A Next-Generation Atmospheric Dynamics Model that Conserves Multiple Properties over Complex Topography on All Scales

We present a new 3D nonhydrostatic atmospheric model that uses a 3D generalization of a mass, energy, vorticity, and potential enstrophy conserving numerical scheme for the 2D shallow water equations to solve the governing equations for 3D fluid motion (i.e. the dynamics). The model uses the altitude coordinate (as opposed to a terrain-following coordinate) in the vertical to avoid the large errors in the pressure gradient terms that can be generated near steep topography in terrain-following coordinates. The altitude coordinate allows inclusion of arbitrarily complex (and steep) topography in the model. A feature of this model not found in most others is that the numerical scheme used for the dynamics conserves domain-summed mass and energy for 3D frictionless flows and domain-summed mass, energy, vorticity, and potential enstrophy for the special case of 2D frictionless barotropic flows. These conservation properties are important to maintain because at least in 2D flows, they have been shown to preserve the correct energy cascade. To evaluate the performance of the dynamics core alone, we present simulations without physics (i.e. dry air with frictionless dynamics) on global, regional, and urban scales and a combined simulation on all three scales using nested grids. We also present a global simulation with surface heating and cooling and demonstrate that the model reproduces the basic general circulation of the atmosphere. Since the dynamics core conserves energy in the absence of external dissipation, the energy in the model tends to accumulate in the smallest scales (i.e. the grid scale) over time. To prevent this, we try several dissipation schemes and discuss the results from each. Finally, to demonstrate the scalability of the parallelized model code, we present timing results from simulations using multiple processors.