期刊
QUARTERLY JOURNAL OF THE ROYAL METEOROLOGICAL SOCIETY
卷 141, 期 690, 页码 1727-1739出版社
WILEY-BLACKWELL
DOI: 10.1002/qj.2474
关键词
reduced lat-lon grids; icosahedral grid; cubed-sphere; shallow-water equations; spectral element method; discontinuous Galerkin; grid generation on the sphere; transfinite interpolation
资金
- Office of Naval Research [PE-0602435N]
- National Science Foundation (Division of Mathematical Sciences) [121670]
- Air Force Office of Scientific Research
- National Academies through a National Research Council fellowship
We test the behaviour of a unified continuous/discontinuous Galerkin (CG/DG) shallow-water model in spherical geometry with curved elements on three different grids of ubiquitous use in atmospheric modelling: (i) the cubed-sphere, (ii) the reduced latitude-longitude, and (iii) the icosahedral grid. Both conforming and non-conforming grids are adopted including static and dynamically adaptive grids for a total of twelve mesh configurations. The behaviour of CG and DG on the different grids are compared for a nonlinear midlatitude perturbed jet and for a linear case that admits an analytic solution. Because the inviscid solution on certain grids shows a high sensitivity to the resolution, the viscous counterpart of the governing equations is also solved and the results compared. The logically unstructured element-based CG/DG model described in this article is flexible with respect to arbitrary grids. However, we were unable to define a best grid configuration that could possibly minimize the error regardless of the characteristic geometry of the flow. This is especially true if the governing equations are not regularized by the addition of a sufficiently large, fully artificial, diffusion mechanism, as will be shown. The main novelty of this study lies in the unified implementation of two element-based Galerkin methods that share the same numerical machinery and do not rely on any specific grid configuration to solve the shallow-water equation on the sphere.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据