Using Explainable AI and Counterfactual Reasoning to Attribute Climate Extremes and Improve Policy Making

The Explainable AI & Climate Extremes Policy research project seeks to develop a rapid and scalable Climate Attribution AI framework that employs convolutional neural networks (CNNs) and explainable artificial intelligence (XAI) to generate dynamically consistent counterfactual scenarios for extreme real-world weather events. The project will produce new explainable frameworks that work for precipitation prediction, provide real-time attribution for precipitation extremes, and focus on both the intensity and frequency of these events under various levels of global warming, with a potential to yield faster, more interpretable attribution assessments.

The project's ultimate outcomes will likely deliver actionable insights for policymakers and aid communities in building resilience to climate impacts.

The project is being undertaken by an interdisciplinary team of faculty mentors and researchers deploying novel research at the intersection of AI, explainability, and climate science.