9:45 AM - 10:00 AM
[SVC32-13] Spatiotemporal variations of hydrothermal cracks at volcanic long-period seismic sources
Keywords:Long-period events, Hydrothermal system, Fluid-filled crack, Water-droplet growth model
Long-period (LP) events in hydrothermal systems, which are characterized by damped harmonic oscillations with dilatational onset motions, have been interpreted as the resonances of cracks filled with misty gas (vapor with small water droplets) formed through condensation of high-temperature water vapor (Taguchi et al., JGR, 2018). Nakano and Kumagai (JpGU, 2024a; VSJ, 2024b) analyzed the transient oscillations between onsets and damped oscillations, and attributed their frequency changes to the Doppler effect caused by the movement of crack edges due to source contraction. Moreover, they showed that the typical duration of the frequency changes corresponds to a timescale of water droplets growing from ~0.01 μm to ~1 μm based on the water-droplet growth (WDG) model (Peters and Meyer, IJHMT, 1994), suggesting that the contraction was driven by progressive water condensation. However, the contraction volume and its temporal variation remain quantitatively unexamined. In this study, we investigated the spatiotemporal variations in LP source contraction by analyzing the frequency changes based on the Doppler effect and the WDG model.
We analyzed 33 LP events observed at Kusatsu-Shirane volcano in 1989–1993, all of which had clearly recorded spectral peaks at all stations. We extracted spectral peak ridges using the wavelet transform (Lilly and Olhede, IEEE, 2009). Furthermore, we estimated the movement directions (θ) and velocities (v) of crack edges based on the Doppler effect from the differences in frequencies between the transient and damped oscillations.
For all the events, maximum or minimum frequencies gradually decreased or increased, respectively, to constant oscillation frequencies during the transient oscillations. According to the Doppler effect, these frequency changes suggest that the contraction volume of a crack source gradually decreased from an initial maximum value to zero. The estimated θ indicated that the crack source contracted from both crack edges in two-thirds of the events. In these events, we estimated the source contraction volumes by assuming that v estimates gradually decreased to zero within the transient oscillations. Following Nakano and Kumagai (2024a), we assumed a constant frequency at the crack source during each event, which is possible by keeping a constant crack width and reducing a length-to-aperture ratio with decreasing gas-weight fraction due to progressive condensation. The estimated contraction volumes were consistent with the crack volume changes previously estimated from frequencies, quality factors, and seismic moments in damped oscillations of these events by Nakano et al. (GJI, 2025). Next, we investigated the condensation process using the WDG model. First, following Nakano and Kumagai (2024b), we estimated the time evolution of misty gas volume by simulating the growth of droplets from an initial uniform radius determined by supersaturation to an assumed final radius. However, the volume change simulated with this model was initially minimum and increased over time, which was not consistent with the contraction volumes inferred from the transient oscillations. We then considered distribution functions for the droplet radius in the WDG model. Our simulations indicated that the contraction volume estimated for the LP event on August 8, 1992 was well reproduced by using a Weibull distribution with the shape and scale parameters of 6 and 15, respectively, minimum and maximum radii of 0.06 μm and 6 μm, respectively, and a total number density of 2.5 × 1013 /m3. We found that the contraction volumes for other LP events were also reproduced by adjusting the parameters of the distribution function. This study demonstrates that the transient oscillations are useful to quantify spatiotemporal variations at LP sources.
We analyzed 33 LP events observed at Kusatsu-Shirane volcano in 1989–1993, all of which had clearly recorded spectral peaks at all stations. We extracted spectral peak ridges using the wavelet transform (Lilly and Olhede, IEEE, 2009). Furthermore, we estimated the movement directions (θ) and velocities (v) of crack edges based on the Doppler effect from the differences in frequencies between the transient and damped oscillations.
For all the events, maximum or minimum frequencies gradually decreased or increased, respectively, to constant oscillation frequencies during the transient oscillations. According to the Doppler effect, these frequency changes suggest that the contraction volume of a crack source gradually decreased from an initial maximum value to zero. The estimated θ indicated that the crack source contracted from both crack edges in two-thirds of the events. In these events, we estimated the source contraction volumes by assuming that v estimates gradually decreased to zero within the transient oscillations. Following Nakano and Kumagai (2024a), we assumed a constant frequency at the crack source during each event, which is possible by keeping a constant crack width and reducing a length-to-aperture ratio with decreasing gas-weight fraction due to progressive condensation. The estimated contraction volumes were consistent with the crack volume changes previously estimated from frequencies, quality factors, and seismic moments in damped oscillations of these events by Nakano et al. (GJI, 2025). Next, we investigated the condensation process using the WDG model. First, following Nakano and Kumagai (2024b), we estimated the time evolution of misty gas volume by simulating the growth of droplets from an initial uniform radius determined by supersaturation to an assumed final radius. However, the volume change simulated with this model was initially minimum and increased over time, which was not consistent with the contraction volumes inferred from the transient oscillations. We then considered distribution functions for the droplet radius in the WDG model. Our simulations indicated that the contraction volume estimated for the LP event on August 8, 1992 was well reproduced by using a Weibull distribution with the shape and scale parameters of 6 and 15, respectively, minimum and maximum radii of 0.06 μm and 6 μm, respectively, and a total number density of 2.5 × 1013 /m3. We found that the contraction volumes for other LP events were also reproduced by adjusting the parameters of the distribution function. This study demonstrates that the transient oscillations are useful to quantify spatiotemporal variations at LP sources.