A quantitative analysis of monitoring signals and eruption explosivity

Abstract

Volcanic hazards are dependent on eruption size and explosivity, thus, the forecasts of these is crucial for emergency management decisions. Monitoring a volcano potentially offers valuable insights to assess when, where, and how explosive a future eruption might be. Data are collected from various monitoring equipment, such as seismometers, tiltmeters, and thermometers, that are installed at different locations and distances, from near-vent to satellite. However, establishing direct links between monitoring signals and even eruption onset remains challenging, especially for volcanoes lacking recent eruptions or without monitoring equipment installed prior to eruption. This challenge extends to eruption explosivity, where establishing links becomes even more difficult. The Global Volcanism Program (GVP) has compiled monitoring data in bulletin reports recorded by observatories and research institutions. These reports start from 1968 and summarise volcanic activity that occurs before, during and after an eruption. Importantly, these reports include descriptions about activity and/or raw data (number of earthquakes, frequencies, plots of the seismic signals or displacements on tiltmeters) from various monitoring equipment, providing a general understanding of precursor activities preceding an eruption. This is potentially key information for forecasting eruption explosivity. This study aims to establish a quantitative link between monitoring signals and eruption explosivity across multiple volcanoes. Data are compiled from 23 volcanoes worldwide, utilising information from the Global Volcanism Program (GVP) database and local volcano observatory reports where accessible. The different descriptions obtained by each class of monitoring equipment—whether seismic, thermal, deformation, SO2 fluxes, or crater alterations—will be statistically categorized and calibrated into predictor variables to be used in machine learning algorithms. We hope to develop a procedure for estimating the explosivity of the next eruption, as a step towards statistically forecasting future eruption styles.