Causal inference provides a powerful framework for unraveling cause-effect relationships, disentangling complex processes within Earth system sciences and quantifying causal effects by combining domain knowledge, machine learning (ML) models, and data from observations and climate model simulations. Learning causal relationships from Earth system data across different timescales is crucial for understanding its complex emergent dynamics and validating climate models, which we count on to simulate future changes. A better understanding of the Earth system helps us design more effective mitigation and adaptation strategies to address open risks with environmental and societal relevance. This session welcomes contributions that highlight innovative approaches, methodologies, and case studies employing causality across Earth sciences. The overall aim is to foster interdisciplinary discussions, encourage collaborations, and promote the development of robust analysis frameworks. We would like to initiate discussions on experiences, challenges, and future development of causal methodologies in Earth system sciences.

The topics of interest for this session include, but are not limited to:
• Causal discovery and inference: algorithms and methods for uncovering causal networks among Earth system processes (e.g., weather, climate, clouds and aerosols, teleconnections, chemical-dynamical interactions);
• Causal effect estimation: statistical techniques to estimate causal effects of anthropogenic or natural forcings in the Earth system;
• Case studies investigating causal pathways and mechanisms driving natural and anthropogenic perturbations such as climate change, land-ocean interactions, teleconnections, extreme events, volcanic eruptions, wildfires, etc.;
• Causal modelling, network analysis, and causal model evaluation: development and application of causal models, network analysis, and graphical models to capture the intricate interconnections and feedbacks within dynamical systems;
• Integration of causal discovery with deep learning (DL) as a promising avenue for improving climate simulations;
• Merging of explainable and causal Artificial Intelligence (AI): discovering governing equations, parameterizations, and relationships based on climate data;
• Challenges and limitations: difficulties associated with the application of causal inference, such as data preprocessing, violations of assumptions, or uncertainty quantification.