Parameterizations are used in a General circulation model (GCM) to represent processes that are not directly simulated, either because the corresponding equations are not included or because the processes are too small for the model resolution. For instance, cloud formation or precipitation are calculated from quantities like radiation, water vapor, and vertical wind speed, depending on the exact parameterization scheme employed.

Much of the differences between GCM experiments to predict the rate of global warming can be attributed to different parameterizations in the models. However, the most important design goal in choosing parameterizations for a GCM is not so much picking the "best" one, as finding a set of parameterizations for the model that is consistent.