What is Fengyu
Fengyu is the world's first full-chain space weather large model jointly launched by the China Meteorological Administration, Nanchang University, and Huawei. The model is based on an innovative chain training structure, combining satellite observation data with numerical model data to achieve full-chain coupled simulation of the solar wind, magnetosphere, and ionosphere, breaking through the computing power and real-time response bottlenecks of traditional models. It performs excellently in nowcasting and extreme event response, with prediction errors controlled at an internationally leading level. The model is widely used in satellite communications, spacecraft orbit management, power protection, and aviation and polar navigation safety assurance, promoting the deep integration of AI and geospace science.
Main functions of Fengyu
- Space weather monitoring and warning: Real-time monitoring of changes in the solar wind, magnetosphere, and ionosphere, providing early warning of extreme events that may affect Earth's space environment.
- Satellite communication and navigation assurance: Based on accurate prediction of space weather changes, reduce satellite communication interference and navigation errors, and ensure the stable operation of satellite systems.
- Spacecraft orbit management: Optimize spacecraft orbit design, avoid orbital deviations caused by space weather changes, and ensure the safety of space missions.
- Power and infrastructure protection: Predict events such as geomagnetic storms in advance, help power systems take protective measures, and reduce impacts on the power grid.
- Aviation and polar navigation safety enhancement: Provide space weather information support for aviation and polar navigation, reducing risks brought by space weather changes.
Technical principles of Fengyu
- Full-chain coupling: The solar wind, magnetosphere, and ionosphere are modeled separately, and based on a coupling optimization mechanism, information interaction and dynamic response simulation across multiple regions and multiple scales are achieved. The chain structure can truly reproduce how the solar wind affects Earth's surrounding environment and the complex interactions between the magnetic field and the ionosphere.
- Data fusion: Combining real satellite observation data with data generated by the independently developed full-chain space weather numerical model to form a data foundation that is mutually complementary and mutually corroborating.
- AI technology: Innovatively deeply couples the core ideas of artificial intelligence with the mechanisms of Earth system science, and based on machine learning algorithms improves the understanding and prediction capability of space weather change processes, especially performing excellently in nowcasting and extreme event response.
- Prediction performance optimization: With prediction performance testing lasting up to one year, "Fengyu" performs excellently in 24-hour nowcasting capability for the solar wind, magnetosphere, and ionosphere regions, and its prediction error for global total electron content is basically controlled within 10%, at an internationally leading level.
Project address of Fengyu
- Project official website：https://www.cma.gov.cn/2011xwzx/2011xqxxw/2011xqxyw
Application scenarios of Fengyu
- Satellite communication and navigation systems: Reduce interference from solar activity on satellite communication signals and lower navigation errors.
- Spacecraft orbit management: Avoid orbital deviations or damage caused by space weather events, and ensure the safe and efficient execution of space missions.
- Power and infrastructure protection: Provide early warning of space weather events such as geomagnetic storms and help power systems take protective measures.
- Aviation and polar navigation safety: Provide space weather information support for aviation and polar navigation, helping to plan safe flight and navigation routes.
- Scientific Research and Education: As an advanced space weather research tool, it provides data support and a technical platform for researchers and educational institutions.