ALBERT

Description: Highlights: • OGCM simulations of the AMOC are highly sensitive to the subarctic freshwater forcing. • Trends in the simulated AMOC are linked to the salinity of the DSOW. • DSOW salinity trends can be traced back to the freshwater transport by the NAC. • The NAC freshwater budget is highly affected by the salinity restoring used in OGCMs. • Modifications in the subarctic precipitation can help to minimize the restoring flux. Global ocean sea-ice models with an atmospheric forcing based on bulk formulations of the air-sea fluxes exhibit spurious trends in key flow indices like the Atlantic Meridional Overturning Circulation (AMOC), constraining their use in investigations of multi-decadal ocean variability. To identify the critical model factors affecting the temporal evolution of the AMOC on time scales of up to 60 years, a series of experiments with both eddy-permitting (0.25°) and non-eddying (0.5°) ocean-ice models has been performed, focusing on the influence of artificial choices for the freshwater forcing, in particular the restoring of sea surface salinity towards climatological values. The atmospheric forcing builds on the proposal for Coordinated Ocean-ice Reference Experiments (CORE), utilizing the refined atmospheric reanalysis products for 1948–2006 compiled by Large and Yeager. Sensitivity experiments with small variations in precipitation (within the observational uncertainty) and sea surface salinity restoring in the subarctic Atlantic produce a wide range of AMOC transports, between upward drifts to more than 22 Sv and nearly-collapsed states with less than 7 Sv, reflecting the excessive role of the salinity feedback in such simulations. In all cases the AMOC is tightly related to the density of the Denmark Strait overflow; changes in that density are governed by the salinity in the Nordic Seas; and in turn, that salinity is strongly affected by the properties of the inflowing North Atlantic water.