The Japan Meteorological Agency is contemplating the introduction of artificial intelligence (AI) into its weather forecasting operations to enhance accuracy by leveraging deep learning technology, according to sources familiar with the matter. To facilitate this integration, a specialized team was established in April to oversee the development of necessary infrastructure and technology. This team is working on combining AI-driven predictions, based on extensive historical weather data analysis, with the agency’s current numerical forecasting models.
The proposed AI system would utilize deep learning algorithms to analyze vast amounts of weather-related data, identifying complex patterns to generate forecasts for temperature, rainfall, and other meteorological phenomena. In certain areas, such as predicting typhoon trajectories, AI-based methods may prove more accurate than traditional forecasting approaches. The agency is expected to unveil additional measures around June aimed at enhancing data functionality through AI and other technological innovations.
Currently, the Japan Meteorological Agency employs a numerical weather prediction model that uses observational data and supercomputer simulations to forecast future atmospheric conditions. Experienced forecasters then analyze these simulations, taking into account real-world factors to issue weather reports and disaster warnings. While AI is already being used to refine simulation outputs, human forecasters will continue to play a crucial role in overseeing AI predictions and issuing final forecasts if the new system is implemented.
Beyond weather forecasting, the agency is also exploring the application of deep learning techniques to process observational data from the Himawari-10 weather satellite, scheduled to commence operations in fiscal year 2029. This move underscores the agency’s commitment to harnessing cutting-edge technology to improve meteorological services.
