A 2D parallel diffusive wave model for floodplain inundation with variable time step (P-DWave)

• We develop a parallel 2D diffusive wave model in Matlab and Fortan.
• We achieved speed-up times ranging from 1.2 to 5.2 using 2 to 12 processors.
• The variable time step method and the process for Wet–Dry fronts kept the solution stable.
• Absolute mass conservation is obtained in all seven tests used to validate the model.

SummaryGrowing advances in remote sensing technologies together with the widespread availability of Digital terrain models (DTM) have intensified the research into two-dimensional (2D) models. Supported by detailed DTM, 2D models can become very accurate tools yet not without an added cost on the computational effort. Floodplain inundation is characterised by a slow varying phenomenon which can last hours, days or even weeks. In this paper we aim to develop a specific parallel diffusive wave model with variable time step suitable for flood inundation. Taking advantage of up to 12 processors, speed-up times ranging from 1.7 to 5.2 and 1.2 to 1.7 are achieved with the Matlab parallel computing toolbox and Fortran OpenMP Application Programming, respectively. The variable time method and the process devised to represent Wet–Dry fronts kept the solution stable and preserved absolute mass conservation. P-DWave performed well against known analytical solutions and dynamic and diffusive models in a total of seven tests.

Figure optionsDownload as PowerPoint slide