What do we really mean by 'parametrization' in weather and climate models?
- Wednesday 27 October 2010, 16:00 to 17:00
- Open to:
- Staff, Students
- Professor Peter Clark (Surrey)
Abstract: Weather and climate models comprise a so-called 'adiabatic dynamical core' and a collection of 'parametrizations' (often given the unfortunate name of 'physics'). The former is a more or less well understood numerical solution of dynamical equations based on the Euler equations. The latter attempt to represent a wide variety of physical processes.
It is widely accepted that differences in these parametrizations between models are largely responsible for differences in model evolution (and especially model 'climate'), but the interaction between them and how they couple with the 'dynamics' can also be key in ways which are very poorly understood. Some of this behaviour undoubtedly arises from violation of the inevitable assumptions and aproximations behind the parametrizations. Unfortunately, practitioners are often very vague as to what these assumptions are, in particular what averaging assumptions have been made. The purpose of this seminar is to expose some of these assumptions for some processes and discuss some very elementary consequences, in the hope of stimulating interest from areas of mathematics with a more rigorous approach to analogous problems.