ClimSim: Supplementary Information

Climate models divide the Earth's atmosphere, land surface, and ocean into a 3D grid, creating a discretized representation of the planet. Earth system models are made up of independent component models for the atmosphere, land surface, rivers, ocean, sea ice, and glaciers. When running as a fully coupled system the "component coupler" handles the flow of data between the components. Within each grid cell of the component models, a series of complex calculations are performed to account for various physical processes, such as phase changes of water, radiative heat transfer, and dynamic transport (referred to as "advection"). Each component model uses the discretized values of many quantities (such as temperature, humidity, and wind speed) as inputs to parameterizations and fluid solvers to output those same values for a future point in time. The atmosphere and ocean components are the most expensive pieces of an Earth system model, which is largely due to the computation and inter-process communication associated with their fluid dynamics solvers. Furthermore, a significant portion of the overall cost is attributed to the atmospheric physics calculations that are performed locally within each grid column. It is important to note that atmospheric physics serves as a major source of uncertainty in climate projections, primarily stemming from the challenges associated with accurately representing cloud and aerosol processes. Traditionally, global atmospheric models parameterize clouds and turbulence using crude, low-order models that attempt to represent the aggregate effects of these processes on larger scales. However, the complexity and nonlinearity of cloud and rainfall processes make them particularly challenging to represent accurately with parameterizations. The MMF approach replaces these conventional parameterizations with a cloud resolving model (CRM) in each cell of the global grid, so that cloud and turbulence can be explicitly represented. Each of these independent CRMs is spatially fixed and exchange coupling tendencies with a parent global grid column. This novel approach to representing clouds and turbulence can improve various aspects of the simulated climate, such as rainfall patterns [2].