Time-lapse monitoring of seismic velocity associated with 2011
Shinmoe-dake eruption using seismic interferometry: an extended Kalman
filter approach
Abstract
Seismic interferometry is a powerful tool to monitor the seismic
velocity change associated with volcanic eruptions. For the monitoring,
changes in seismic velocity with environmental origins (such as
precipitation) are problematic. In order to model the environmental
effects, we propose a new technique based on a state-space model. An
extended Kalman filter estimates seismic velocity changes as state
variables, with a first-order approximation of the stretching method. We
apply this technique to three-component seismic records in order to
detect the seismic velocity change associated with the Shinmoe-dake
eruptions in 2011 and 2018. First, ambient noise cross-correlations were
calculated from May 2010 to April 2018. We also modeled seismic velocity
changes resulting from precipitation and the 2016 Kumamoto earthquake,
with exponential type responses. Most of the results show no significant
changes associated with the eruptions, although gradual inflation of the
magma reservoir preceded the 2011 eruption by one year. The observed low
sensitivity to static stress changes suggests that the fraction of
geofluid and crack density at about 1 km depth is small, and the crack
shapes could be circular. Only one station pair west of the crater shows
the significant drop associated with the eruption in 2011. The gradual
drop of seismic velocity up to 0.05% preceded the eruption by one
month. When the gradual drop began, volcanic tremors were activated at
about 2 km depth. These observations suggest that the drop could be
caused by damage accumulation due to vertical magma migration beneath
the summit.