Name Surname
Leveraging Traditional and Machine Learning Approaches for Petrological Monitoring and Eruptive Style Forecasting in Open Vent Volcanic Systems
Main Supervisor: Réka Lukács (ELTE)
Co-Supervisors: Rosa Anna Corsaro (INGV) | Chiara Cristiani (Dipartimento della Protezione civile)
Objectives
The project aim at investigating the potential of machine learning, including generative AI approaches, to model petrological data in near real-time volcanic monitoring scenarios. The main objective is to correlate petrological signals with geophysical and geochemical monitoring data to enhance volcano monitoring and eruptive style forecasting. The primary goal is to develop rapid and robust workflows for correlating different sources of information (petrologic, geophysical, and geochemical), provide AI driven summarization, and possibly go in the direction of forecasting eruptive intensities and style modulations. The focus will be on eruptive sequences at open vent volcanic systems like the Etna volcano, which offers an extensive record of sampling and monitoring (both geophysical and geochemical) with already available data.
Epected Results
- Probing the potentials and limitations of data-driven machine learning-based modelling tools to identify critical petrological signals, enhancing petrological monitoring and eruptive style forecasting;
- Develop ML based workflows of petrological monitoring to support near real-time assessments of magma dynamics, enabling more accurate predictions of successive eruption intensities and style modulations.