Event image

ABSTRACT:
Dynamically adaptive unstructured mesh numerical modelling allows model resolution to be concentrated in  regions of relatively increased dynamical importance. For the modelling of the ocean this could allow active eddying regions to be resolved using locally higher resolution, while more quiescent regions are  treated using lower resolution. These methods have the potential to increase the efficiency of eddy resolving calculations of the large-scale ocean.

The quasi-geostrophic equations are used to test the viability of this approach for the simulation of  high Reynolds number large scale ocean dynamics. The simplified dynamical equations capture larger scale dynamical features (geostrophic eddies, or the ocean mesoscale) but filter out smaller and faster scale features, thereby providing a classic simplified test bed for the study and modelling of the ocean mesoscale. The dynamically adaptive numerical model is constructed via a combination of the FEniCS system and the Fluidity-ICOM model. The absolute performance of the approach is compared against a simple finite difference approach, and open questions regarding the configuration and application of such a model are discussed.

ADDITIONAL INFORMATION:
James studies the geophysical fluid dynamics of the ocean in general, and of ocean eddies in particular,  involving a combination of theoretical study and high-resolution numerical modelling. He is developing an increasing interest in automated code generation methods, and in particular in inverse methods which utilise automatically derived discrete adjoint models.
http://www.maths.ed.ac.uk/people/show?person=364