Mathematikgebäude, M614

The complexity of ocean coastlines and bottom topography and need to concentrate resolution on dynamically important features of circulation makes methods based on unstructured meshes especially attractive to ocean modeling. However, the mainstream of numerical ocean modeling goes through standard finite differences. The main reason is their much higher numerical efficiency and algorithmic simplicity. I will present our results pertaining to the development of Finite-Element Ocean general circulation Model (FEOM) at the Alfred Wegener Institute for Polar and Marine Research (AWI), and briefly outline its performance in a number of projects run at the AWI. Our typical meshes include from 0.5 to 5M 3D nodes, and we are interested in long integration times measuring up to hundred years (climate prediction). This calls for dedicated efforts toward increasing the overall numerical efficiency on parallel architectures and, in particular, that of iteractive solvers. There is also need in higher-order advection schemes allowing to reduce the numerical diffusion to much lower values (to minimize spurious mixing of watermasses with different temperature and salinity). In conclusion, I will briefly characterize other efforts worldwide aimed at designing ocean circulation model working on unstructured grids, and present our future plans.

Dr. S. Danilov

Alfred Wegener Institute for Polar and Marine Research, Bremerhaven