Causal networks for climate model evaluation and constrained projections

Abstract

Global climate models are central tools for understanding past and future climate change.The assessment of model skill, in turn, can benefit from modern data science approaches.Here we apply causal discovery algorithms to sea level pressure data from a large set ofclimate model simulations and, as a proxy for observations, meteorological reanalyses. Wedemonstrate how the resulting causal networks (fingerprints) offer an objective pathway forprocess-oriented model evaluation. Models withfingerprints closer to observations betterreproduce important precipitation patterns over highly populated areas such as the Indiansubcontinent, Africa, East Asia, Europe and North America. We further identify expectedmodel interdependencies due to shared development backgrounds. Finally, our networkmetrics provide stronger relationships for constraining precipitation projections under climatechange as compared to traditional evaluation metrics for storm tracks or precipitation itself.Such emergent relationships highlight the potential of causal networks to constrain long-standing uncertainties in climate change projections.