The powerful help of supercomputers made it possible to predict, five months in advance, the volcanic eruption of the Sierra Negra in the Galapagos Islands, which awakened on June 26, 2018. The result was obtained thanks to the predictive model developed by researchers led by the University United States of Illinois at Urbana-Champaign, who tested it with the extremely high computing performance of the iForge and Blue Waters supercomputers: these devices are capable of performing respectively one trillion and one million billion operations per second.
The study, published in the journal Science Advances, makes it possible to take a big step forward in one of the greatest challenges of volcanology.
The predictive model of volcanic eruptions, developed by researchers led by Patricia Gregg, had already attracted attention for successfully recreating the unexpected eruption of the Okmok volcano in Alaska, which occurred in 2008. To test the new update, the authors of the study chose to focus on the Sierra Negra volcano, which is an excellent test bed: "It is a well-behaved volcano", comments Gregg, "because before each eruption it shows all the telltale signs we would expect to see , such as gas emissions and increased seismic activity ".
The supercomputers completed the task in January 2018: "Our model predicted that the strength of the rocks that contain the Sierra Negra magma chamber would become very unstable between June 25 and July 5," explains Gregg, "with consequent high probability of eruption ".
The awakening of the volcano, therefore, took place only one day later than the expected date.
According to the researchers, the study shows the power to incorporate high-performance computing, previously unavailable to the volcanic forecasting community, into practical research.