Element-Based High-Order Galerkin Methods for Atmospheric Numerical Modeling

Abstract. Atmospheric numerical modeling has been going through evolutionary changes over the past decade. One major reason for this trend is due to the recent paradigm change inscientific computing, triggered by the arrival of petascale computing resources with core counts in the tens of thousands to hundreds of thousands range. To meet the challenges of building a new generation atmospheric general circulation model, the National Center for Atmospheric Research (NCAR, USA) has developed a computationally efficient and scalableatmospheric modeling framework known as the High-Order Method Modeling Environment (HOMME). For the spatial discretization, HOMME primarily employs element-based Galerkin approaches such as the spectral-element and discontinuous Galerkin (DG) methods, on a cubed-sphere tiled with quadrilateral elements. The DG method has several computationally attractive properties, which include local conservation, geometric flexibility, high-order accuracy and high-parallel efficiency. Recently, the HOMME framework has been further extended to include non-hydrostatic dynamics based on fully compressible Navier-Stokes system. In this talk I will describe the DG discretization as implemented for the 3D global model with an operator-split time integration approach.