Switching from seismic to seismo-acoustic harmonic tremor at a transition of eruptive activity during the Shinmoe-dake 2011 eruption
This paper reports a sequence of harmonic tremor observed during the 2011 eruption of Shinmoe-dake volcano, Kyushu, Japan. The main eruptive activity started with subplinian eruptions, followed by lava effusion. Harmonic tremor was observed as seismic waves during the final stage of the effusive eruption. The tremor observed at this stage had unclear and fluctuating harmonic modes. In the atmosphere, however, many impulsive acoustic waves indicating small surface explosions were observed. When effusion stopped and explosive degassing began, harmonic tremor was observed as acoustic waves in the air and in the seismic data, and the harmonic modes became clearer and more stable. This transition in the character of the harmonic tremor coincided with rapid deflation of the lava that had accumulated in the crater. Based on these observations, and laboratory experiments reproducing the features of the wave fields, it is concluded that the harmonic tremor sequence at Shinmoe-dake was generated by gas flowing through channels in the gradually solidifying lava. Comparing our results with the few cases of similar transition observed at other volcanoes, we expect that the transition indicates changes in magma rheology and degassing conditions in the crater, and therefore of changes in eruptive activity. Key words: Harmonic tremor, infrasound, lava deflation, degassing, lava viscosity, bubbles.
Key wordsHarmonic tremor infrasound lava deflation degassing lava viscosity bubbles
Active volcanoes generate a variety of oscillations in the ground and in the air. Among them is what is called ‘harmonic tremor (HT)’. As the name indicates, HT is a sustained oscillation consisting of a fundamental mode and its overtones. The regularly repeating wavefield suggests some non-linear feedback mechanism, and volcanoes are often compared to flow-driven musical instruments (Julian, 1994; Lesage et al., 2006; Rust et al., 2008), in which either magma or gas may be ‘blowing’ the instrument. Although the unique features of HT oscillations have attracted researchers for decades, sources have not been exclusively determined for any observed cases.
Previous studies of HT have primarily focused on seismic observations. However, observations of volcanically-generated infrasound are increasing, and HT has also been found in the atmosphere (Hagerty et al., 2000; Lees et al., 2008). Within the same period of activity, HT is sometimes transmitted both in the air and in the ground (SAHT: Seismo-Acoustic HT), but other times transmitted only in the ground (SHT: Seismic HT) (Lees et al., 2008). The accompanying paper (Lyons et al., 2013) presents a laboratory experiment simulating the feature of switching between SHT and SAHT and applies the model to observations at Fuego volcano, Guatemala. Here, we add a new observation of SHT-SAHT switching that occurred in the 2011 eruption of Shinmoe-dake in Kyushu, Japan. In this case, the conditions and timing of eruptive activity when the switching occurred were well constrained for the first time.
2. Observation at Shinmoe-Dake
2.1 Stations and instruments
Locations of the stations used in this work are summarized in Fig. 1.
The Earthquake Research Institute of the University of Tokyo operates several permanent telemetered broadband seismic stations (Trillium 120P) near the volcano. An infrasonic microphone (Hakusan, SI102, 0.1 –1000 Hz) was installed on the 6th of December 2010 to SMN, the station nearest to Shinmoe-dake crater. The same type of infrasound microphones were added to other stations, KRS and EBS, on the 29th and 30th of January 2011, respectively, after the eruption started. All the stations were recorded continuously with a sample rate of 100 Hz.
A permanent light-sensitive monitoring camera is operated by Aira-Isa Regional Development Bureau of Kagoshima prefecture about 3 km west of Shinmoe-dake. The elevation of the camera location is 1366 m, which is slightly higher than the west side of the crater rim of Shinmoe-dake (1350 –1360 m). The image (640 × 480 pixels, 30 fps) was transferred to the bureau in real time and saved every 2 seconds during the eruption. The angle and zoom of the camera were not fixed but remotely controlled for monitoring purposes. The scale of an image was calculated using the positions of the topographic features (Fig. 1(b)). The distance between the peaks on the north and south sides of the crater rim is 753 m, and their elevations are 1396 m and 1419 m, respectively.
2.2 Switching from SHT to SAHT
2.3 Lava emplacement and deformation in the crater
An aerial photo taken in the evening of January 31 (Fig. 6(b)) shows that the vent glows yellow-red, while other parts of the surface are black. The color of hot lava observed by naked eyes in darkness is yellow-red at the temperature of 700 –800°C, dark red at 600 –700°C, and black below 500–550°C (Miyagi and Maejima, 2012). Although colors of digital images are not necessarily the same as those seen by the naked eye, and depend on the camera and its setting, the color distribution of the yellow-red vent (Fig. 6(b)) is similar to the data presented by Miyagi and Maejima (2012) for a basalt surface at 843° C photographed by a Nikon D90.
A very strong explosion occurred at 7:54 on February 1 and a petrologic study of dense rock ejected by the explosion showed that the lava consisted of 30 vol.% phe-nocrysts and melt with 63.9 wt% of SiO2 (57.9 wt% SiO2 in the whole rock), and that the magma temperature was 960–980°C (Suzuki et al., 2013). Using the composition of the melt (Suzuki et al., 2013) and water solubility at the atmospheric pressure (Moore et al., 2008), viscosity of the melt is estimated (Giordano et al., 2008) to be 106 Pa s at 1000°C, 109 Pa s at 800°C, and more than 1011 Pa s below 700°C. The effective viscosity of the lava is larger by a factor due to the effect of phenocrysts (Marsh, 1981; Lejeune and Richet, 1995).
Processing the camera images reveals that the surface of the lava dome rapidly subsided by 10 m in an hour (3 × 10−3 m/s) and by 20 m in less than 6 hours (10−3 m/s) following 18:10. The subsidence corresponds to the apparent strain rate in the order of 10−5 s−1, considering the bottom and the top of the dome to be 1271 m (Fig. 1(b)) and ~1370 m (Fig. 7(c)), respectively. The first SAHT (Fig. 4(a)) was observed at 21:47. After February 1, the lava surface was sometimes visible when the camera was zoomed but was not clear.
There is a possibility that the surface subsidence shown in Fig. 7(c) was a result of an illusion due to change of the volcanic glow. If a crack had opened on the camera side slope of the lava, the top might be hidden behind the glows emitted from the crack and apparently seem to be lowered. However, the subsidence had certainly occurred by the morning of February 1, and there is no evidence that the subsidence occurred in different periods during the night.
2.4 Time histories
The Japan Meteorology Agency (JMA) operates a tilt-meter (Mitsutoyo, JTS-3B, 5 mV/μrad) in a 98-m-deep borehole at KITK (Fig. 1(a)). The east-west component of tilt at KITK is shown to illustrate the volume decrease of the magma chamber during the eruptions (Fig. 8(a)). The 3 steps on the 26th and 27th of January correspond to the volume change during the 3 subplinian eruptions, which were estimated to be 1 –2 × 106, for the individual steps (Kozono et al., 2013), and the following gradual slope corresponds to the volume change of 5 × 106 m3 during the lava effusion (Kozono et al., 2013). The volume decrease (figure 5 of Kozono et al., 2013) and the west-down tilt (Fig. 8(a)) halted at 19:00 on the 31st of January.
The RMS amplitudes (1 min. time window) of seismic and acoustic signals at SMN in the 1 –7 Hz frequency range are shown in Fig. 8(b) and 8(c), respectively. Both the seismic and acoustic amplitudes have small but dense spikes from January 29 to 31 during frequent small explosions. Spikes in Fig. 8(b) and 8(c) become stronger with longer intervals after February 1, when vulcanian activity started. The open triangles indicate SHT events, and the solid triangles indicate SAHT events. Figure 8 and photos in Fig. 6 clearly show that the switching from SHT to SAHT, the lava shrinkage, and the transition from frequent small explosions to intermittent vulcanian explosions all occurred on the night of January 31 when magma flow from the chamber stopped.
Based on these observations, we consider that at least the SAHT was generated by gas flow, not by magma flow. Because at Shinmoe-dake SHT and SAHT have common harmonic modes, and they occurred subsequently in the same period of the eruption sequence, we assume an identical gas-driven oscillator for both of SHT and SAHT.
3.1 A model for SHT-SAHT switching
HT started as SHT in the last stage of lava discharge.
The harmonic modes were unclear and unstable at this stage.
During the SHT events, the acoustic wavefield consisted of impulses associated with small explosions.
SAHT started after the lava discharge stopped.
The harmonic modes in SAHT are clearer and more stable than those in SHT.
The fundamental frequency was variable both in SHT and SAHT, and there were no systematic difference in the frequency between SHT and SAHT.
In the laboratory, we observed transitions between bubbling (9(c)) and an open conduit (9(d)), which is a feature of gas transport through such visco-elastic shear-thinning fluid with appropriate combinations of viscosity and gas flux (Divoux et al., 2009, 2011). In the bubbling regime, only pressure sensors beneath the fluid detect pressure perturbations generated by the valve, while acoustic waves in the air are impulsive. The regularity of the valve oscillation is disturbed by pressure fluctuations associated with bubble growth and release from the tube (9(c)). HT generated by the valve oscillation is transmitted to the air when a conduit is established through the fluid (9(d)), which only occurred with high fluid viscosity, resulting in clear and more stable harmonics. When we use water instead of the visco-elastic fluid (9(b)), the pressure fluctuation generated by bubbles is too large to establish regular oscillation of the valve and no HT is generated.
These observations are similar to the Shinmoe-dake HT events. At the beginning of lava effusion, viscosity was too low and deformation of the lava flow at the crater floor was too large to establish HT, as in Fig. 9(b). SHT started at the late stage of the lava effusion, and during the period of SHT degassing continued through the vents near the center of the pancake dome (Fig. 6(a), (b)). The vent (the open circle in Fig. 1(b)) was almost directly above the initial source of lava effusion (the open star in Fig. 1(b)), and a low-viscosity pathway might have been formed from the source to the vents, through which gas escaped preferentially, along with magma that fed the lava dome. Because the temperature along the pathway was kept high and the viscosity low, a stable open conduit was not formed and we consider the gas transport mechanism to have been bubble-dominated. This gas transport mechanism is similar to the bubbling regime in non-newtonian shear-thinning fluid (Fig. 9(c)). After effusion stopped, the degassing rate decreased and the vent was closed (Fig. 6(c), (d)). However, slow degassing continued and pressure accumulated in the lava, and progressive degassing and crystallization of the lava may have also increased its effective viscosity. When the overpressure in the lava became too high, gas was released by a vulcanian explosion and generated an open channel. The increased viscosity of the magma allowed the channel to remain open long enough to permit focused degassing through the channel and the transmission of HT into the atmosphere, as in Fig. 9(d).
3.2 Comparison with other cases
It occurs in eruptions with mixed explosive and effusive behavior.
SHT tends to be associated more frequently with strombolian activity, while SAHT is more often associated with vulcanian activity.
All the volcanoes erupted relatively low viscosity magma of basaltic-andesite to low-silica andesite during the events.
An absolute classification between strombolian and vulca-nian activity has not been established, and varying terminology has been adopted to describe activity of different volcanoes or even of the same volcano (Marchetti et al., 2009). Here in feature (2) above, ‘strombolian’ and ‘vulca-nian’ activity indicate a period with small frequent explosions and that with stronger explosions at longer intervals, respectively.
Below we overview the few reported cases of SHT-SAHT switching from the point of view of the above commonality. Although it is somewhat speculative, the aim of this section is to draw attention to the switching phenomena and increase reported cases. Then, the present hypothesis can be tested.
3.2.1 Arenal volcano, April –May, 1997
Arenal has been regularly active since the strong 1968 eruption that opened 3 new craters in 200-m intervals on the west flank of the volcano. The activity in the period of 1984 –1998 was characterized by mixed strombolian and effusive eruptions from the highest crater, which is about 200 m from the summit (Alvarado and Soto, 2002; Wadge et al., 2006). Although the discharged magma is basaltic andesite and sometimes forms a lava pool in the crater, its viscosity is sometimes large enough to allow the generation of pyroclastic flows (Alvarado and Soto, 2002).
Hagerty et al. (2000) were the first to report SHT-SAHT switching. They had seismometers and microphones 2 km from the summit crater of Arenal volcano during April –May, 1997. Several episodes of HT were recorded during the experiment, some of which were recognized to be SAHT. Only one SAHT example (~25 minutes on 7:30 on April 27, 1997) is reported in the paper, and has the following characteristics (Hagerty et al., 2000). Two small explosions occurred, one at about 8 min before the onset of SAHT and the other 45 min after the first explosion. The infrasound amplitudes at 2 km were a few Pa for the explosions and less than 0.5 Pa for the tremor. For two hours starting at ~4:00, or about 3 hours before the SAHT event, small regularly repeating explosions were recorded by both the seismometers and the microphones for nearly two hours. Their amplitudes were less than 1 Pa, and the time between two successive explosions varies between 0.7 –15 s throughout this period. In the sense that this SAHT event occurred with the transition from frequent explosions (intervals 0.7 –15 s) to less frequent explosions (an interval of 45 min), it is similar to the Shinmoe-dake case. However, lack of information on the features of infrasound signal during SHT or temporal distribution of SHT and SAHT prevents further comparison here.
3.2.2 Reventador volcano, August –September, 2005
Reventador is one of the most active volcanoes in Ecuador. The paroxysmal eruption of November 3, 2002 was very similar to the Shinmoe-dake 2011 eruption. It started with subplinian activity (VEI 4 with andesitic pyroclastic falls and flows) with limited seismic precursory signals. The activity shifted between the 6th and 21st of November to an effusive stage characterized by the emission of two lava flows (andesite to low-silica andesite) (Ridolfi et al., 2008). Following the spring 2005 lava flows, strombolian activity was first observed from the crater in June 2005.
Lees et al. (2008) performed seismic and acoustic observations for Reventador volcano at 2 –8 km from the vent during August 2 to September 13, 2005. Activity varied considerably over the duration of the six-week seismic deployment. Fluctuations in volcanic activity were reflected in the seismicity, which was relatively mild at the onset of their installation period. Starting on the next day, however, explosion levels picked up and several episodes of high ex-plosivity were observed (Lees et al., 2008). Over the course of the observations, several sequences of SHT or quasi-periodic tremor were recorded. The harmonic modes were not stable and occasionally lasted for only 30 s. However, significant temporal variations in frequency content were not observed even when amplitudes fluctuated considerably. There was also not a strong correlation between seismic and acoustic signals in these episodes. SAHT events were observed during periods of chugging, which is observed at various volcanoes such as Arenal, Costa Rica (Benoit and McNutt, 1997), Karymsky, Russia (Lees et al., 2004), San-gay (Johnson and Lees, 2000) and Tungurahua (Ruiz et al., 2006), Ecuador, but SHT-SAHT switching have not been reported at Karymsky, Sangay and Tungurahua. The chugging sequence is characterized by regularly repeating pulses in both seismic and acoustic signals starting after an explosion with a certain time delay. No clear transition separating periods of SHT and SAHT was reported in the case of Reventador, and all of explosions, SHT, and SAHT could be observed on the same day (Lees et al., 2008).
3.2.3 Fuego volcano, 2008 –2009
Fuego volcano is one of the most active volcanoes in Central America and erupts basaltic andesite, sometimes producing subplinian columns and pyroclastic flows that affect the local population (Rose et al., 1978; Lyons et al., 2010). Fuego has been continuously active since the onset of the preset period of activity which began with a VEI 2 eruption on May 21, 1999. Despite limited observations, HT has been frequently observed during temporary campaigns (Lyons et al., 2010, 2013). Lyons et al. (2013) observed SHT during a period of strombolian activity in June 2008 that include the effusion of short lava flows, while the following year SAHT was recorded during a period of mild vulcanian activity without associated lava effusion. In this case, the SHT-SAHT switching was associated with subtle changes in magma production rate, which led to a switch from strombolian to vulcanian activity. Lyons et al. (2013) propose that a lower magma production rate in 2009 led to degassing crystallization and an increase in the magma viscosity, thereby allowing the production of open degassing channels in the uppermost portion of the magma column and transmission of HT into the atmosphere.
Switching from SHT to SAHT was observed during the Shinmoe-dake 2011 eruption. Although similar phenomena have been observed at several volcanoes, the sequence was for the first time correlated with the growth and shrinkage of the lava in the crater and compared with the entire sequence of the eruption activity. SHT-SAHT switching has been generated in the laboratory with gas flow through a feedback oscillator and a visco-elastic fluid (Lyons et al., 2013). The model results show that SAHT was only generated in visco-elastic fluid with sufficiently high viscosity. We propose that the SHT-SAHT switching sequence at Shinmoe-dake may be associated with gradually solidifying lava in the crater. Initially the viscosity was too low and the movement of the lava was too dynamic to establish HT. At the last stage of the lava effusion, HT occurred only in the conduit with impulsive waves in the air, reflecting bubble-regime degassing. After the effusion stopped, the viscosity of the lava accumulated in the crater became high enough to sustain open degassing channels. The cessation of lava discharge and subsequent degassing of the lava dome may have caused the flattening of the lava. SHT-SAHT switching has been observed at other volcanoes erupting magma with similar compositions, and in mixed activity consisting of effusive, strombolian, and vulcanian styles. SHT-SAHT switching may be a good indicator of changes in the conditions of magma rheology and discharge rates of magma and gas.
We are grateful to the Yomiuri Shimbun, Asia Air Survey Co. Ltd., and Japan Meteorology Agency for providing the aerial photos of the Shinmoe-dake crater, and Aira-Isa Regional Development Bureau of Kagoshima prefecture for providing video images. Information to estimate the properties of the lava was provided by Y. Suzuki. We also thank M. Takeo, J. M. Lees, and V. Vidal for useful discussions. The manuscript has been significantly improved by constructive comments by the two reviewers (J. B. Johson and T. Nishimura) and the editor (S. R. McNutt). This study has been supported by the MEX Grant-in-Aid for Scientific Research (22540431, 21200052), and JSPS Fellowship Summer Program (SP11-039).
- Goto, A. and J. B. Johnson, Monotonic Infrasound and Helmholtz Resonance at Volcan Villarrica (Chile), Geophys. Res. Lett., 38, L06301, doi:10.1029/2011gl046858, 2011.Google Scholar
- Imura, R. and T. Kobayashi, Eruptions of Shinmoedake Volcano, Kirishima Volcano Group, in the last 300 years, Bull. Volcanol. Soc. Jpn., 36, 135–148, 1991 (in Japanese with English abstract).Google Scholar
- Johnson, J. B., R. Sanderson, J. Lyons, R. Escobar-Wolf, G. Waite, and J. M. Lees, Dissection of a composite volcanic earthquake at Santiaguito, Guatemala, Geophys. Res. Lett., 36, doi:10.1029/2009gl039370, 2009.Google Scholar
- Kozono, T., H. Ueda, T. Ozawa, T. Koyaguchi, E. Fujita, A. Tomiya, and Y. J. Suzuki, Magma discharge variations during the 2011 eruptions of Shinmoe-dake volcano, Japan, revealed by geodetic and satellite observations, Bull. Volcanol., 75, doi:10.1007/s00445-013-0695-4, 2013.Google Scholar
- Lyons, J. J., M. Ichihara, A. Kurokawa, and J. M. Lees, Switching between seismic and seismo-acoustic harmonic tremor simulated in the laboratory: Insights into the role of open degassing channels and magma viscosity, J. Geophys. Res., 2013.Google Scholar
- Miyagi, I. and Y Maejima, Estimation of eruptive temperatures based on the color data of digital cameras, Bull. Volcanol. Soc. Jpn., 57, 45–50, 2012 (in Japanese with English abstract).Google Scholar
- Nakao, S., Y. Morita, H. Yakiwara, J. Oikawa, H. Ueda, H. Taka-hashi, Y. Ohta, T. Matsushima, and M. Iguchi, Volume change of the magma reservoir relating to the 2011 Kirishima Shinmoe-dake eruption—Charging, discharging and recharging process inferred from GPS measurements, Earth Planets Space,65, this issue, 505 –515, doi:10.5047/eps.2013.05.017, 2013.CrossRefGoogle Scholar
- Ridolfi, F., M. Puerini, A. Renzulli, M. Menna, and T. Toulkeridis, The magmatic feeding system of El Reventador volcano (Sub-Andean zone, Ecuador) constrained by texture, mineralogy and thermobarometry of the 2002 erupted products, J. Volcanol. Geotherm. Res., 176, 94 –106, 2008.CrossRefGoogle Scholar
- Ripepe, M., E. Marchetti, C. Bonadonna, A. J. L. Harris, L. Pioli, and G. Ulivieri, Monochromatic infrasonic tremor driven by persistent degassing and convection at Villarrica Volcano, Chile, Geophys. Res. Lett., 37, L15303, doi:10.1029/2010gl043516, 2010.Google Scholar
- Rust, A. C., N. J. Balmforth, and S. Mandre, The feasibility of generating low-frequency volcano seismicity by flow through a deformable channel, in Fluid Motions in Volcanic Conduits, edited by S. J. Lane and J. S. Gilbert, pp. 45 –56, Geol. Soc., London, 2008.Google Scholar
- Sakai, T., H. Yamasato, and K. Uhira, Infrasound accompanying C-type tremor at Sakurajima Volcano, Bull. Volcanol. Soc. Jpn., 41, 181 –185, 1996 (in Japanese with English abstract).Google Scholar
- Suzuki, Y., A. Yasuda, N. Hokanishi, T. Kaneko, S. Nakada, and T. Fu-jii, Syneruptive deep magma transfer and shallow magma remobiliza-tion during the 2011 eruption of Shinmoe-dake, Japan—Constraints from melt inclusions and phase equilibria experiments—, J. Volcanol. Geotherm. Res., 257, 184 –204, 2013.CrossRefGoogle Scholar
- Tsutsui, M., K. Tomita, and T. Kobayashi, Fumarolic activity since December 2003 and volcanic activity during the Meiji and Taisho eras (1880 –1923) of Ohachi Volcano, Kishirima Volcano Group, Southern Kyushu, Japan, Bull. Volcanol. Soc. Jpn., 50, 475 –489, 2005 (in Japanese with English abstract).Google Scholar
- Yokoo, A., T. Tameguri, M. Iguchi, and K. Ishihara, Sequence and characteristics of the 2007 eruption at Showa Crater of Sakurajima Volcano, Ann. Disas. Prev. Res. Inst., Kyoto Univ., 51B, 267 –273, 2008 (in Japanese with English Abstract).Google Scholar