• The motion patterns typical of volcanoes can be predicted with a high rate of effectiveness as of data sent from the slopes of the Llaima Volcano in La Araucanía Region. This has been possible thanks to the work by Dr. Max Chacón, professor at the Department of Informatics Engineering of Universidad de Santiago, who developed a software program for this purpose. “In Chile, it is essential to increase the knowledge on volcanoes. In this way, we will be able to face emergency situations like eruptions, and even predict them and make timely decisions,” Dr. Chacón said.

Due to its location on the Pacific Ring of Fire, the Chilean territory has the second largest volcanic chain in the world, with more than 2,000 volcanoes, many of them located in the southern part of the country. Although they are considered among the most active volcanoes in Latin America, only 43 of them are being monitored. The main concern about these geological structures is their potential for eruptions and their seismic activity is a key factor in prevention. 

Researchers of Universidad de Santiago, Universidad de La Frontera, and the Southern Andean Volcano Observatory, Ovdas by its acronym in Spanish, conducted the study “Pattern recognition applied to seismic signals from the Llaima Volcano: an analysis of the events’ features”, which is available in the Journal of Volcanology and Geothermal Research.

According to Dr. Max Chacón, professor at the Department of Informatics Engineering of Universidad de Santiago and one of the developers of the program, “Based on the machine learning approach, we developed a piece of software that identifies the type of the seismic signal as of data sent by sensors located on the slopes of the volcano,” with an 80% accuracy in the Llaima Volcano.

“Volcano seismic signals are not related to the earthquakes typical of tectonic plate collisions, as one could deduce from a first interpretation of these phenomena. These movements are specific to volcanoes and they are caused by their distinctive features, like their activity, magma movement, gas movement, stiffness of components, etc.

Previously, the research team had studied the Villarrica Volcano, detecting the existence of three characteristic seismic event patterns: the LP (Long Period) event, which is related to the pressure of gas and other fluids in the conduit; the Tremor, which is related to changes in gas and magma densities; and lastly, the VT (Volcano Tectonic) event, which is associated to the fracture of the solid parts of the volcano or the conduits.

The researchers used the program again in a study on the Llaima Volcano, but they also tried to identify the features of data sent by the sensors in the volcano. Among these data, the amplitude, frequency and phase of the signals, and the way in which they appear together in each seismic signal, were particularly considered.

“With this, we tried to give more information to volcanologists, so that they could identify the signals more easily, also analyzing the way in which these features appeared in seismic events,” Dr. Chacón, an expert in system models, said.

According to Dr. Chacón, one of the most interesting conclusions was the similarity of the results in the two volcanoes, what contradicts the current idea that each volcano has unique seismic movements. The researcher says that for now, the hypothesis is that the similarity is due to the proximity between the two volcanoes; however, this has not been proved.

For the above, there are two potential steps to follow in the context of this research. First, a comparative study on these volcanoes to measure the exact differences between them; and second, the improvement of the program with the purpose of detecting the correlations between these seismic signals at the moment of the eruption.

“In Chile, we live with many volcanoes, so it is essential to increase the knowledge about them. The more we know, the better we will be able to face emergency situations like eruptions, and even predict them and make timely decisions,” Dr. Chacón said.


Translated by Marcela Contreras