Skip to content

Automatic assessment of volcanic activity of Hakone volcano, Japan -Introduction of volcanic unrest index and automatic detection of earthquakes-

Ryo Kurihara1

Abstract

In volcanic areas, multiple types of observations, such as seismic, geodetic, and geochemical observations, are applied. At Hakone Volcano in central Japan, our research institute has primarily conducted many observations. At this volcano, we experienced a small phreatic eruption in 2015 and an episode of unrest in 2019. From these events, several issues have been identified, such as how to interpret multiple types of observational data and how to automatically acquire and analyze the data.  In order to interpret multiple types of data, we have developed a Volcanic Unrest Index tailored to Hakone Volcano, based on Potter et al. (2015). We assign criteria corresponding to indices 1, 2, 3, and 4 to each type of data, such as the duration of swarm earthquakes, the amount of crustal deformation, and the compositional ratios of volcanic gases. The analysis uses data within a 30-day time window and has been designed to automatically output the results every day by shifting the time window forward by one day.  On the other hand, observation data is acquired through manual work, such as on-site measurements of volcanic gas and the picking of earthquake waveforms. Therefore, we are currently developing automatic detection methods for volcanic earthquakes based on the matched filter technique and machine learning. Using these methods, we aim to detect not only volcanic-tectonic earthquakes but also deep low-frequency (or long-period) earthquakes and small earthquakes occurring at very shallow depths.