MarsMPAS - a new global mesoscale circulation model for Mars

We present a new Mars global circulation model based on the NCAR Model for Prediction Across Scales (MPAS). MPAS employs an unstructured horizontal computational mesh that allows for arbitrary grid point location, including cases of globally uniform mapping and selective zooming. The model scales very efficiently across massively parallel supercomputers allowing very high (global mesoscale) resolution simulations to be performed with practical amounts of computational resources. The quality of the numerics and the lack of grid distortions (such as the "polar points") means that MPAS is much superior to existing models for problems involving polar dynamics and tracer transport. Significantly, the MPAS-atmosphere model was developed at NCAR to use the same "physics" modules as the WRF modeling system. As such, MPAS and WRF may be considered alternate dynamical cores within a single modeling system; for which a large collection of physics for various planets has been developed within the planetWRF system. The MPAS model solves the fluid dynamic equations for a fully compressible, non-hydrostatic atmosphere and can explicitly resolve convective motions induced by CO2 condensation. Just like its design for the Earth, MarsMPAS can be serve as a cloud resolving model for CO2, and for water ice clouds and dust. With the MarsWRF physics, MarsMPAS reproduces orbiter observations of the temporal and spatial variation of air temperature, and the seasonal lander surface pressure cycles. Examining argon enchancement at the southern winter pole (a sensitive inducator of both polar heat balance and tracer transport fidelity), MarsMPAS is able to achieve a maximum argon enhancement factor of 5 to 6, in better agreement with observations than predictions from any previous Mars GCM. Benchmarks suggest that MarsMPAS is highly scalable in terms of parallel computation, and much larger time steps (factor of 10 larger than those used in MarsWRF) are permissible for mesoscale simulations (15km or smaller). We also address the use of height-based coordinates in MPAS - specifically the suspected "pressure cooker effect" due to the large seasonal change in both total atmospheric mass and global mean air temperature - and quantitatively demonstrate that it is not of concern for reasonable global atmospheric model top heights (>75km).