Many volcano scientists spend their time monitoring volcanoes to predict if, or when, they’ll next erupt. The challenge for these scientists is to forecast the exact probability of an eruption and the moment it will occur.
To address this problem, a team of researchers developed an early-detection system to determine when to issue alerts about volcanic eruptions. Their motivation was that previous alert systems weren’t accurate enough because they couldn’t detect underground volcanic processes that occur right before an eruption.
From 2014 to 2023, the researchers conducted a series of tests to accurately detect temporary, low-frequency vibrations using an instrument called a seismometer. The researchers focused on these transient signals because they come from earthquakes and other ground phenomena, which scientists often rely on to predict volcanic eruptions. These phenomena include ground tilting caused by fracturing from the movement of magma and volcanic gases. The scientists named these signals the “Jerk” signals because they occur due to sudden movements.
The team set up their experiment at Piton de la Fournaise, a volcano at La Réunion Island off the east coast of Madagascar. They discovered the Jerk signals by detecting and analyzing mysterious short-term signals from existing and experimental seismic data. During real-time experiments, they detected signals 8 kilometers (about 5 miles) from the volcano at the Rivière de l’Est seismic station. These signals resulted from changes in ground elevation as magma moved, and they preceded eruptions by minutes and sometimes hours.
The researchers measured the Jerk signals in units of newton-meters per second, which conveys how fast physical changes occurred in or around the volcano. In addition to real-time experiments, the team checked the Jerk signals’ accuracy by analyzing the station’s past data to determine when these signals would have occurred before an eruption.
To avoid false alarms during their experiments, the researchers needed to distinguish between the signal amplitudes of ocean tides and the transient Jerk signals. So, they used computational software to predict the signals from ocean tides. This was necessary because the instruments that record ground phenomena are also sensitive to tidal movements.
The Jerk system sent its first automatic alarm in June 2014, 1 hour before the first volcanic tremor, signalling that magma was at the Earth’s surface. The team recorded the last signal in their study on July 2, 2023, 40 minutes before the eruption, at a slow 1.5 newton-meters per second. From 2014 to 2023, other Jerk detections occurred between a few minutes and up to 8 hours before an eruption.
To further assess the strength of the Jerk signals, the team examined previous eruption records at Piton de la Fournaise. They found that for the 24 eruptions between 1998 and 2010, a Jerk alarm would have gone off 83% of the time. Of the 48 eruptions that occurred between 1998 and 2023, including the eruptions during their real-time tests, about 42 Jerk alarms would have occurred. However, they found that the Jerk signals were more accurate during their real-time analysis than they would have been for the eruptions 20 years earlier.
Although the researchers successfully detected pre-eruption signals, they noted that the Jerk alarms occasionally arrived too late due to the data processing time, which takes at least 10 to 15 minutes. They reported 2 late alarms: one on June 11, 2019, and the other on February 10, 2020. Therefore, after 10 years of tracking the Jerk signals, the team evaluated their success rates at 22 successful alarms out of 24 eruptions, for a 92% positive rate.
Before using the Jerk signals, many volcanologists struggled to predict the exact moment a volcano erupts. The team concluded that, in addition to the usual methods for detecting pre-eruption earthquakes, observatory scientists could now use Jerk signals to confidently notify authorities in advance of volcanic eruptions, so they can alert people living near volcanoes worldwide.
