A stabilization for three‐dimensional discontinuous Galerkin discretizations applied to nonhydrostatic atmospheric simulations. (15th December 2015)

• Record Type: Journal Article
• Authors: Blaise, Sébastien; Lambrechts, Jonathan; Deleersnijder, Eric
• Is Part Of: International journal for numerical methods in fluids. Volume 81:Number 9(2016)
• Page Start: 558
• Page End: 585
• Publication Date: 2015-12-15
• Subjects: discontinuous Galerkin -- atmospheric model -- Euler equations -- geophysical flow -- computational fluid dynamics; Fluid dynamics -- Mathematics -- Periodicals; 532
• DOI: 10.1002/fld.4197
• Languages: English
• ISSNs: 0271-2091

Abstract

Summary: A discontinuous Galerkin nonhydrostatic atmospheric model is used for two‐dimensional and three‐dimensional simulations. There is a wide range of timescales to be dealt with. To do so, two different implicit/explicit time discretizations are implemented. A stabilization, based upon a reduced‐order discretization of the gravity term, is introduced to ensure the balance between pressure and gravity effects. While not affecting significantly the convergence properties of the scheme, this approach allows the simulation of anisotropic flows without generating spurious oscillations, as it happens for a classical discontinuous Galerkin discretization. This approach is shown to be less diffusive than usual spatial filters. A stability analysis demonstrates that the use of this modified scheme discards the instability associated with the usual discretization. Validation against analytical solutions is performed, confirming the good convergence and stability properties of the scheme. Numerical results demonstrate the attractivity of the discontinuous Galerkin method with implicit/explicit time integration for large‐scale atmospheric flows. Copyright © 2015 John Wiley & Sons, Ltd. Abstract : A stabilization, based upon a reduced‐order discretization of the gravity term, ensures the balance between pressure and gravity effects. Validation with a stability analysis and numerical experiments confirming the correct convergence rate.