The Meteorological Agency's research arm recently began developing an artificial intelligence-based system to predict high-wind events severe enough to impact the safety of land and air transportation, officials said.
The system, which the Meteorological Research Institute aims to develop over a period of four years, is also expected to help self-driving cars avoid accidents caused by strong wind, including tornados and downbursts.
Major wind systems develop quickly and their random nature has made it difficult to forecast their potential severity or location.
East Japan Railway Co. (JR East) has been operating its own system to detect possible gusts following a derailment caused by strong winds in 2005, but its application has been limited.
By monitoring air current data obtained from weather radar, the Meteorological Research Institute's prototype system has been able to detect the formation of high wind events with 90 percent accuracy, the officials said.
The institute needs to collect more data on how winds are generated in warmer periods of the year, they said.
Cumulonimbus clouds — tall, dense, vertical clouds observed mainly in summer — cause unpredictable air currents, making wind prediction more complicated, according to the agency.
