10:00 AM - 10:15 AM
[SVC31-05] Temporal variations in source properties of long-period seismic events and their relations with magmatic degassing
Keywords:volcano seismicity, Kusatsu-Shirane volcano, waveform inversion
Long-period (LP) events, which are observed commonly in shallow parts at active volcanoes, are interpreted as the oscillations of a fluid-filled crack. Their waveforms are characterized by clear damped oscillations with the oscillation frequency (f), quality factor (Q), and seismic moment (M0), which depend on source properties of the crack. The observed values of f, Q, and M0 often show temporal variations. Taguchi et al. (2018) suggested that LP events at Kusatsu-Shirane volcano were generated by resonances of a crack filled with misty gas, which were triggered by volumetric changes associated with water condensation from vapor derived from magma degassing. Temporal variations in f, Q, and M0 may be related to magma degassing, but its relation remains unknown. To clarify this relation, we developed an efficient method to estimate M0, which has not been systematically determined in previous studies, and estimated the mass of vapor intruded into the crack from f, Q, and M0. We analyzed 202 LP events observed in 1990−1993 at Kusatsu-Shirane volcano, and investigated the LP source processes related to magmatic degassing at this volcano.
We used f and Q values of the lowest frequency peaks in the individual LP waveforms. We performed inversion using mean amplitudes of LP waveforms band-pass filtered around the lowest peaks to estimate M0 values of the individual events. Although full-waveform inversion was only applicable to low-frequency waveforms up to around 1 Hz due to the difficulty to fit waveforms including phases, which are more strongly affected by heterogenous structures than amplitudes, our inversion approach provided stable results using high-frequency waveforms up to 6 Hz. Furthermore, we developed a new method to estimate crack size and acoustic properties of misty gas in the crack from f, Q, and M0 by using the analytical and empirical formulas for f (Maeda and Kumagai, 2017) and Q (Taguchi et al., 2021) of crack resonances, respectively, under the assumption that crack length L is proportional to aperture d and its change Δd. We then estimated the mass of H2O of misty gas (mH2O) in the crack for the individual LP events.
We found that f -1 was positively correlated with mH2O. If vapor intruded into the crack was derived from magmatic degassing, mH2O is proportional to the mass of degassed magma. The observed temporal variations in f in 1990−1993 were divided into two phases: short-term fluctuations in 1990 and systematic decreasing and increasing trends in 1991 and 1992−1993. Based on our estimated relation of f to mH2O, we interpreted magma degassing in these periods as follows. In 1991 and 1992−1993, the degassed magma gradually increased and decreased, but it rapidly fluctuated in 1990, suggesting that magma supply style was different in these periods. In this way, we can infer magma activity from observed temporal variations in f of LP events, which can be easily determined by spectral analysis compared to Q and M0. Although there was no eruption in 1990−1993, provision of larger amount of vapor into the crack may have led to a phreatic eruption. To monitor f of LP events is thus useful to infer magmatic activity and to assess the risk of eruptions.
We used f and Q values of the lowest frequency peaks in the individual LP waveforms. We performed inversion using mean amplitudes of LP waveforms band-pass filtered around the lowest peaks to estimate M0 values of the individual events. Although full-waveform inversion was only applicable to low-frequency waveforms up to around 1 Hz due to the difficulty to fit waveforms including phases, which are more strongly affected by heterogenous structures than amplitudes, our inversion approach provided stable results using high-frequency waveforms up to 6 Hz. Furthermore, we developed a new method to estimate crack size and acoustic properties of misty gas in the crack from f, Q, and M0 by using the analytical and empirical formulas for f (Maeda and Kumagai, 2017) and Q (Taguchi et al., 2021) of crack resonances, respectively, under the assumption that crack length L is proportional to aperture d and its change Δd. We then estimated the mass of H2O of misty gas (mH2O) in the crack for the individual LP events.
We found that f -1 was positively correlated with mH2O. If vapor intruded into the crack was derived from magmatic degassing, mH2O is proportional to the mass of degassed magma. The observed temporal variations in f in 1990−1993 were divided into two phases: short-term fluctuations in 1990 and systematic decreasing and increasing trends in 1991 and 1992−1993. Based on our estimated relation of f to mH2O, we interpreted magma degassing in these periods as follows. In 1991 and 1992−1993, the degassed magma gradually increased and decreased, but it rapidly fluctuated in 1990, suggesting that magma supply style was different in these periods. In this way, we can infer magma activity from observed temporal variations in f of LP events, which can be easily determined by spectral analysis compared to Q and M0. Although there was no eruption in 1990−1993, provision of larger amount of vapor into the crack may have led to a phreatic eruption. To monitor f of LP events is thus useful to infer magmatic activity and to assess the risk of eruptions.