Quantifying Aerosol Radiative Forcing Uncertainty Within an Earth System Model Using Perturbed Parameter Ensembles

Aerosols affect climate through aerosol-radiation interactions (ARI) and aerosol-cloud interactions (ACI), resulting in a negative radiative forcing (cooling) masking part of the warming by Greenhouse gases. However, climate models face significant uncertainties in quantifying the aerosol cooling effect due to uncertainties in aerosol processes and properties, posing challenges for accurate long-term climate predictions and projections. Our study aims to reduce these uncertainties using a perturbed parameter ensemble (PPE) using the ECHAM6-HAM2.3 model. We perturb 24 aerosol parameters across 180 simulations for both preindustrial and present-day atmospheres and will analyze the resulting aerosol forcing uncertainties and their underlying causes. Both convolutional neural network and Gaussian processes machine learning techniques are used to emulate the ensemble. This work lays the foundation for future work where the PPE will be used in combination with novel satellite observations by the NASA PACE and ESA EarthCare mission to reduce climate model uncertainty.