On October 29, at the 2026 Fengyun Satellite User Conference, the inaugural batch of artificial intelligence models tailored for extreme weather, built upon data from Fengyun meteorological satellites, was officially unveiled. This marks a significant leap for Fengyun satellite applications, transitioning from real-time monitoring to forward-looking forecasting. The newly released intelligent cloud image interpretation model for typhoons seamlessly integrates multi-channel cloud imagery and associated data from Fengyun satellites. This integration facilitates the automated analysis of a typhoon's position, structure, intensity, and evolving patterns, culminating in the generation of detailed interpretation reports. Presently, the model boasts comprehensive interpretation capabilities, encompassing seven distinct categories of typhoon attributes and amalgamating over 20 real-time satellite products. It performs interpretations every 15 minutes and produces interpretation texts within a mere 27 seconds.

Another notable model unveiled at the conference is the severe convective short-term warning model. This model harnesses continuous observational data from the Fengyun-4 satellite collected over the preceding two hours, augmented by AI technology, to facilitate future forecasting. Its innovative approach lies in dissecting cloud image evolution into two key components: the 'overall movement of large systems' and the 'local generation and dissipation of small systems.' This decomposition enables the model to advance warning times, providing alerts up to four hours in advance.