Item

Abstract

Climate models are known to suffer from various biases and uncertainties. In the subpolar North Atlantic typical biases among models from the Coupled Model Intercomparison Project phase 6 (CMIP6) are found in the mean surface temperature and salinity, and in the mean sea ice concentration, which can affect the air-sea interaction.In this study, we are investigating the diversity of CMIP6 models with respect to their response of the Atlantic Meridional Overturning Circulation (AMOC) to the North Atlantic Oscillation (NAO) in pre-industrial control experiments. This response is sensitive to the mean state of the North Atlantic. We focus on two categories of models: Models that are rather warm-salty versus models that are rather cold-fresh within the subpolar gyre of the North Atlantic. Warm-salty models tend to have a lower sea ice cover in the Labrador Sea (LS) and larger LS heat loss during a positive NAO, compared to cold-fresh models. They also have a weaker stratification in the LS. Sub-surface density changes 1 to 3 years after the NAO are larger in the warm-salty models and establish a zonal density gradient that can cause a stronger delayed AMOC response via the thermal wind balance.These findings stress the need for improvement of the North Atlantic mean state in climate models. Uncertain mean states might further contribute to the uncertainty in AMOC future projections.