5:15 PM - 6:45 PM
[SVC26-P05] Volcanic long-period events triggered by progressive water condensation
Keywords:Long-period events, Condensation process in hydrothermal system, fluid-filled crack
Volcanic long-period (LP) events in hydrothermal systems show waveforms with damped oscillations and multiple spectral peaks. Moreover, transient oscillations are observed for about 1 s between initial motion and damped oscillations, and the peak frequency (f) is constant throughout the event. The damped oscillations are interpreted to be generated by the resonance of a crack filled with misty gas (vapor with small water droplets) in hydrothermal systems (e.g., Kumagai and Chouet, JGR, 2000), but the physical process generating transient oscillations is yet to be clarified. Taguchi et al. (JGR, 2018) proposed that LP events are excited by rapid condensation of high-temperature water vapor from a crack edge. However, it is not known what kind of relationship exists between instantaneous condensation and transient oscillations. Therefore, we investigated the physical process during transient oscillations by analyzing their waveform characteristics.
In the running spectrum of an LP event at Kusatsu-Shirane volcano on August 8, 1992 (the lowest peak frequency of 4.4 Hz), large amplitude signals were seen in a wide frequency range during the onset portion, indicating that these signals represent the response of a pulse-like excitation. Then, we extracted only this response by applying a bandpass filter in 1–2.5 Hz. We found that the synthetic waveforms calculated by using the moment function having a step-like decrease with a duration of 0.5 s and seismic moment (M0) of 109 Nm matched well to the observed waveforms. The volume change (ΔVe) corresponding to the estimated M0 was about 1 m3. The crack volume (V) and gas-weight fraction (n) of misty gas in the crack during the damped oscillations were estimated to be 200 m3 and 0.4, respectively (Nakano and Kumagai, JpGU, 2023), which correspond to the vapor volume of 400 m3 just before the pulse-like excitation. Therefore, the volume change (ΔV) before and after the pulse-like excitation (i.e., during the transient oscillations) was 200 m3, indicating that ΔVe is much smaller than ΔV. This difference can be explained by considering progressive condensation so that the crack continued to contract even after the pulse-like excitation. As mentioned above, f was constant throughout the event. However, if condensation progresses during the transient oscillations, f can be changed as n decreases due to condensation. Therefore, we investigated the conditions necessary for progressive condensation while keeping f constant. According to the analytical formula (Maeda and Kumagai, GJI, 2017), f depends on sound speed a, crack length L, and crack stiffness C. Here, C = 3(a/α)2(ρf/ρs)(L/d), where α and ρs are the P-wave velocity and density of the surrounding rock, ρf is fluid density, and d is crack thickness. Assuming that α and ρs are constant and there is no mass change of water in the crack during the event, a, ρf, and V are uniquely determined by n. L and L/d are determined depending on V and crack width W. Our numerical investigations assuming that W is constant based on the results of Nakano and Kumagai (2023) indicated that f remains constant regardless of n when L/d is inversely proportional to n.
The above results suggest that the LP event was excited by the following process. Pressure and temperature in the crack should have been constant during condensation to keep saturation conditions. After some vapor rapidly condensed at a crack edge, V decreased with decreasing n as condensation progressed, in which the increase in fluid pressure was compensated by deformation of the crack with increasing L/d to balance lithostatic pressure. In this way, pressure and temperature were kept constant during the transient oscillations, and as a result, f remained constant because L/d increased with decreasing n. Our results indicate that progressive water condensation in hydrothermal systems plays an important role in triggering LP events.
In the running spectrum of an LP event at Kusatsu-Shirane volcano on August 8, 1992 (the lowest peak frequency of 4.4 Hz), large amplitude signals were seen in a wide frequency range during the onset portion, indicating that these signals represent the response of a pulse-like excitation. Then, we extracted only this response by applying a bandpass filter in 1–2.5 Hz. We found that the synthetic waveforms calculated by using the moment function having a step-like decrease with a duration of 0.5 s and seismic moment (M0) of 109 Nm matched well to the observed waveforms. The volume change (ΔVe) corresponding to the estimated M0 was about 1 m3. The crack volume (V) and gas-weight fraction (n) of misty gas in the crack during the damped oscillations were estimated to be 200 m3 and 0.4, respectively (Nakano and Kumagai, JpGU, 2023), which correspond to the vapor volume of 400 m3 just before the pulse-like excitation. Therefore, the volume change (ΔV) before and after the pulse-like excitation (i.e., during the transient oscillations) was 200 m3, indicating that ΔVe is much smaller than ΔV. This difference can be explained by considering progressive condensation so that the crack continued to contract even after the pulse-like excitation. As mentioned above, f was constant throughout the event. However, if condensation progresses during the transient oscillations, f can be changed as n decreases due to condensation. Therefore, we investigated the conditions necessary for progressive condensation while keeping f constant. According to the analytical formula (Maeda and Kumagai, GJI, 2017), f depends on sound speed a, crack length L, and crack stiffness C. Here, C = 3(a/α)2(ρf/ρs)(L/d), where α and ρs are the P-wave velocity and density of the surrounding rock, ρf is fluid density, and d is crack thickness. Assuming that α and ρs are constant and there is no mass change of water in the crack during the event, a, ρf, and V are uniquely determined by n. L and L/d are determined depending on V and crack width W. Our numerical investigations assuming that W is constant based on the results of Nakano and Kumagai (2023) indicated that f remains constant regardless of n when L/d is inversely proportional to n.
The above results suggest that the LP event was excited by the following process. Pressure and temperature in the crack should have been constant during condensation to keep saturation conditions. After some vapor rapidly condensed at a crack edge, V decreased with decreasing n as condensation progressed, in which the increase in fluid pressure was compensated by deformation of the crack with increasing L/d to balance lithostatic pressure. In this way, pressure and temperature were kept constant during the transient oscillations, and as a result, f remained constant because L/d increased with decreasing n. Our results indicate that progressive water condensation in hydrothermal systems plays an important role in triggering LP events.