5:15 PM - 6:45 PM
[SVC31-P10] Forward modeling of source mechanisms for explosion earthquakes at Sakurajima volcano
1. Introduction
At Sakurajima volcano, explosive eruptions of Vulcanian style occur several dozen to several hundred times a year. The hypocenter distribution and source mechanism of the explosion earthquakes that are excited by these eruptions provide information on the behavior of fluids such as ash and gas in the conduit during eruptions, and are important for understanding the eruption mechanism.
Tameguri et al. (2002) decomposed the waveforms of explosion earthquakes at Sakurajima into three phases: the compressional first motion ‘P phase’, the dilatational motion ‘D phase’, and the largest motion ‘LP phase’ (Rayleigh wave), and estimated their mechanisms as an isotropic expansion and cylindrical contraction at a depth of 2 km, and an isotropic expansion and horizontal contraction at a very shallow depth, respectively, by waveform inversion assuming a semi-infinite homogeneous medium. After Miyamachi et al. (2013) estimated a seismic wave velocity structure by analysis of the the seismic experiment data, Nishimura (2022, VSJ) reported that the hypocenters of the P phase are around 300 m depth beneath the crater bottom and Hasegawa and Nishimura (2023, VSJ) pointed out that a simpler source process possibly explains the multiple waveforms, when considering the topography and the velocity structure.
In this study, we estimated the source mechanism of the main motion, LP phase, of the Sakurajima explosion earthquakes, comparing it with the theoretical waveforms considering the topography and the velocity structure.
2. Data and Method
We use waveforms of the explosion earthquakes on November 20, 21, December 2, 5, 2022, observed at three seismic stations with high signal-to-noise ratios (Akobara, Amidagawa, and Nabeyama) of six JMA stations. Because the LP phase is dominant at 0.5-2 Hz, a 0.2-2 Hz bandpass filter is applied to the waveform after removing the seismometer properties. The LP phase is observed in 3 s from the P-phase arrival time. Since the transverse component is half smaller than the radial component within 2-3 s from the P-phase arrival time, we assume symmetry about the vertical axis through the crater center and analyze only the vertical and radial components in the same time window.
Theoretical waveforms are calculated using the finite difference method. The topography is assumed to be a smooth mountain with a height of 1 km and a spread of 4 km from the summit, with a crater of 500 m in a diameter and 250 m in depth (750 m above sea level) at the summit, based on a topographic map of Sakurajima. The velocity structure is assumed to be a four-layer horizontal structure with a linear gradient in each layer, following Miyamachi et al. (2013). The epicenter is fixed just below the crater, and the source depth is varied from the crater bottom to the sea level at 100 m intervals to find the optimal value. The source time function is a Gaussian function. The type of source mechanism is chosen from isotropic, pipe, horizontal crack, and downward single force.
3. Results
In all four events, the theoretical waveforms agree well with the observed ones in the case of vertical expanding cracks or downward single forces but not in the case of isotropic or pipe sources. A source time function with a pulse width of 1.3 s or less can generally explain the observed waveforms.
In the case of the crack model, the optimal hypocentral location is 300 m above sea level (450 m depth from the crater bottom), and the seismic moment is about 10^11-10^12 Nm when the pulse width is assumed to be 1 s. This value roughly corresponds to the moment of the LP source estimated by Tameguri et al. (2013). In the case of a downward single force, the optimal hypocentral location is 700 m above sea level (50 m below the crater bottom), and the force value is about 10^8 N when the pulse width is 1 s. In both cases, there is an uncertainty of about 200 m in the source depth.
The direct P wave from the model estimated from the LP phase arrives 0.5 s after the P phase in the observation. These results indicate that a vertically dominant seismic mechanism, about 0.5 s after the triggering P-phase source, excites the LP phase in explosion earthquakes at Sakurajima volcano.
Acknowledgements
We use seismic waveform data from the Japan Meteorological Agency Volcano Observation Network.
At Sakurajima volcano, explosive eruptions of Vulcanian style occur several dozen to several hundred times a year. The hypocenter distribution and source mechanism of the explosion earthquakes that are excited by these eruptions provide information on the behavior of fluids such as ash and gas in the conduit during eruptions, and are important for understanding the eruption mechanism.
Tameguri et al. (2002) decomposed the waveforms of explosion earthquakes at Sakurajima into three phases: the compressional first motion ‘P phase’, the dilatational motion ‘D phase’, and the largest motion ‘LP phase’ (Rayleigh wave), and estimated their mechanisms as an isotropic expansion and cylindrical contraction at a depth of 2 km, and an isotropic expansion and horizontal contraction at a very shallow depth, respectively, by waveform inversion assuming a semi-infinite homogeneous medium. After Miyamachi et al. (2013) estimated a seismic wave velocity structure by analysis of the the seismic experiment data, Nishimura (2022, VSJ) reported that the hypocenters of the P phase are around 300 m depth beneath the crater bottom and Hasegawa and Nishimura (2023, VSJ) pointed out that a simpler source process possibly explains the multiple waveforms, when considering the topography and the velocity structure.
In this study, we estimated the source mechanism of the main motion, LP phase, of the Sakurajima explosion earthquakes, comparing it with the theoretical waveforms considering the topography and the velocity structure.
2. Data and Method
We use waveforms of the explosion earthquakes on November 20, 21, December 2, 5, 2022, observed at three seismic stations with high signal-to-noise ratios (Akobara, Amidagawa, and Nabeyama) of six JMA stations. Because the LP phase is dominant at 0.5-2 Hz, a 0.2-2 Hz bandpass filter is applied to the waveform after removing the seismometer properties. The LP phase is observed in 3 s from the P-phase arrival time. Since the transverse component is half smaller than the radial component within 2-3 s from the P-phase arrival time, we assume symmetry about the vertical axis through the crater center and analyze only the vertical and radial components in the same time window.
Theoretical waveforms are calculated using the finite difference method. The topography is assumed to be a smooth mountain with a height of 1 km and a spread of 4 km from the summit, with a crater of 500 m in a diameter and 250 m in depth (750 m above sea level) at the summit, based on a topographic map of Sakurajima. The velocity structure is assumed to be a four-layer horizontal structure with a linear gradient in each layer, following Miyamachi et al. (2013). The epicenter is fixed just below the crater, and the source depth is varied from the crater bottom to the sea level at 100 m intervals to find the optimal value. The source time function is a Gaussian function. The type of source mechanism is chosen from isotropic, pipe, horizontal crack, and downward single force.
3. Results
In all four events, the theoretical waveforms agree well with the observed ones in the case of vertical expanding cracks or downward single forces but not in the case of isotropic or pipe sources. A source time function with a pulse width of 1.3 s or less can generally explain the observed waveforms.
In the case of the crack model, the optimal hypocentral location is 300 m above sea level (450 m depth from the crater bottom), and the seismic moment is about 10^11-10^12 Nm when the pulse width is assumed to be 1 s. This value roughly corresponds to the moment of the LP source estimated by Tameguri et al. (2013). In the case of a downward single force, the optimal hypocentral location is 700 m above sea level (50 m below the crater bottom), and the force value is about 10^8 N when the pulse width is 1 s. In both cases, there is an uncertainty of about 200 m in the source depth.
The direct P wave from the model estimated from the LP phase arrives 0.5 s after the P phase in the observation. These results indicate that a vertically dominant seismic mechanism, about 0.5 s after the triggering P-phase source, excites the LP phase in explosion earthquakes at Sakurajima volcano.
Acknowledgements
We use seismic waveform data from the Japan Meteorological Agency Volcano Observation Network.