Flowchart of the methodology designed for this analysis. Source: The forefront of geoscience (2024). DOI: 10.3389/feart.2024.1342468
A team of physicists, geologists, and signal theorists from the University of Granada in Spain have developed a machine learning-based algorithm designed to predict when Mount St. Helens will erupt.
According to a study published in the journal The forefront of geoscienceThe research team built a unique algorithm to use historical data to predict past eruptions.
Scientists have long tried to predict when specific volcanoes will erupt, and as part of that effort have placed sensors on or near volcanoes to learn more about the type of volcanic activity that leads to an eruption.
So far, such efforts have not been of much use: There is so much conflicting data that it is still impossible to predict eruptions for more than a day or two, and signals that indicate an imminent eruption at one volcano may mean absolutely nothing at all at another.
In this new effort, the team improved their chances of predicting eruptions by studying a single volcano: Mount St. Helens in Skamania County, Washington.
This work involved collecting as much recorded data as possible about the volcano over the years, including before and during the 1980 eruption, one of the most powerful volcanic eruptions to occur in recent history in North America, and also one of the deadliest and most economically damaging.
The researchers used the data to train machine learning algorithms that they hoped would find patterns that human researchers had missed, and they also added mathematical formulas to mine meaning from the seismic signals, such as the amount of pressure buildup and stored energy.
The researchers then trained the algorithm on data from several eruptions and asked it to predict whether an eruption was imminent before any previous eruptions. The team found that the algorithm was 95% accurate in predicting past eruptions at least three days in advance.
The new effort comes after other teams monitoring the volcano reported seeing activity at the site that could indicate another eruption is imminent, including 350 earthquakes since February, many of which occurred several miles below the crater floor.
For more information:
Pablo Rey-Devesa et al. “A universal machine learning approach to predicting volcanic eruptions using seismological signatures” The forefront of geoscience (2024). DOI: 10.3389/feart.2024.1342468
© 2024 Science X Network
Quote: Machine learning algorithm proves highly accurate in predicting Mount St. Helens eruption (July 1, 2024) Retrieved July 1, 2024 from https://phys.org/news/2024-07-machine-algorithm-highly-accurate-mount.html
This document is subject to copyright. It may not be reproduced without written permission, except for fair dealing for the purposes of personal study or research. The content is provided for informational purposes only.
