19.11.2018: FB1-Seminar

11:00 Uhr, Hörsaal, Düsternbrooker Weg 20

The atmosphere and oceans are densely populated by non-linear, turbulent processes that strongly impact the transport of heat, carbon etc. in the climate system. In the atmosphere, these processes can take the form of large weather systems on spatial scales of ∼ 1000 km. The oceanic equivalent have smaller spatial scales of 10 − 100 km and are known as mesoscale eddies. When the Navier-Stokes equations are discretised onto a grid of finite resolution and get a finite time step, budgets of momentum, energy, etc. are truncated at given spatial and temporal scales. For a climate model to adequately represent oceanic mesoscale eddies would require a horizontal resolution of < 10 km, which is computationally prohibitive, even with the world’s largest supercomputers. Recent state-of-the-art climate models use a horizontal resolution of ∼ 30 − 100 km, thus partly resolving mesoscale eddies.

Because the momentum and energy budgets are truncated, numerical ocean models use parameterisations that mimic the statistical behaviour of small-scale processes and their impact on the ocean flow. A common choice is diffusion with either a Laplacian or Bi-Laplacian operator and a constant viscosity coefficient. Such parameterisations remove turbulent energy at the smallest resolved scales, which is unrealistic as we know that some of that energy feeds back to the larger-scale flow. Thus, if the smallest resolved scale is ~30 km, much of the mesoscale eddy energy will be lost instead of interacting with the resolved flow.

In this talk, I will present a new method to parametrize mesoscale eddies: modelling the eddy stress tensor. When applied to Lagrangian particles traced in the Southern Ocean, the parametrization increases the kinetic energy at all resolved spatial scales, increases dispersion rates, and increases horizontal diffusivity. When the parametrization is applied to an eddy-permitting ocean model, ORCA025, the resulting flow becomes more similar to the flow in a high-resolution model, ORCA0083. The results strongly indicate that implementing the parametrization into a numerical ocean model, would greatly improve its realism, especially in eddy-rich regions such as the Gulf Stream or Southern Ocean.