Primary Convener:
Donald Lucas, Lawrence Livermore National Laboratory

Convener:
Gemma Anderson, Lawrence Livermore National Laboratory
Vipin Kumar, University of Minnesota Twin Cities
Duncan Watson-Parris, University of California San Diego

Chair:
Donald Lucas, Lawrence Livermore National Laboratory
Gemma Anderson, Lawrence Livermore National Laboratory
Duncan Watson-Parris, University of Oxford

Cutting-edge deep learning methods are revolutionizing climate and Earth system science. Emerging techniques such as diffusion models, normalizing flows, variational autoencoders, long short-term memory networks, and vision transformers, are unlocking new possibilities for understanding and predicting complex environmental phenomena. Presentations will highlight applications of these methods to critical climate challenges, including modeling extreme floods, tracking wildfire progression, and detecting drought patterns with unprecedented precision. By leveraging the power of advanced neural network architectures and generative AI methods, researchers are pushing the boundaries of predictive accuracy, spatial resolution, and temporal insights in climate science. The session will foster interdisciplinary collaboration, showcasing how deep learning is transforming traditional approaches to data analysis and simulation. Attendees will gain insights into the latest advancements and learn how these tools can be applied to improve climate resilience and decision-making in the face of growing environmental challenges.

Index Terms
3305 Climate change and variability
0555 Neural networks, fuzzy logic, machine learning
1622 Earth system modeling
1942 Machine learning

Cross-Listed:
NH - Natural Hazards
IN - Informatics
NG - Nonlinear Geophysics
A - Atmospheric Sciences

Suggested Itineraries:
Machine Learning and AI
Open Science and Open Data

Neighborhoods:
3. Earth Covering