A weak AMOC in a cold climate: Causes and remedies for a bias in the low-resolution version of the UK Earth System Model

The UK Earth System Model (UKESM) is currently being developed by the UK Met Office and the academic community in the UK. The low-resolution version of UKESM has got a nominal grid cell size of 150 km in the atmosphere (Unified Model [UM], N96) and 1° in the ocean (NEMO, ORCA1). In several preliminary test configurations of UKESM-N96-ORCA1, we find a significant cold bias in the northern hemisphere in comparison with HadGEM2 (N96-ORCA025, i.e. 0.25° resolution in the ocean). The sea surface is too cold by more than 2 K, and up to 6 K, in large parts of the North Atlantic and the northwest Pacific. In addition to the cold bias, the maximum AMOC transport (diagnosed below 500 m depth) decreases in all the configurations, displaying values between 11 and 14 Sv after 50 years run length. Transport at 26°N is even smaller and hence too weak in relation to observed values (approx. 18 Sv). The mixed layer is too deep within the North Atlantic Current and the Kuroshio, but too shallow north of these currents. The cold bias extends to a depth of several hundred metres. In the North Atlantic, it is accompanied by a freshening of up to 1.5 psu, compared to present-day climatology, along the path of the North Atlantic Current. A core problem appears to be the cessation of deep-water formation in the Labrador Sea. Remarkably, using earlier versions of NEMO and the UM, the AMOC is stable at around 16 or 17 Sv in the N96-ORCA1 configuration. We report on various strategies to reduce the cold bias and enhance the AMOC transport. Changing various parameters that affect the vertical mixing in NEMO has no significant effect. Modifying the bathymetry to deepen and widen the channels across the Greenland-Iceland-Scotland sill leads to a short-term improvement in AMOC transport, but only for about ten years. Strikingly, in a configuration with longer time steps for the atmosphere model we find a climate that is even colder, but has got a more vigorous maximum AMOC transport (14 Sv instead of 12 Sv). Conversely, if the isopycnal diffusivity is augmented by a factor of 1.5, we find a warming and an even weaker AMOC transport. This brings us to further strategies to modify the atmosphere-ocean fluxes of heat and freshwater.