Spherical Diffusion Model Accelerates Century-Scale Climate Projections from Months to Hours
UC San Diego and the Allen Institute for AI developed Spherical DYffusion, a hybrid AI-physics model that compresses 100 years of climate simulation into 25 hours. It uses generative AI combined with physics-driven data to produce accurate long-term climate forecasts far faster than traditional methods.
This innovation teaches the value of integrating AI with domain-specific physics to drastically speed up complex simulations without sacrificing fidelity. It pushes scientists and practitioners to explore hybrid modeling approaches for efficient problem-solving.
The collaborative team at UC San Diego and the Allen Institute for AI applied this model to produce actionable climate insights with unprecedented speed, influencing policy and research timelines.
Step 1: Visit UC San Diego’s project page for Spherical DYffusion (https://today.ucsd.edu/story/nine-breakthroughs-made-possible-by-ai). Step 2: Download the model code and datasets provided. Step 3: Run the model on your climate data using the recommended computational setup to generate accelerated long-term climate projections within a day.