By Giuliano Camiletti

Edited by Gisela A. Morán & Jessica Pi

One winter morning on June 4, 2011, the Puyehue-Cordón Caulle volcano erupted with the energy of 70 atomic bombs. Its ashes spread massively, nearby towns were evacuated, roads blocked, and hundreds of flights were canceled. However, its eruption wasn’t a surprise. The volcano began to show signs of activity in April of the same year, swarms of earthquakes that grew in frequency and intensity, reaching more than 300 earthquakes per minute in early June: the eruption was imminent. Signals like these are used to monitor the activity of volcanoes to predict their eruption and be prepared for when it occurs. 

The prediction of volcanic eruptions still represents, for many volcanoes, a scientific problem to be solved. The progress in volcano monitoring techniques has rocketed (measurement of earthquakes, observations with satellite images), and it is possible to have an accurate picture of their activity in real time. However, making accurate predictions of when the eruption will happen is unlikely, at least in humans' time. Why?

Some signs indicate volcanic activity, such as earthquakes evidencing the movement of magmas (molten rock under volcanoes, Fig. 2), the deformation of the volcano's surface, or the emission of gases. As pictured, The Puyehue volcano showed the signs and soon after erupted. However, the presence of these signs does not always translate into a volcanic eruption. Sometimes a volcano presents all the expected signs before the eruption, and it does not occur. On the other hand, other times, the eruption occurs suddenly without the expected signs, or it is triggered by an external and unpredictable event such as an earthquake unrelated to volcanic activity. For example, the case of the Rabaul volcano, New Guinea, showed signs of unrest in the years 1984-1985, while the eruption in September 1994, only some activity was detected less than a day before. 

There is much geological evidence that gives us information about the behavior of volcanoes. Moreover, there are multiple signals before the eruption. However, we cannot know exactly when it will occur, and timing is everything if we want to be prepared for a catastrophe. A great source of information about the volcanic activity is past eruptions. This gives us great information about the behavior of a volcano, i.e., how often it erupts, whether it is violent or not, and its size. However, many eruptions seem to occur randomly, with no apparent "memory" of the volcano's past behavior.

The impediment to making a reliable prediction in most of the world's volcanoes is an invitation to continue research to find those unknown unknowns and variables that have the potential to generate deadly eruptions. However, there is hope on the horizon. The emergence of new technologies allows volcanologists to combine artificial intelligence with monitoring techniques, increasing the potential to simultaneously observe many volcanoes and eventually increase the probability of predicting eruptions.

If you want to learn more about this topic the author recommends the following resources:

• "How AI and satellites could help predict volcanic eruptions"

• DE BELLO VULCANICO 40-year scientific effort of ‘predicting the unpredictable’ since the 1980 eruption of Mount Saint Helens

• Documental: Fire of Love (2022)