Ian Grooms, Department of Applied Mathematics, University of Colorado Boulder

Backscatter and other hazards of ocean model development

Standard practice in scientific computing is to write some physical quantity of interest as the solution of a PDE, and then to find an accurate solution of the PDE using numerical methods. This is not possible with climate, neither now or in the future, for two reasons: we do not know enough to formulate the problem as the solution of a PDE, and even if we could we could not resolve the solution numerically. We instead address ourselves to the question, 'What model can I construct whose solution approximates the true climate as accurately as possible, while being tractable with available computational resources?' The standard approach to this question is to formulate new PDEs that describe the dynamics of some vaguely-defined 'resolvable part' of the solution. These PDEs necessarily include terms that represent the effects of the unresolved part of the solution on the resolved part. Formulating these terms is called 'parameterization.' Global ocean models in the class used for climate prediction are generally deficient in kinetic energy on the range of scales that should be resolvable on their grid. Backscatter parameterizations attempt to rectify this by injecting kinetic energy back into the resolved scales. This talk reviews some recent work developing backscatter parameterizations in the Modular Ocean Model, version 6 (MOM6) for the Community Earth System Model (CESM).