China's First AI Weather Forecast System 'Fengqing' Enters Operational Use
China's first fully self-controlled AI-based weather forecast system, "Fengqing", has officially commenced operations. It significantly enhances the prediction capabilities for extreme weather events.
China's first fully domestically controlled AI-powered system for global short- to medium-range ensemble weather forecasting, "Fengqing" V1.0, has passed its operational review and officially entered national meteorological service.
The system was trained and iterated using the "AI4S" supercomputing cluster at the National Supercomputing Center in Tianjin. It represents China's inaugural business-grade AI ensemble model for weather forecasting, ensuring complete domestic control over data, models, and computing power. "Fengqing" signifies a major advancement in supporting the development of indigenous AI technology for meteorology.
"Fengqing" leverages the computing infrastructure of the National Supercomputing Center in Tianjin and its large-scale, highly parallel AI4S scientific computing cluster. The system processes massive amounts of meteorological data, optimizes models, and enhances accuracy. Its core computing power, underlying algorithms, and data sources are all domestically developed, establishing an independent and controllable technological framework for intelligent weather forecasting.
The system integrates a vision transformer architecture with meteorological physics constraint algorithms. It can predict global weather phenomena up to 11 days in advance. Its predictive accuracy and control over probabilistic errors reportedly surpass the latest AI ensemble forecasting system from the European Centre for Medium-Range Weather Forecasts (ECMWF). Supported by Tianjin's supercomputing capabilities, "Fengqing" can perform global meteorological element derivations within minutes, improving the prediction of extreme weather events and long-term trends.
Upon its operational launch, the system will support China's meteorological efforts in disaster prevention, transportation, energy management, environmental protection, and public services. The development team plans to continue optimizing the model's accuracy and forecast horizons.