SCALE-DG combines a high-order discontinuous Galerkin dynamical core with the physics infrastructure of the SCALE library, providing a flexible platform for both regional and global atmospheric simulations.

Brief Description

Features

  • Support both 3D regional and global simulations only in SCALE-DG
    • This feature is provided by a model framework provided by our DG library
  • Execute parallel simulations in supercomputers
    • Process and thread parallelizations are performed by MPI library (we use 2D domain compositions) and OpenMP, respectively.
  • Generate initial data for idealized experiments
  • Treat NetCDF files as input and output data
  • Provide a variety of idealized regional and global benchmark test cases

Dynamical core

  • Governing equation: 3D fully compressible non-hydrostatic equations
  • Grid system: hexahedral finite element and horizontally curvilinear coordinate system
    • For the global mode, cubed sphere coordinates are used
    • Topography is treated using a terrain-following coordinate
  • Spatial discretization: nodal discontinuous Galerkin method (e.g., Hesthaven and Warburton, 2007)
    • Polynomial order associated with numerical accuracy can be arbitrarily chosen.
    • Numerical flux: Rusanov flux
    • Stabilization mechanisms: Inherent numerical dissipation with numerical fluxes and explicit modal filtering
  • Temporal discretization: various type of Runge-Kutta (RK) schemes
    • Full explicit (horizontally and vertically explicit; HEVE)
      • Classical 4s4o RK scheme (where 4s4o means 4 stage and 4th-order accuracy)
      • Strong-stability preserving (SSP) RK schemes: 3s3o, 4s3o, 5s3o (Higueras and Roldan, 2018), 10s4o (Ketcheson, 2008)
    • Horizontally explicit and vertically implicit (HEVI)
      • Implicit and explicit RK schemes: ARK232 (Giraldo et al., 2013), ARK324 (Kennedy and Carpenter, 2003)
  • Tracer transport
    • Preserving the non-negativity is ensured by a limiter (Light and Durran, 2016) and SSP RK schemes in conjunction with fundamental spatial discretization by nodal DGM.

Physical processes

Physical parameterizations include schemes provided by the SCALE library as well as simplified schemes implemented within FE-Project/FElib.

◦: scheme using SCALE library components

  • Turbulence process
    • Smagorinsky (1963) and Lilly (1962)-type sub-grid scale model corrected by Brown et al. (1994) and Scotti et al. (1993)
    • MYNN level 2.0 boundary layer scheme (Nakanishi and Niino, 2006)
  • Cloud microphysics process
    • 3-class 1 moment bulk scheme◦ (Kessler, 1969)
    • 6-class 1 moment bulk scheme◦ (Tomita, 2008)
    • Large-scale condensation scheme
  • Cumulus convection process
    • Moist convective adjustment (Manabe et al., 1965)
  • Radiation process
    • Gray longwave radiation scheme (Byrne and O’Gorman, 2013; Frierson et al., 2006)
    • Two-band longwave and one-band shortwave scheme (Geen et al., 2016; Vallis et al., 2018)
  • Surface process
    • Ocean model: slab ocean or fixed initial SST
    • Surface flux scheme
      • Simplified bulk formulation of momentum, heat, and latent flux with constant coefficients for idealized experiments
      • Constant surface-flux option◦ for heat and moisture, while simplified bulk formulation for momentum

Documents

The description of SCALE-DG is available at Document page.

Example of numerical experiments

Simulation results of standard test cases such as baroclinic wave and Held-Suarez experiments are shown in Gallery page.