11:00 AM - 11:15 AM
[SVC31-08] The source mechanism of explosion earthquakes at Sakurajima volcano observed from the very-long-period band
1. Introduction
Explosion earthquakes occur with explosive eruptions and have been observed at various volcanoes. Among them, Sakurajima is the most active volcano in Japan, and explosive earthquakes occur frequently. Explosion earthquakes were explained by a model using a cylindrical source at a depth of 1-2 km obtained by analysis for the very long-period (VLP) band (Uhira and Takeo, 1994). On the other hand, Green's function calculation in the waveform inversion analysis used in this study assumes an anelastic layered halfspace structure and does not consider the effects of topography. Considering the actual topography, the source location and mechanism may change. VLP signals have been implicated in source volume changes caused by the movement of fluids such as magma and volcanic gases in the subsurface (e.g., Chouet and Matoza, 2013). Understanding the source mechanism of the VLP signal of explosive earthquakes may lead to understanding the properties of the magma involved in the eruption and the eruptive process. In this study, we analyzed the data on the eruption at 20:05 on 24 July 2022, when the volcanic ejecta was observed scattering beyond 2.4 km on the monitoring camera, and the volcanic plume rose more than 2000m above the crater rim.
2. VLP event waveform and Method
The amplitude spectrum of the analysis event showed a flat spectrum at 0.4-1Hz and 2-4Hz on the high-frequency side with a boundary of 0.25Hz. On the other hand, the low-frequency side shows a flat spectrum from 0.25 Hz to 0.125 Hz and drops toward 0.067 Hz on the lower frequency side. In this study, a band-pass filter of 0.067-0.25Hz on the low-frequency side was applied. In this study, a band-pass filter of 0.067-0.25 Hz was applied to observed waveforms of the eruption earthquake, and the waveform inversion method (Ohminato et al., 1998; Auger et al., 2006) was used to perform source mechanism analysis. A grid search was performed for the source position by dividing the topographic data from the digital elevation model into 40m cube meshes. The source location was determined by grid search, and the search area was defined as 1600 m east-west, 1400 m north-south, and 3000 m below sea level from the surface in the horizontal plane surrounding the crater, based on the distribution of source locations of explosive earthquakes (Nishimura, 2022). A point source is assumed for five models: six moment tensor and three single force, six moment tensor, three single force, tensile crack, and cylindrical source.
3. Results and Discussion
The mechanism that explained the waveforms best among the assumed models was the six-component moment tensor mechanism, and the source location at that time was located about 200 m below the ground surface near the mouth of the Minamidake crater, shallower than in the previous study (Uhira and Takeo, 1994). The cylindrical source estimated in the previous study had the worst waveform fitting of the assumed mechanisms. The moment tensor six-component source time function had predominant amplitude values for the diagonal components, with minimal amplitude values for the off-diagonal components compared to the diagonal components. The time-series variation of the principal components of the source time function was generally stable, with no significant change with time. The principal axes of the six moment components are (3.94,1.39,1.15)×1013 N m, which is also about 1.2:1:3.4. The elastic constants are λ=2μ if the medium, rock, is close to the liquidus temperatures, and λ=μ otherwise. The elastic constants are considered for two cases, λ=μ and λ=2μ. For λ=μ, one sub-horizontal aperture crack is considered; for λ=2μ, two aperture cracks are estimated: a sub-horizontal aperture crack and a sub-vertical one with a 21% contribution to the former. The obtained source time function shows the deflation-inflation sequence. It is inferred that a phenomenon similar to the bubble growth model of Chouet et al. (2006), in which a volatile bubble in the magma expands after contraction and the pressure recovers, is taking place.
Acknowledgment: We used broadband seismic data provided by the Osumi office of River and National Highway, Ministry of Land, Infrastructure, Transport and Tourism (MLIT). This study was supported by ERI JURP 2022-M-15 in Earthquake Research Institute, the University of Tokyo.
Explosion earthquakes occur with explosive eruptions and have been observed at various volcanoes. Among them, Sakurajima is the most active volcano in Japan, and explosive earthquakes occur frequently. Explosion earthquakes were explained by a model using a cylindrical source at a depth of 1-2 km obtained by analysis for the very long-period (VLP) band (Uhira and Takeo, 1994). On the other hand, Green's function calculation in the waveform inversion analysis used in this study assumes an anelastic layered halfspace structure and does not consider the effects of topography. Considering the actual topography, the source location and mechanism may change. VLP signals have been implicated in source volume changes caused by the movement of fluids such as magma and volcanic gases in the subsurface (e.g., Chouet and Matoza, 2013). Understanding the source mechanism of the VLP signal of explosive earthquakes may lead to understanding the properties of the magma involved in the eruption and the eruptive process. In this study, we analyzed the data on the eruption at 20:05 on 24 July 2022, when the volcanic ejecta was observed scattering beyond 2.4 km on the monitoring camera, and the volcanic plume rose more than 2000m above the crater rim.
2. VLP event waveform and Method
The amplitude spectrum of the analysis event showed a flat spectrum at 0.4-1Hz and 2-4Hz on the high-frequency side with a boundary of 0.25Hz. On the other hand, the low-frequency side shows a flat spectrum from 0.25 Hz to 0.125 Hz and drops toward 0.067 Hz on the lower frequency side. In this study, a band-pass filter of 0.067-0.25Hz on the low-frequency side was applied. In this study, a band-pass filter of 0.067-0.25 Hz was applied to observed waveforms of the eruption earthquake, and the waveform inversion method (Ohminato et al., 1998; Auger et al., 2006) was used to perform source mechanism analysis. A grid search was performed for the source position by dividing the topographic data from the digital elevation model into 40m cube meshes. The source location was determined by grid search, and the search area was defined as 1600 m east-west, 1400 m north-south, and 3000 m below sea level from the surface in the horizontal plane surrounding the crater, based on the distribution of source locations of explosive earthquakes (Nishimura, 2022). A point source is assumed for five models: six moment tensor and three single force, six moment tensor, three single force, tensile crack, and cylindrical source.
3. Results and Discussion
The mechanism that explained the waveforms best among the assumed models was the six-component moment tensor mechanism, and the source location at that time was located about 200 m below the ground surface near the mouth of the Minamidake crater, shallower than in the previous study (Uhira and Takeo, 1994). The cylindrical source estimated in the previous study had the worst waveform fitting of the assumed mechanisms. The moment tensor six-component source time function had predominant amplitude values for the diagonal components, with minimal amplitude values for the off-diagonal components compared to the diagonal components. The time-series variation of the principal components of the source time function was generally stable, with no significant change with time. The principal axes of the six moment components are (3.94,1.39,1.15)×1013 N m, which is also about 1.2:1:3.4. The elastic constants are λ=2μ if the medium, rock, is close to the liquidus temperatures, and λ=μ otherwise. The elastic constants are considered for two cases, λ=μ and λ=2μ. For λ=μ, one sub-horizontal aperture crack is considered; for λ=2μ, two aperture cracks are estimated: a sub-horizontal aperture crack and a sub-vertical one with a 21% contribution to the former. The obtained source time function shows the deflation-inflation sequence. It is inferred that a phenomenon similar to the bubble growth model of Chouet et al. (2006), in which a volatile bubble in the magma expands after contraction and the pressure recovers, is taking place.
Acknowledgment: We used broadband seismic data provided by the Osumi office of River and National Highway, Ministry of Land, Infrastructure, Transport and Tourism (MLIT). This study was supported by ERI JURP 2022-M-15 in Earthquake Research Institute, the University of Tokyo.