Artificial intelligence has emerged that predicts whether or not lightning will occur in a 30km radius depending on atmospheric pressure, temperature, relative humidity, and wind speed.
Lightning is a disaster that not only electrocuts people or animals, leading to death, but also damages electrical installations and buildings, and sometimes even massive wildfires. A system for predicting lightning is already in practical use, but the equipment for predicting in the system is expensive. A device that receives radio waves generated by lightning is installed in all directions to collect information and calculate the probability of lightning through computational processing at a central processing place.
The EPFL research team at Lausanne Federal Institute of Technology developed a system that predicts lightning under four conditions: atmospheric pressure, temperature, relative humidity, and wind speed by learning data from the Swiss meteorological station for the past 10 years. This AI prediction has a distance error of 30 km, and the predicted hit rate of a lightning strike within 10 minutes is 71%. At the time of thesis publication, there were many misinformations.
However, since the four conditions, such as atmospheric pressure, are data that can be measured simply, if AI reaches the level of practical use, the equipment cost required for lightning prediction can be significantly reduced.
The research team explained that the currently practical system requires expensive data obtained from radar or satellites, and processing itself is slow and complex.In contrast, AI can predict the occurrence of lightning in remote locations outside the radar and satellite range. Related information can be found here .