Evaluation of different meshing criteria for areas exposed to flooding

Abstract

This paper investigates the relationship between criteria used for the unstructured mesh generation of two dimensional shallow water models and model result convergence. A physically based, localized truncation error remeshing method is recursively applied to an ocean to estuary domain. The adapted meshes are tested in a tidal study with the wet and dry option turned on: their quantitative performance, made at three stations within the estuary, shows that the method improves the reproduction of the measured flow velocity, especially at the upstream station, as a consequence of an increased mesh resolution within the estuary.A comparison between four alternative discretization criteria (coarse estuary discretization; fine estuary discretization, multi-criteria approach and its a-posteriori, error adapted mesh) is made under the same model conditions of the previous test. Although the study does not identify a best mesh, it is possible to see how the coarsest estuary discretization has the lowest performance indices, while the finest uniform estuary discretization outperforms the other meshes just in reproducing the water elevation. The a-posteriori, error adapted mesh gives a more accurate representation of the of the wetted area extension than the multi-criteria mesh it is derived from.