Author(s): Sam Jamieson; Julien Lhomme; Grant Wright; Ben Gouldby
Linked Author(s):
Keywords: Local time step; Frozen-flux; Rapid flood modelling; Regional flood modelling
Abstract: Simulating large scale, region wide flood inundation continues to be problematic in terms of simulation accuracy and computational run times. Use of a fine resolution mesh is highly computationally demanding, yet coarsening the computational mesh can smooth topographic features and hence reduce the accuracy of flow process representation. It is possible to maintain the efficiency of a coarse mesh by using a sub-grid representation to ensure topographic accuracy; however, in some approaches the grid edges are not necessarily aligned with topographic features, and so are unlikely to be able to accurately represent flux blockage effects. The Rapid Flood Spreading Method (RFSM) is a computational approach specifically designed to overcome current modelling limitations for large scale and probabilistic applications. Irregular grids are automatically defined around‘topographic depressions’such that any high points (dykes, railways embankments, etc. ) are always fully represented on the cell interfaces. In the past decade, this approach has been incorporated into a number of models of increasing sophistication. The latest version of the model, RFSM-EDA (Rapid Flood Spreading Method–Explicit Diffusion wave with Acceleration term) provides a step change in performance over current techniques, and has previously been shown to provide comparable accuracy relative to conventional 2D models, at significantly faster computational speeds, for both small and large scale applications. However, as with all explicit schemes the computational efficiency can be constrained by the use of a single global time step, which is further restricted by the highly irregular mesh. As a result there is scope for reductions in computational runtimes through the implementation of a Local Time Step (LTS) scheme. This paper illustrates the implementation of a new Local Time Step method which is algorithmically simpler than alternative approaches. It is applied to three test cases: an analytical solution, a small scale floodplain inundation, and a regional scale flood event namely the northern avulsion element of the 2010 River Indus flooding in Pakistan, which led to~2000 fatalities and displaced some 20 million people from their homes. The results show that the LTS scheme has equivalent accuracy to the global time step (GTS) method. The efficiency of the scheme is case-specific; at times it only marginally decreases runtime, whilst at others the runtimes are reduced by up to 32% . The LTS scheme shows promise as a simple method for reducing runtimes without detriment to predictive ability, and further indicates that RFSM-EDA is suitable for regional scale flood inundation modeling, and thus has clear application to national flood risk assessment and real-time flood prediction.
Year: 2014