RismadarVoice Reporters, August 11, 2026
China is increasingly deploying artificial intelligence in weather forecasting as extreme weather events intensify, with locally developed systems emerging as strong competitors to traditional forecasting models.
Chinese AI systems, including Fengwu, developed by Shanghai AI Laboratory, Huawei’s Pangu and Fudan University’s Fuxi, are being tested alongside conventional forecasting systems, particularly during typhoon season.
Unlike traditional weather models that rely on supercomputers to simulate atmospheric physics, AI systems analyse vast amounts of historical weather data to generate forecasts much faster and at lower computing costs.
Researchers say some Chinese AI models can match or outperform conventional systems on certain measures of accuracy. Fengwu, for instance, has been reported by its developers to outperform Google’s GraphCast on about 80 per cent of evaluated weather variables.
The technology could help authorities improve preparedness for floods, evacuations and transport disruptions by providing faster and more accurate information about severe weather.
However, experts say AI forecasting still has limitations. While the systems have shown strong performance in predicting typhoon tracks, they remain less reliable in forecasting storm intensity and have not been fully tested for major climate events.
Techwind CTO Sun Zhi said Fengwu predicted Typhoon Dolphin’s landfall in China five days in advance to within about 30 minutes and 30 kilometres of the actual location.
He stressed that extensive scientific research would still be required before AI forecasts could be fully trusted for long-term climate predictions.
Despite rapid advances in the technology, AI forecasting is expected to complement rather than replace conventional weather forecasting systems in the near future.
