Machine learning in Volcanology

Magmatic processes controlling volcanic eruptions take place at inaccessible depth of tens of kilometers within the Earth’s lithosphere and control the frequency, magnitude and intensity of volcanic eruptions. Because about 1 billion people life around active volcanoes, understanding magmatic processes is not only a scientific endeavor but also of paramount importance for the mitigation of volcanic hazard. As we cannot directly access the plumbing system of volcanoes, all information we gather on magmatic processes are indirect, multidimensional and of complex nature. It is for this reason that machine learning applications to the study of volcanic system is becoming increasingly important.

70% of the active volcanoes on Earth are poorly studied. I will present an application of machine learning that allowed us to unveil relationships between the explosivity of volcanoes and easily measurable parameters, which paves the way for the rapid assessment of volcanic hazard at poorly studied volcanoes.

About Luca Caricchi: I am petrologist and volcanologist interested in the processes controlling the transport, accumulation, and eruption of magma on Earth. I use a combination of classic petrology, modelling, machine learning and statistics to quantify magma flux through the lithosphere and unveil its control on the frequency and magnitude of volcanic eruptions and the formation of ore deposits.