A Global Version of the MM5 Mesoscale Model

Increases in computing power have led to a blurring of the roles of various atmospheric models. Global models can now be run with resolutions in the tens of kilometers. For regional models it is now also possible to expand the domain to a global size with grid sizes in the same range. To make MM5 global, we have taken two polar stereographic domains, one centered on each pole, and allowed them to interact by providing each other boundary conditions at the equator. This exchange occurs after each time-step, providing a smooth join. The expansion of MM5 to global scales enables it to run without the need for lateral boundary conditions, and so in theory it can run indefinitely given just initial conditions. The primary role of this model is for medium-range prediction and simulations of up to two weeks, a time scale that represents the limits of predictability with state-of-the-art forecast models. Another limitation to this time scale is sea-surface temperature variation that the model cannot update, except climatologically. However, running the model for five-day forecasts, as is now done regularly, illustrates some issues related to the scalability and transferability of physical parameterizations. Currently the model is run globally with a nominal grid size of 120 km, corresponding to a real grid size that varies from 128 km at the poles to 64 km at the equator due to the map-scale factor. The microphysics and convective options used were developed with finer resolution regional applications in mind, and the cloud/radiation interaction lacks any treatment of cloud fractional coverage. Nevertheless, despite the same physics being applied globally, the model behavior appears to be realistic, giving some confidence in both the scalability and transferability of the parameterizations for this application. The model physics had to be adapted for multi-day time scales of the forecasts to account for land-surface time-variability. To this end, we have developed a simple snow-cover prediction model based on surface heat and moisture budgets. MM5 Version 3 also has a full land-surface/snow/vegetation model as a more sophisticated option that can use climatologically varying vegetation fraction. The model is still deficient in gravity wave drag and its treatment of the stratosphere, which probably remain the main limits on its use for long-term general circulation simulations. Interactive vegetation, sea-ice, ocean, and /lake models would also be beneficial for climate applications.