Can we predict the formation of tornadoes?

Even weather experts often have to fail when predicting a tornado. It was the same on May 16, when a tornado plowed trees and covered roofs in the Viersen district. The German Weather Service (DWD) only found out about it afterwards. "We can use weather models and Doppler radar devices to predict severe weather and so-called super cells, from which tornadoes could arise, but the uncertainty is very high," says DWD meteorologist Andreas Friedrich. That is why they work with a few hundred volunteer storm alarms who reported tornadoes in sight. In the Viersen district there was just no one on site at the moment.

Even in the USA, which is particularly plagued by tornadoes, where weather services work with thousands of, in some cases paid, storm chasers and mobile measurement technology, the false alarm rate is still 75 percent. Brian Elbing from Oklahoma State University wants to improve the tornado predictions with a kind of eavesdropping. "Storms emit infrasound up to two hours before a tornado hits the ground, and we can record this sound with microphones," says the researcher. The measurements also worked from a great distance and could make dangerous on-site research by storm chasers superfluous.

Infrasound can be measured and interpreted from great distances

Infrasound has very low frequencies, sometimes well below 20 Hertz, and is also produced by earthquakes or surf waves. People perceive it, if at all, more as a roar, somewhere on the border between hearing and feeling. "Like sound in general, infrasound also contains a lot of information about its source and surroundings," says Elbing. You just have to decipher it. In addition, infrasound can still be measured at great distances, because low frequencies are difficult to stop. "Everyone who has annoying neighbors who turn on music at night and then only hears the bass is familiar," says the researcher.

Using infrasound to predict tornados is not a completely new idea and Elbing is not the only one researching it. Nevertheless, there is a lack of "listening experience" in order to be able to interpret sound measurements reliably. So far there have been only a few reliable studies. At a conference of the Acoustical Society of America in Minneapolis, he recently presented the latest measurement results: the acoustic fingerprint of a tornado that swept through Oklahoma about a year ago, about 20 kilometers away from three infrasound microphones that his team placed in a triangular arrangement on the roof of a university building had mounted.

From the measurement data, the researchers were able to read both the direction and the 46-meter diameter of the tornado. "It exactly matched the official specifications for the width of the damage lane to the nearest meter," says Elbing. There is also some evidence that the amplitude of the sound waves allows conclusions to be drawn about the strength of the tornado. The first characteristic signals were seen around ten minutes before the tornado was born. In future measurements, the researchers also want to use drones in order to obtain even more information and better understand what makes the sound of the rotating air masses.

Some animals are probably not dependent on such details. For example, a group of goldwing woodwingers is said to have run away from a storm with 84 tornadoes in 2014, one or two days before meteorologists had the storm on their radar. An internal, infrasound-based alarm system may have warned them. However, nothing is known about the false alarm rate.