Representation of mesoscale in a hierarchy of Met Office ocean model configurations

Using an explicit representation or a parametrisation of the ocean mesoscale affects not only the mean state of the ocean but also the climate variability. However, the choice of resolution is constrained by computational costs. Ocean models developed at the Met Office are used for a vast range of applications from short-range coupled NWP forecasts to Earth system models. Shorter range predictions can run with higher resolution models while climate models are very constrained. To support all applications, we developed a hierarchy of three ocean models: eddy parametrising (1°), eddy-present (1/4°) and eddy-rich (1/12°) resolution models. In the eddy parametrising configuration, mesoscale eddies are not resolved. In the eddy-present configuration, the resolutions allow some mesoscale eddies to be captured in the low and mid-latitudes. In the eddy-rich configuration, eddies are present at most latitudes.

In the 1° configuration, the mesoscale turbulence associated with eddies cannot be solved explicitly. The lateral turbulent fluxes are assumed to depend linearly on the lateral gradients of large-scale quantities requiring second order operators. A lateral diffusion of momentum on geopotential surfaces with a Laplacian viscosity is used. The effect on the large scale is represented using Gent and McWilliams (1990) proposed parameterisation of mesoscale eddy-induced turbulence. In the 1/4° and 1/12° configurations, the more scale selective biharmonic operator is used. It ensures the stability of the model while not interfering with the resolved mesoscale activity. To improve the circulation and biases in the Southern Ocean, a weak GM parametrisation is added in the eddy-present and eddy-rich models.

30-year forced integrations with the model hierarchy are assessed. Global and large-scale temperature and salinity biases are similar across the resolutions.  The largest differences occur in regions with strong mesoscale activity (Western boundary currents, Southern Ocean). Eddy-present and eddy-rich models significantly improve the representation of the Western boundary currents, both in position and strength. Improving the Western boundary currents has large impacts on temperature and salinity biases.

We review the results of Moreton et al (2020) on eddies in the Met Office hierarchy of models. The surface properties of eddies in eddy-present and eddy-rich coupled models are evaluated using an eddy tracking algorithm on SSH anomalies. Results show that relative to eddy-present, eddy-rich resolution simulates more (+60%) and longer-lasting (+23%) eddies, in better agreement with observations. The representation of eddies in Western Boundary Currents and the Southern Ocean compares well with observations at both resolutions.  However, a common deficiency in the models is the low eddy population in subtropical gyres. Despite a grid spacing larger than the Rossby radius of deformation at high-latitudes, eddy-present resolution only allows for eddy growth in these regions a lower rate than seen in observations and eddy-rich resolution. The westward displacement of eddies in eddy-rich model (mainly in the Agulhas region) is increased compared to the eddy-present model. The size of eddies is found to be dependent on model grid resolution.