10:45 AM - 11:00 AM
[SVC26-07] Precursor seismic activity ditected approximately 15 minutes before the start of the major eruption of Hunga Tonga=Hunga Ha'apai volcano on January 15, 2022
Keywords:Volcanic eruption, Hunga Tonga=Hunga Ha'apai, Volcanic earthquake, Volcano monitoring, Eruption sequence
The massive eruption on January 15, 2022, at Hunga Tonga=Hunga Ha'apai (HTHH) volcano in Tonga was the largest in recent history, with plumes exceeding 50 km in height and tsunami that impacted worldwide, including Japan, over 7000 km away. Signals from the eruption were captured by various observation instruments around the world, with which studies have been performed to reveal the effects of this eruption on the atmosphere and the spatio-temporal evolution of the waves. However, questions remain about the onset of the largest eruption, which began around 04:00 UTC on January 15, and the precursory processes.
Cronin et al. (2023, IAVCEI) noted a small eruption at 02:57 on January 15 and considered it the beginning of the main eruption. However, our analyses of satellite images and seismic waveforms showed that this small eruption was not the beginning but a part of the sporadic eruptions that had continued from the previous day (Horiuchi et al. JpGU 2023). Besides, Borrero et al. (2022, Pure Appl. Geophys.), who had conducted a domestic tsunami survey and interviews immediately after the eruption, noted the start time of the eruption as 03:47 but did not report any specific evidence of this timing. In this study, we analyze the signals starting around 03:47 at two seismic stations, both about 750 km away from the volcano, to clarify when the subsurface events associated with the large-scale eruption started.
The data used were three-component broadband seismograph data from MSVF (in Fiji, 758 km from the crater) and FUTU (in Wallis-Futuna, 752 km) downloaded from IRIS Web Services. After correcting for the instrument responses, the data were low-pass filtered below 0.15 Hz and decimated to 1 Hz.
First, we assume that the target signal is a Rayleigh wave. Considering the Rayleigh wave particle motion, the horizontal velocity component and the Hilbert-transformed vertical velocity component should have a phase shift of either zero or 180 degrees. Therefore, the absolute value of the cross-correlation between the two should be large near zero delay time. We examined the cross-spectral density and coherence of the east-west velocity and Hilbert-transformed vertical velocity and found that the signal power and correlation were high in the frequency range of 0.04-0.08 Hz. This means that the oscillations are due to Rayleigh waves dominating in 0.04-0.08 Hz. Furthermore, based on the ratios of the cross-correlation function of the Hilbert-transformed vertical velocity to the east-west component and that to the north-south component, we found that at both MSVF and FUTU stations, the Rayleigh wave was from the direction of the HTHH volcano.
Next, we compared the MSVF-FUTU arrival time differences of the signals around 03:47 and those from the M5.8 earthquake that occurred at 04:14:45 beneath HTHH. We focused on the frequency band of 0.07-0.09 Hz, where the signals of both events had significant power. We detected the arrival times in the cross-correlation functions of the Hilbert-transformed vertical velocity to the east-west component at MSVF and to the north-south component at FUTU. We found that the signal arrival from the M5.8 earthquake at MSVF was 28-32 seconds (average 31 seconds) after FUTUI, and the arrival of the oscillation around 03:47 at MSVF was 23-40 seconds (average 33 seconds) after FUTU. The discrepancy in the arrival time differences was within 10 seconds, which is insignificant since the cross-correlation function was calculated in a 64-second-long window, and the frequency band of interest corresponds to periods longer than 10 seconds.
Both direction and arrival time differences support that the 03:47 oscillation is a Rayleigh wave from HTHH, which we regard as a precursor of the huge eruption. Besides, the FUTU station records the peak of the M5.8 signal at 04:18:45, which is 4 minutes after the reported onset. Then, we estimate the onset of the precursor a little after 03:46 using the peak recorded at FUTU at 03:50:13.
Referring to the seismograph variations from the previous day, the signal discovered in this study was significant in its magnitude and high vertical-horizontal correlation. On the other hand, no apparent change was observed in the satellite images. Our results indicate that some events leading to the large-scale eruption on January 15 occurred underground or on the seafloor around 03:44, supporting the witness's report by Borrero et al. (2022).
Cronin et al. (2023, IAVCEI) noted a small eruption at 02:57 on January 15 and considered it the beginning of the main eruption. However, our analyses of satellite images and seismic waveforms showed that this small eruption was not the beginning but a part of the sporadic eruptions that had continued from the previous day (Horiuchi et al. JpGU 2023). Besides, Borrero et al. (2022, Pure Appl. Geophys.), who had conducted a domestic tsunami survey and interviews immediately after the eruption, noted the start time of the eruption as 03:47 but did not report any specific evidence of this timing. In this study, we analyze the signals starting around 03:47 at two seismic stations, both about 750 km away from the volcano, to clarify when the subsurface events associated with the large-scale eruption started.
The data used were three-component broadband seismograph data from MSVF (in Fiji, 758 km from the crater) and FUTU (in Wallis-Futuna, 752 km) downloaded from IRIS Web Services. After correcting for the instrument responses, the data were low-pass filtered below 0.15 Hz and decimated to 1 Hz.
First, we assume that the target signal is a Rayleigh wave. Considering the Rayleigh wave particle motion, the horizontal velocity component and the Hilbert-transformed vertical velocity component should have a phase shift of either zero or 180 degrees. Therefore, the absolute value of the cross-correlation between the two should be large near zero delay time. We examined the cross-spectral density and coherence of the east-west velocity and Hilbert-transformed vertical velocity and found that the signal power and correlation were high in the frequency range of 0.04-0.08 Hz. This means that the oscillations are due to Rayleigh waves dominating in 0.04-0.08 Hz. Furthermore, based on the ratios of the cross-correlation function of the Hilbert-transformed vertical velocity to the east-west component and that to the north-south component, we found that at both MSVF and FUTU stations, the Rayleigh wave was from the direction of the HTHH volcano.
Next, we compared the MSVF-FUTU arrival time differences of the signals around 03:47 and those from the M5.8 earthquake that occurred at 04:14:45 beneath HTHH. We focused on the frequency band of 0.07-0.09 Hz, where the signals of both events had significant power. We detected the arrival times in the cross-correlation functions of the Hilbert-transformed vertical velocity to the east-west component at MSVF and to the north-south component at FUTU. We found that the signal arrival from the M5.8 earthquake at MSVF was 28-32 seconds (average 31 seconds) after FUTUI, and the arrival of the oscillation around 03:47 at MSVF was 23-40 seconds (average 33 seconds) after FUTU. The discrepancy in the arrival time differences was within 10 seconds, which is insignificant since the cross-correlation function was calculated in a 64-second-long window, and the frequency band of interest corresponds to periods longer than 10 seconds.
Both direction and arrival time differences support that the 03:47 oscillation is a Rayleigh wave from HTHH, which we regard as a precursor of the huge eruption. Besides, the FUTU station records the peak of the M5.8 signal at 04:18:45, which is 4 minutes after the reported onset. Then, we estimate the onset of the precursor a little after 03:46 using the peak recorded at FUTU at 03:50:13.
Referring to the seismograph variations from the previous day, the signal discovered in this study was significant in its magnitude and high vertical-horizontal correlation. On the other hand, no apparent change was observed in the satellite images. Our results indicate that some events leading to the large-scale eruption on January 15 occurred underground or on the seafloor around 03:44, supporting the witness's report by Borrero et al. (2022).