Data-driven imaging of volcanic plumbing systems using machine learning based thermo-barometry on various minerals - Volcanology, physical modelling statistics (VAMOS) FNS project Petrology, volcanology and statistics

VAMOS uses information retrieved from mineral and glass chemistry as the target of numerical and

statistical inversion modelling to link the thermal and chemical architecture of volcanic plumbing

systems and the magnitude of well-studied past eruptions. On the base of these results, we will identify

patterns in the eruptive record heralding eruptions of different magnitudes and use statistical emulation

to estimate the volume of eruptible magma present today within the volcanic plumbing systems of

the investigates volcanoes. To estimate eruption duration, we will invert chemical information on

samples collected during past eruptions (e.g. 2021 eruption of Tajogaite) using numerical and statistical

modelling, to define the best proxies identifying the waning stage of eruptions. VAMOS will establish

a workflow to estimate the magnitude and duration of future eruptions for any volcano on Earth.