In the Catatumbo river basin, in Venezuela, south of Lake Maracaibo, there is a storm that lasts about 160 days a year and generates about 300 lightning bolts per hour. It is the place on Earth where lightning strikes the most , so there is a good chance that some will strike us, although we do not know when or where.
However, except in particular places like this and other tropical areas, where thunderstorms are much more frequent than in more temperate regions, predicting when and where lightning will strike is much more difficult.
There are 1,800 storms in progress worldwide at any given time. About 40,000 storms a day. Every day more than 17 million lightning strikes fall on Earth; 200 per second. All in all, we now have a powerful tool to know where the next one will fall .
Algorithmic prediction
The tool has been developed by the School of Engineering of the Federal Polytechnic School of Lausanne (EPFL), specifically by a group of researchers from the Electromagnetic Compatibility Laboratory led by Farhad Rachidi . To create it, a machine learning algorithm was used that was trained to recognize the conditions that lead to lightning strikes.
Once trained, the system made predictions that were correct almost 80% of the time. In addition, it predicted in real time where lightning will strike with a margin of 10 to 30 minutes in a radius of 30 kilometers.
The machine learning algorithm training was conducted with large amounts of data combining meteorological records collected over a ten-year period from 12 Swiss weather stations, located in both urban and mountainous areas. Four parameters were taken into account : atmospheric pressure, air temperature, relative humidity and wind speed.
In addition, because the data was easily acquired in real time (something that had never been achieved before), predictions could be made very quickly, also issuing alerts even before the storm itself formed. That is, it is the closest thing to forecasting the real time (the chronological and the meteorological).
This surprising predictive capacity may thus contribute to an ambitious European project called Laser Lightning Rod , which aims to develop systems to protect the population against lightning. Launched in 2017, the project involves sending short laser pulses of several terawatts into the atmosphere during a storm in order to trigger lightning, guiding it to a specific location and away from vulnerable areas.
In this way, it is intended to cushion the catastrophic effects of some storms: estimates of the average number of people who die from lightning each year range between 6,000 and 24,000 worldwide (men are struck by lightning 6 times more than women). Lightning is also the cause of power outages, forest fires, and damage to electronics and infrastructure amounting to millions of euros each year.
Thanks to the prediction of lightning, then, it is expected to reduce a large amount of negative impact on material and human capital that, since always, has been exposed to the meteorological whim.
