AI typhoon track forecasts outperform physics models in Japan
Artificial intelligence models are providing more accurate typhoon track forecasts than traditional physics-based systems. While these tools are already in use by private services, researchers are working to integrate them into national disaster prevention strategies.
Artificial intelligence is demonstrating superior accuracy in predicting the paths of typhoons, with recent data showing these models outperforming conventional physics-based systems. Private weather service Weathernews Inc. has integrated AI into its forecasting, while the Japan Meteorological Agency is actively researching the technology to enhance its own predictive capabilities.
The advantage of AI was highlighted during the movement of Typhoon Champi, which formed over the Pacific Ocean on 15 October. By 19 October, traditional physics-based models used by the Japan Meteorological Agency and international institutions largely predicted the storm would track towards Taiwan and China. In contrast, AI models correctly identified that the typhoon would move northward near Okinawa before curving over the East China Sea. The storm followed the AI-predicted path, turning sharply upon approaching the Nansei Islands.
Munehiko Yamaguchi of the Meteorological Research Institute noted that AI is particularly effective at analysing large-scale atmospheric flows, such as the Pacific high-pressure system and westerlies that influence typhoon trajectories. According to Weathernews, track deviations for the first 26 typhoons of the previous year were approximately 30 per cent lower when using AI compared to traditional physics-based models.
Despite these gains in tracking, experts warn that the technology is not yet a complete solution. Yamaguchi explained that AI models currently struggle to forecast typhoon intensity, which complicates predictions regarding wind speeds and rainfall. Consequently, the Japan Meteorological Agency continues to rely primarily on physics-based models while using AI as a supplementary reference.
Conventional physics-based models function by inputting data such as temperature, air pressure and water vapour levels into supercomputers to calculate outcomes based on the laws of physics. AI models, by contrast, learn from vast amounts of historical weather data to identify tendencies and regularities, searching for patterns similar to current conditions to estimate atmospheric movements.
Private sector adoption remains robust, with Weathernews providing AI-generated track forecasts to its paid subscribers. Agricultural businesses have also begun to rely on these tools for disaster prevention. For instance, Naoto Yamada, vice president of Tanemaki Joso, an agricultural corporation in Joso, Ibaraki Prefecture, reported that his organisation consistently consults AI forecasts due to their high accuracy.
The practical benefits of this technology were evident when Typhoon Dujuan approached the Kanto region in September. While the Japan Meteorological Agency's initial forecast projected a track towards western Japan or Okinawa, AI models indicated a potential approach to Kanto, which is centred around Tokyo. Based on the AI-generated information, Tanemaki Joso was able to take early precautions, including installing flood barriers and arranging staff shifts about a week in advance.
Looking forward, researchers anticipate that rapid technological development may soon address the limitations of AI in intensity forecasting. Yamaguchi stated that AI models have become an essential component of modern meteorology and that the focus must now shift toward integrating these tools effectively into disaster prevention strategies.
Elected News
-
Bangladesh Seeks UN Women Support on Digital Harassment and Women’s Empowerment
-
Bangladesh, Australia Discuss Human Trafficking, Irregular Migration and Rohingya Crisis
-
Bangladesh Plans Internet-Free Access to Mobile Banking and Financial Apps
-
DU to Open Undergraduate Admission Applications on 11 November, Tests Begin 5 December