Disentangling model structural and parametric causes of uncertainty

We reveal structural inconsistencies in microphysical properties in a global climate model, using a large ensemble of model variants that densely sample model parameter uncertainties related to clouds, aerosols, radiation and precipitation. Our method disentangles the effects of structural and parametric uncertainties and is used to quantify the limiting impact of structural uncertainties on observational constraint. Using high-dimensional visualization tools, we interrogate the impact of observational constraints on model parametric uncertainty. Additionally, we quantify the remaining uncertainty in aerosol-cloud interaction forcing after observational constraint and identify specific observations with the greatest potential to further constrain models.