Bi-polar freshwater flux see-saw anomalies impacting the north atlantic thermohaline circulation

We present three scenarios using a global 1/8 degree eddy-permitting ocean, land and sea-ice general circulation model forced with hourly MERRA GEOS-5 atmospheric data from 1992-2014. In all cases, dense water production in the northern Labrador and Norwegian Seas is fairly well represented during the first decade of the simulation. In the reference case, after the turn of the century, surface haloclines pervade these dense water formation regions, and the overturning of deep North Atlantic watermasses becomes unrealistically weak. In a second case, we attempted to de-stabilize the overturning circulation by reducing precipitation North of 40N. The AMOC index at this latitude increases by roughly 23% for a 15% decrease in net North Atlantic freshwater forcing.

A freshwater flux adjustment is used in order to maintain model sea-ice concentrations closer to observations. Of the 0.54 Sv of Atlantic freshwater forcing between 40N and Bering Strait in the reference simulation , 27% is externally applied in order to insulate sea-ice from warmer underlying watermasses, mainly on the continental shelves.

In a third experiment, static Antarctic ice shelves are thermodynamically coupled to the ocean. Meltwater injection, in ice shelf cavities, facilitates sea-ice production on nearby continental shelves. The amount of external freshwater required to maintain warm season Antarctic sea-ice reduces by 26% to 0.17 Sv in the presence of a time-average meltwater contribution of 0.06 Sv. Significant changes occur on most high-latitude shelf regions, including in the North Atlantic. Inter-hemispheric propagation of coastally-trapped anomalies results in increased ventilation near deepwater formation sites in the Norwegian and Labrador Seas, and a roughly 20% increase in time-average AMOC strength at 40N, relative to the reference case, for a less than 10% increase in freshwater forcing south of 60S.

These case studies demonstrate strong oceanic teleconections between the poles, as well as the degree to which freshwater forcing bias, including contributions from glaciers, may be limiting our ability to accurately simulate AMOC in general circulation models.