AI and the World's Hard Problems · Earth & climate science · arXiv · 2026-10-22
Tropical Cyclone Attributes in Sparse Autoencoder Features of an AI Weather Model
To learn how AI weather models represent tropical cyclones internally, the authors analyzed sparse autoencoder activations in GraphCast over 2,115 six-hourly timesteps from 205 cyclones, relating them to position, intensity and pressure structure across five Layer-8 decompositions and 16 processor layers. Each attribute relates to multiple features, position and pressure-structure relations overlap in a smaller subset, and this organization recurs across independent SAEs and layers without feature alignment. In deeper layers the leading position feature aligns better with position and pressure structure six hours ahead than at present, suggesting attribute-relation organization is a reproducible level of description for interpreting cyclone-related information in GraphCast.