Clouds remain major sources of uncertainty in Earth System Models (ESM). Their large scale properties (albedo, lifespan or precipitation efficiency) depend on the interplay of processes (turbulence, convection, cloud and aerosol microphysics) that happen on spatial and temporal scales that are too fine to be directly resolved. It is important to parameterize each of those processes accurately but also to couple them consistently. This is often challenging because our understanding of certain processes is still limited, or because of the sheer variability observed in nature. Increasing computational resources won't immediately solve these issues, but present new opportunities, provided that our research software can adapt to the evolving computer architectures.

In this talk I will describe the sub-grid scale parameterization stack of the CliMA ESM. I will focus on the unified EDMF-based parameterization of turbulence and convection, and the way it can be coupled to a variety of cloud and aerosol microphysics parameterizations. The coupling is accomplished by sampling inputs over the implied subgrid-scale distributions of conserved variables and the assumed vertical precipitation structure. I will show results for different cloud regimes, discuss the advantages of the unified parameterization approach, and highlight how the simulation accuracy can improve thanks to data-driven calibration of the free parameters.

Finally, I will discuss some of the modern research software development aspects such as model design choices that allow for easy free parameter calibrations, clear interface between parameterizations that ensures separation of concerns, as well as software documentation and testing. I will showcase how software design and workflow patterns can help break the complexity of the problem and lower the barrier for contributing to the project.