After a long series of discussions, we have decided to study two grids for dynamical cores (3D fluid solvers on a sphere, which are the basis for climate simulations). A careful choice of grids has to be made to deal with the pole problem, where converging meridians on a sphere lead to a vanishing computational time step at the poles (also known as resolution clustering). There is no perfect grid and any choice of grid generally has to deal with resolution clustering, grid imprinting and computational modes (artificial waves that are an artifact of the grid, rather than real physics). We are currently studying the modified Yin Yang and icosahedral grids. If you have an opinion on the choice of grids, we would love to hear it.
Posted on: Wed, 23 Oct 2013 18:33:09 +0000