A strategy for solving the complete 2D shallow water equations on a Cartesian grid is described. Two schemes are used: a Godunov approach with a first order in time and space scheme based on Roe's approximate Riemann solver and a raster-based approach which uses the mass conservation equation in uniform flow assumption. A single processor version of the algorithm which sweeps the complete grid is first demonstrated. Subsequently a multiple processor version is demonstrated for use on a High-performance Grid Computer. Two different underlying domain division techniques are described: the classic Halo swap with non periodic boundary conditions and an independent flood plain treatment. The latter is demonstrated with communication between processors based on the Message Passage Interface libraries, MPI. The purpose of the research was to assess how fine a grid resolution is feasible based on an analysis of computational times for flood inundation prediction.
展开▼
