A highly efficient 2D flood model with sub-element topography

Abstract

The need for large-scale and regional probabilistic simulations means that two-dimensional (2D) inundation models are still limited by computational requirements. In addition to parallelisation and physical process simplification, attempts to reduce runtimes typically involve coarsening the computational mesh, which can smooth important topographic features and hence limit accuracy. This paper presents a new 2D flow model that uses an enhanced diffusion-wave algorithm, and incorporates sub-element topography in a computational mesh that adapts to the terrain features. The model utilises a fine topographic resolution without having to use a fine computation mesh, and so achieves fast computational runtimes. The model has been tested against the UK Environment Agency's 2D benchmarking tests, and even though the model is designed to operate at larger spatial scales than those in the benchmarking tests, it is shown to provide comparable accuracy relative to a selection of conventional 2D models, at significantly faster computational speeds. The model therefore has the potential to offer a step change in performance of large-scale probabilistic flood mapping and systems flood risk analysis modelling.