Uncertainty in climate projections carries large costs, in both economic and human terms, through the difficulty it creates for adaptation and mitigation planning. Anthropogenic aerosols remain one of the largest single sources of that uncertainty in the climate forcing. This application is for computing time aimed at reducing this uncertainty through 1) structural model development, coupled with 2) perturbed parameter ensemble evaluation for quantified and constrained uncertainty together with 3) process evaluation using long term measurements in innovative ways. The aim is to produce a model with robustly quantified and reduced uncertainty in future projections. The project is funded both through the Swedish Research Council and Knut and Alice Wallenberg Foundation. Firstly, structural model development is needed due to aerosol models often relying on 20-year-old code which has structural uncertainties that may introduce major errors. Secondly, to explore, understand and reduce the parametric uncertainty in the model, we will use perturbed parameter ensembles (PPEs) together with statistical emulators to sample the model parameter space comprehensively, so that implausible model variants can subsequently be excluded on the basis of the process measures. Thirdly, process evaluation is needed because model evaluation has traditionally focused on state variables such as aerosol optical depth, rather than on the relationships between variables along causal chains. Here the focus shifts to those relationships — "process measures" — which are used to evaluate and constrain the model: aerosol loss rates and aerosol production in successive work packages. Previous PPE studies have had limited success because model variants with very different responses to emission changes can produce equally plausible aerosol states. We hypothesize that process measures provide a stronger constraint on aerosol forcing, because they relate directly to the causes of uncertainty within the model rather than to its emergent state. The computational work is carried out with the Norwegian Earth System Model (NorESM). Alongside the perturbed parameter ensembles, the project includes a set of targeted, protocol-defined sensitivity simulations contributed to the AeroCom Phase-4 experiment CI-NPF, in which nucleation rate and the availability of condensable vapor are perturbed independently to determine whether the modelled aerosol size distribution is formation-rate limited or growth limited, and how that behavior propagates into the effective radiative forcing from aerosol–cloud interactions. These simulations serve the same scientific goal by a complementary route: where the PPE samples parameter uncertainty broadly, the CI-NPF ensemble isolates a specific causal chain that the process measures are designed to constrain, and places NorESM's behavior in a multi-model context. These results will pave the way for the ongoing development of the future generations of Earth simulations. The resources will be used by the applicant, co-applicant, together with master and PhD students of the applicants.