Project Phase 5 (CMIP5) show poor performance in simulating the AMM-related wind?SST relationship, but the possible causes remain unclear. This study assesses the representation of the AMM and associated wind?SST relationship in the climate models from CMIP6. Results show that most of the CMIP6 models can reasonably reproduce the observed spatial pattern of the AMM, with significant SST and wind anomalies in the northern tropical Atlantic and weak anomalies in the equatorial?southern oceans. However, the simulated wind?SST relationship associated with the AMM varies among the models. In particular, several models fail to capture the observed wind?SST relationship; that is, the simulated wind anomalies peak in boreal spring as in the observations, but no obvious peak occurs in the corresponding SST anomalies. Further analysis suggests the models that fail to capture the observed wind?SST relationship tend to simulate a stronger mean trade wind and a thicker mixed layer in the northern tropical Atlantic, leading to a weaker ocean?atmosphere coupling and, thus, a weaker SST response to the wind forcing. Moreover, there exists a significant out-of-phase relationship between the strength of ocean?atmosphere coupling and mean mixed layer depth among the models, supporting the impact of mean state biases on the AMM variability in the models.