Shingle 2.0: generalising self-consistent and automated domain discretisation for multi-scale geophysical models

Abstract

Abstract. The approaches taken to describe and develop spatial discretisations of the domains required for geophysical simulation models are commonly ad hoc, model-or application-specific, and under-documented. This is particularly acute for simulation models that are flexible in their use of multi-scale, anisotropic, fully unstructured meshes where a relatively large number of heterogeneous parameters are required to constrain their full description. As a consequence, it can be difficult to reproduce simulations, to ensure a provenance in model data handling and initialisation, and a challenge to conduct model intercomparisons rigorously. This paper takes a novel approach to spatial discretisation, considering it much like a numerical simulation model problem of its own. It introduces a generalised, extensible, selfdocumenting approach to carefully describe, and necessarily fully, the constraints over the heterogeneous parameter space that determine how a domain is spatially discretised. This additionally provides a method to accurately record these constraints, using high-level natural language based abstractions that enable full accounts of provenance, sharing, and distribution. Together with this description, a generalised consistent approach to unstructured mesh generation for geophysical models is developed that is automated, robust and repeatable, quick-to-draft, rigorously verified, and consistent with the source data throughout. This interprets the description above to execute a self-consistent spatial discretisation process, which is automatically validated to expected discrete characteristics and metrics.Library code, verification tests, and examples available in the repository at https://github.com/shingleproject/ Shingle. Further details of the project presented at http:// shingleproject.org.

OceanMesh2D 1.0: MATLAB-based software for two-dimensional unstructured mesh generation in coastal ocean modeling

“…The mesh size functions are constructed on a structured grid that relates every point in the meshing domain to a desired mesh size h or, more precisely, a triangular edge length (hence edgefx). Defining the mesh size function on a structured grid has advantages over an unstructured one (Conroy et al, 2012;Engwirda, 2017) in relation to computational efficiency when storing, querying, interpolating, and performing calculations. Further, bathymetric data are often defined on structured grids, and in these cases, computing the mesh size function directly on the same grid can minimize seabed interpolation error for the mesh size function calculation.…”

“…One major drawback of the distance mesh size function is that the minimum mesh size will be placed evenly along straight stretches of shoreline. If the distance mesh size function generates too many vertices, a feature mesh size function that places resolution according to the geometric width of the shoreline should be employed instead (Conroy et al, 2012;Koko, 2015). In this function, the feature size (e.g., the width of channels and/or tributaries and the radius of curvature of the shoreline) along the coast is estimated by computing distances to the medial axis of the shoreline geometry.…”

“…To address this issue, the ocean modeling community has developed approaches and tools to support the automated generation of unstructured meshes for coastal circulation problems (Hagen et al, 2002;Bilgili et al, 2006;Gorman et al, 2006Gorman et al, , 2008Lambrechts et al, 2008;Conroy et al, 2012;Engwirda, 2017;Candy and Pietrzak, 2018;Avdis et al, 2018;Remacle and Lambrechts, 2018). Most works have either tried to minimize topo-bathymetric interpolation error on the mesh (e.g., Gorman et al, 2006) or construct the mesh based on resolving relevant physical processes in the domain and/or preserving the geometry of the shoreline boundary (e.g., Conroy et al, 2012;Engwirda, 2017). An iterative a posteriori method that aims to keep the local Published by Copernicus Publications on behalf of the European Geosciences Union.…”

“…In a related previous work, the Advanced Mesh generator (ADMESH; Conroy et al, 2012) implemented a DistMesh2D-based coastal ocean mesh generator in MAT-LAB. In this work, we build on many of the ideas described in ADMESH with the following primary improvements: (a) a focus on computational efficiency to enable the software to become practically useful even for large geophysical datasets (e.g., ∼ 1 km resolution global topo-bathy) in the MATLAB scripting language; (b) the inclusion of preprocessing and post-processing workflows; (c) a greater variety of mesh size functions and flexibility in their application, which offers more control over mesh resolution placement; and, critically, (d) code written in an open-source environment for the benefit of the community.…”

MPAS‐Ocean Simulation Quality for Variable‐Resolution North American Coastal Meshes

“…This includes MPAS (Ringler et al, ), FESOM (Danilov et al, ; Wang et al, ), ICON (Korn, ), FVCOM (Chen et al, ), SCHISM (Zhang et al, ), SLIM (Kärnä et al, ), and Fluidity (Davies et al, ). Mesh creation tools such as Shingle 2.0 have been developed to produce high‐quality reproducible meshes efficiently (Candy & Pietrzak, ). Models differ in the arrangement of variables on the underlying computational grid and in the numerical techniques employed, with both unstructured triangle‐ and polygon‐based finite‐volume and finite‐element type discretization schemes adopted in various frameworks.…”