Modeling urban flood inundation in a parallel computing environment

Algorithms for parallel processing have been described and demonstrated for grids containing over six million cells. This has been achieved by the use of the freely available MPI libraries. The efficiencies are good and demonstrate that such a technique can be used in urban areas without loss of representation of physical phenomena. The implementation was successful for both a large cluster and a smaller dual-core PC system. Further studies will be carried out to analysis the parallel efficiency. Work is also underway to implement a multi-block approach that maintains the Cartesian grids used here, but that allows for different resolutions to be used in different sub-domains as appropriate. Further simulations for Carlisle at a resolution of 1m will be carried out with a DEM that better represents the main channel.