Process Evaluation for Constraining uncertainty in Climate models (ProcECCo)

Uncertainty in climate projections can cost trillions of dollars and many lives due to challenges in adaptation and mitigation. Anthropogenic aerosols are widely acknowledged as one of the largest sources of uncertainty in the climate forcing.

This project aims to reduce uncertainty in climate projections by developing a new framework for climate model evaluation, focusing on observable process relationships in combination with a robust uncertainty quantification framework.The innovation is mainly based on two factors: 1) traditionally, model evaluations have focused on state variables like such aerosol optical, rather than relationships between variables along causal chains.

In this project, I shift the focus to relationships between variables – “process measures” – using these to evaluate and constrain the models (WP2: aerosol loss rates, 2025, WP3: aerosol production, 2026) This is done within an uncertainty quantification framework, using perturbed parameter ensembles (PPEs, WP1, 2024) together with emulators to comprehensively sample the parameter space, from which we can then exclude unlikely model variants based on the process measures (WP6, 2027).

Previous PPE studies have had limited success because models with very different response to emission changes can produce equally plausible aerosol states.

We hypothesize that process measures will provide a stronger constraint on the aerosol forcing because they relate directly to the causes of uncertainty in the model.