Dynamic strain anomalies detection at Stromboli before 2019 vulcanian explosions using machine learning.

Characterizing the dynamics of explosive activity is impelling to build tools for hazard assessment at open-conduit volcanoes: machine learning techniques are a feasible choice. During the 2019 Stromboli experienced two paroxysmal eruptions, occurred in two different volcanic phases, giving us the possibility to conceive and test an early-warning algorithm on a real use case. Among the changes observed in the weeks preceding the first paroxysm, one of the most significant is represented by the shape variation of the ordinary minor explosions, filtered in the very long period (2--50 s) band, recorded by a Sacks-Evertson strainmeter installed nearby. Starting from these observations, the usage of two independent methods to classify strain transients falling in the ultra long period (50--200 s) frequency band, allowed us to validate the robustness of the approach. This classification leads us to establish a link between VLP and ULP shape variation and volcanic activity, especially related to the first paroxysm. Previous warning times at Stromboli are of a few minutes only, while our approach could permit to anticipate this time to several days by detecting medium-term shape changes.