China is combining artificial intelligence with conventional numerical weather forecasting to improve typhoon predictions. For Typhoon Bavi this year, a hybrid model produced an accurate track forecast 108 hours in advance, according to Niu Zeyi, assistant researcher at the Shanghai Typhoon Institute under the China Meteorological Administration. Niu said typhoon patterns have become more unusual and harder to forecast, and that tracks are strongly influenced by the subtropical high, whose position and shape are difficult to predict—an area where AI models can help. His team merged AI strengths with traditional numerical methods to forecast both trajectory and intensity. Accurate early forecasts help place personnel and resources where they are most needed. China’s meteorological system uses Fengyun satellites, weather radar, and automatic observation stations to feed real-time data into high-performance computers, with forecasters combining model output and experience.
China is combining artificial intelligence with conventional numerical weather forecasting to improve typhoon predictions. For Typhoon Bavi this year, a hybrid model produced an accurate track forecast 108 hours in advance, according to Niu Zeyi, assistant researcher at the Shanghai Typhoon Institute under the China Meteorological Administration. Niu said typhoon patterns have become more unusual and harder to forecast, and that tracks are strongly influenced by the subtropical high, whose position and shape are difficult to predict—an area where AI models can help. His team merged AI strengths with traditional numerical methods to forecast both trajectory and intensity. Accurate early forecasts help place personnel and resources where they are most needed. China’s meteorological system uses Fengyun satellites, weather radar, and automatic observation stations to feed real-time data into high-performance computers, with forecasters combining model output and experience.