Given the lethality and destructiveness of lightning, knowing in advance when and where it will occur will help reduce impact and injury, foreign media reported. A new artificial intelligence-based system could help, using only standard weather station data.
Developed by a team from the Electromagnetic Compatibility Laboratory of the EPFL Institute of the Federal Institute of Technology in Lausanne, Switzerland, the system “trains” the AI algorithm using readings of four basic weather parameters in the database (atmospheric pressure, temperature, relative humidity and wind speed).
The readings were collected over the past decade by 12 Swiss weather stations in cities and mountainous areas and cross-referenced with records from lightning detection and positioning systems. This allows the AI algorithm to understand which weather conditions are lightning-related in a given area.
The resulting computer system is said to be able to predict 10 to 30 minutes in advance when and where lightning will occur within a 30-kilometer (19-mile) radius, with an accuracy rate of almost 80 percent. These numbers will be increased as technology develops further.
“Current systems are slow and complex, and they need to get expensive external data by radar or satellite,” says Dr. Amirhossein Mostajabi. Our method uses data that can be obtained from any weather station. This means that we can cover remote areas outside the radar and satellite range and where communication networks are not available. “
A paper on the study was recently published in the journal Climate and The Climate And Climate Science.