China Releases First Batch of AI Models for Extreme Weather from Fengyun Satellites

The operational cadence — a 15-minute interpretation cycle and sub-30-second reporting — is the more consequential detail than the model count, since it moves satellite data from archive material toward something closer to a live feed for forecasters. Pairing that with a nowcasting model that stretches warning lead time speaks to the gap between detection and usable notice, which is where extreme-weather losses are usually decided. The audience is primarily national meteorological services and disaster-response agencies rather than markets. Whether these Fengyun-based models get validated against events beyond China's own basins is the open question.