What is it about?
Thunderstorms are hard to predict. Traditional methods are slow or short-sighted (1-2 hours). We built an AI system called DDMS using "diffusion" technology—similar to AI art generators. It learns from satellite images to predict not just where storms move, but how they grow and die. This extends accurate warnings to 4 hours.
Featured Image
Photo by NASA on Unsplash
Why is it important?
Double the warning time means more time to evacuate and protect infrastructure, saving lives and money. Also, unlike expensive ground radars, this satellite-based AI works over oceans and in developing countries lacking infrastructure, making high-quality weather warnings accessible globally.
Perspectives
The team believes integrating diffusion AI is key to modeling the chaotic nature of clouds. The system is highly transferable and could power a global, low-cost thunderstorm warning network by adapting to various satellites.
Kuai Dai
Harbin Institute of Technology
Read the Original
This page is a summary of: Four-hour thunderstorm nowcasting using a deep diffusion model for satellite data, Proceedings of the National Academy of Sciences, December 2025, Proceedings of the National Academy of Sciences,
DOI: 10.1073/pnas.2517520122.
You can read the full text:
Resources
Contributors
The following have contributed to this page
