Cubed-sphere modelling activities at CSIRO

John McGregor 
CSIRO Oceans and Atmosphere, Melbourne

  16 May, 1 pm, in 2155

Abstract:
The Conformal-Cubic Atmospheric Model (CCAM) has been developed at CSIRO over a number of years. CCAM is formulated on the conformal-cubic grid, and employs 2-time-level semi-Lagrangian semi-implicit numerics. The model is quite mature. It has mainly been used in dynamical downscaling studies of climate change, but is also used for specialized numerical weather prediction applications. CCAM employs reversible staggering for the wind components (McGregor, MWR, 2005), producing good wave dispersion behavior and also good behaviour for the kinetic energy spectra. For treatment of non-hydrostatic flow, CCAM utilizes the highly efficient equations of Miller and White (QJRMS, 1984).

Recently the CCAM code has been generalized to utilize the Uniform Jacobian (UJ) variation of the cubed-sphere grid. This grid is derived from the conformal-cubic grid to provide equal area for every grid cell. Since the grid lines are no longer orthogonal, covariant and contravariant velocity components are required. Apart from the complications of the velocity components, most of the CCAM semi-Lagrangian approach may be used, including reversible staggering of the contravariant velocity components to switch between values at cell centres and cell edges. The solver for the Helmholtz equation is a little more complicated than for CCAM. A split-explicit version of CCAM has also been developed, solving the primitive equations in flux form.

A major application of CCAM is for downscaling climate change simulations of coupled atmosphere-ocean GCMs. The CSIRO downscaling strategy utilizes forcing from the sea-ice and Sea Surface Temperatures (SSTs) provided by chosen GCMs. Because the GCMs may have some significant biases in their SSTs, the SSTs from each GCM are corrected for their monthly biases in both mean value and variance, as calculated from their 30-year present-day climatologies. The same monthly SST corrections are applied throughout the simulations from each GCM. The talk will include results from some recent CORDEX regional climate downscaling simulations