Addressing Complexity in Global Aerosol Climate Model Cloud Microphysics

In a quest to represent the Earth system, climate models have become more and more complex. This generates problems, for example, hindering model interpretability. This study contributes to a regain of model understanding and proposes simplifications to decrease scheme complexity. We reflect on the reasons for model complexity and the problems it generates or deepens, connecting perspectives from atmospheric science and the philosophy of climate science. Using an emulated perturbed parameter ensemble of the cloud microphysics (CMP) process efficiencies, we investigate the sensitivity of the model to process perturbations. The sensitivity analysis characterizes the scheme and model behavior, contrasting it to physical process understanding as well as an alternative CMP formulation (comparing the 2M (Lohmann et al., 2007, ) to the P3 scheme (Morrison & Milbrandt, 2015, ; Dietlicher et al., 2018, )). For the 2M scheme, ice crystal autoconversion dominates the model sensitivity in the ice phase. The P3 scheme removes this artificial process and thus shows more balanced sensitivities. Model behavior sometimes aligns with process understanding, but many process sensitivities are masked by other more dominant processes or the model finally responds differently due to adjustments. We identify processes that the model is not sensitive to and test their simplification. For example, heterogeneous freezing or secondary ice production are drastically simplifiable. Depending on one's modeling vision one may interpret this study's findings as pointing to simplification potential in the CMP scheme or the need for process representation improvements where the model behavior does not tally with our physical understanding.