Parameterizations in earth system and atmopsheric models frequently include free parameters that are poorly constrained by observations or theory. In and ongoing collaboration with
Juliane Muller in the Computational Research Division we are developing a computationally cheap framework to calibrate a parameterization using surrogate models. We are testing the new approach for the parameterization of turbulence and shallow clouds with the goal to provide more accurate representations of, e.g., cloud feedbacks in climate models.