Stochastic GM+E: An energetically-informed stochastic backscatter scheme for ocean models

Global ocean models at resolutions that do not resolve mesoscale eddies lack variability, not just on scales that they cannot represent because they are below the grid scale, but also on resolvable scales. This research develops a backscatter parameterizations that increases variability on the resolved scales of a non-eddying model. The parameterization acts on the model's momentum equations, and sets the rate of backscatter, viz. the rate at which energy is injected to the resolved scales, proportional to the rate at which the Gent-McWilliams (GM) parameterization removes energy from the resolved scales. This models the physical process whereby mesoscales convert large-scale potential energy to kinetic energy, and then transfer that kinetic energy towards larger scales. These parameterization is implemented in the MOM6 ocean model, and results are presented on its impact in simulations at nominal 2/3-degree resolution. Stochastic GM+E acts primarily in the Southern Ocean, the North Atlantic Current, and the Kuroshio Extension, where it impacts SST variability and southern-hemisphere sea ice extent.