Resolution-dependent interannual Gulf Stream, heat content, and AMOC variability in the CMIP5

Low-resolution (0.25 degrees) OGCM simulations often show significant biases in North Atlantic temperature mean and variability. Nearshore mean temperatures tend to be up to 5 degrees too high, and temperatures in the North Atlantic Current region tend to be around 5 degrees too low. The position of the Gulf Stream is also highly meridionally variable on time scales interannual and longer, in contrast to altimetric observations. Higher-resolution (i.e. eddy-permitting) simulations resolve various baroclinic instability processes and tend to evince a stronger, meridionally stable Gulf Stream. This allows for more effective large-scale transport and tends to ameliorate these biases. The effect of increased model resolution on Gulf Stream variability, as well as its forcing of AMOC is not yet well understood.

Here we use a simple metric for determining Gulf Stream position within model output. With this method, we find that many of the members of the CMIP5 evince strong meridional variability in Gulf Stream position as compared to observations. Employing a time-dependent position index, we also find that within these members, this enhanced variability dominates North Atlantic horizontal and vertical temperature exchanges on time scales interannual and longer. In contrast, a coupled GCM run at eddy-permitting resolution (0.1 degrees) shows virtually no large-scale monotonic meridional Gulf Stream position variability. Basin-scale temperature variability is characterized by an integrated response to atmospheric forcing. 

This work underlines the importance of model resolution in effective simulation of Gulf Stream and North Atlantic interannual and decadal variability, especially in determining AMOC variability and change in historical simulation.