Relation between single very long period pulses and volcanic gas emissions at Mt. Asama, Japan

Transcription

1 GEOPHYSICAL RESEARCH LETTERS, VOL. 38,, doi: /2011gl047555, 2011 Relation between single very long period pulses and volcanic gas emissions at Mt. Asama, Japan Ryunosuke Kazahaya, 1 Toshiya Mori, 1 Minoru Takeo, 2 Takao Ohminato, 2 Taku Urabe, 2 and Yuta Maeda 3 Received 24 March 2011; revised 28 April 2011; accepted 28 April 2011; published 15 June [1] Multiple volcanic observations conducted at Mt. Asama, Japan, provide evidence of a link between single very longperiod (VLP) seismic pulses and volcanic gas emissions. SO 2 flux measurements were conducted on 2 June 2009, when Mt. Asama was producing ash free eruptions with VLP pulses. Gas bursts from a vent at the crater bottom following the VLP pulses provided an excellent opportunity to examine the relation directly. The SO 2 emission for each eruption was calculated by integrating high temporal SO 2 flux data obtained by the SO 2 imaging system and subtracting the contribution from quiescent degassing from fumaroles around the crater bottom. A seismic moment of VLP pulse was estimated by the waveform inversion. We observed seven eruptions and obtained the proportional relation between VLP pulse moment and SO 2 emission. The relation determined is consistent with the VLP source model; these observational results are the first report of a quantitative comparison between single VLP pulse moment and volcanic gas emission. Citation: Kazahaya, R., T. Mori, M. Takeo, T. Ohminato, T. Urabe, and Y. Maeda (2011), Relation between single verylong period pulses and volcanic gas emissions at Mt. Asama, Japan, Geophys. Res. Lett., 38,, doi: /2011gl Introduction [2] Very long period (VLP) seismic signals are thought to be generated by the mass transport and volume change of fluids (magmas and volatiles) at depth within a volcano [e.g., Chouet et al., 2010]. These seismic signals, with periods longer than 2 s, are commonly observed at many volcanoes around the world exhibiting different levels of volcanic activity. Based on observations, cracks undergoing a sequence of inflation and deflation [e.g., Chouet et al., 2005] and an expansion of slugs within the magma column [e.g., Ripepe et al., 2005] have been proposed as the processes generating VLP signals. The VLP source models have been also tested in laboratory experiments [James et al., 2004] and in numerical simulations [O Brien and Bean, 2008], with the results suggesting that slug ascent through a volcano conduit likely produces VLP pulses. The relation between volcanic gas emissions and VLP seismic signals has been documented only by video camera observations of the gas exhalation accompanying seismic pulses, which did not provide enough 1 Geochemical Research Center, University of Tokyo, Tokyo, Japan. 2 Earthquake Research Institute, University of Tokyo, Tokyo, Japan. 3 National Research Institute for Earth Science and Disaster Prevention, Tsukuba, Japan. Copyright 2011 by the American Geophysical Union /11/2011GL quantitative data in order to compare them efficiently [e.g., Chouet et al., 2010; Dawson et al., 2010]. There are few quantitative gas flux studies on these volcanic seismic signals, and these studies only compared long term (daily to monthly) flux changes and seismic data [Fischer et al., 1994; Ripepe et al., 2005]. Conventional SO 2 flux observational techniques using a correlation spectrometer or a UV spectrometer require scanning of the plume, which takes generally more than several tens of seconds. Because of this limitation, the comparison of SO 2 emissions with volcanic geophysical streams has been difficult. Recently, SO 2 visualization techniques have been developed [Bluth et al., 2007; Mori and Burton, 2006], providing the possibility of measuring SO 2 flux at unprecedented frequencies (2 Hz) and allowing the comparison of SO 2 flux fluctuations directly with seismic signals. This innovation will yield new insights into degassing and eruptive processes. For example, Nadeau et al. [2011] have reported a linkage between volcanic tremor and degassing. [3] In this study, we examined the relationship between VLP pulses and volcanic gas emissions at Mt. Asama, Japan, through a combination of high temporal resolution SO 2 flux measurement using a SO 2 imaging system, vent process monitoring using a video camera set at the crater rim, and seismic observation using broadband seismometers installed at the summit. Gas burst eruptions from a vent at the crater bottom following VLP pulses gave us an optimal opportunity to investigate the relationship. 2. Mt. Asama [4] Mt. Asama, located 150 km northwest of Tokyo (Figure 1a), is one of the most active volcanoes in Japan. Its main rock types are andesite and dacite. The summit of Mt. Asama consists of a crater about 450 m in diameter and about 250 m deep. VLP seismic signals have been observed at Mt. Asama since 2003 when a broadband seismometer was installed at the crater rim [Yamamoto et al., 2005]. Yamamoto et al. [2005] concluded that similar seismic pulses had previously occurred, at least since September This VLP seismic activity decreased after July 2004, coinciding with a dike intrusion [Aoki et al., 2005]. During 2008, the number of VLP pulses increased before eruptions. The most recent eruption occurred on 2 February 2009, causing the opening of a small vent at the center of the crater bottom. Since then, ash free degassing bursts from the vent have been frequently observed after the occurrence of a VLP pulse. Inner crater video images and the seismic data show that each gas burst from the vent starts s (av. 29 s) after the onset of a VLP pulse. The durations of VLP pulses are s, and most gas bursts continue over several 1of5

2 SO 2 flux time series obtained by this method showed a good agreement with that of conventional traverse methods in another field campaign conducted at Sakurajima [Kazahaya et al., 2010]. [7] The vent located at the crater bottom was monitored by a video camera installed on the west side of the crater rim (Figure 1b). This camera acquired one image per second. For the seismic observation, VLP signals were recorded by a network of broadband seismometers (Guralp CMG 3T 360 s and Nanometrics Trillium 120 s sensors; Figure 1a). Since VLP signals are so feeble they can hardly be recognized, the seismic recording station needs to be installed as close as possible to the source. A typical VLP pulse accompanying a gas burst is shown in Figure 2b. Figure 1. (a) Map of Mt. Asama. The square and circles show Asama Volcano Observatory (AVO), located about 4 km east from the crater, and the seismic stations around the summit, respectively. The map was illustrated by using a digital elevation model data with 10 m resolution provided by Geospatial Information Authority of Japan. (b) Photographs of the crater bottom taken by the inner crater camera set at the western crater rim. (left) Before an eruption; quiescent stage and only passive degassing from the fumaroles. (right) During the eruption; ash free gas burst eruption accompanying a VLP pulse from the vent at the crater bottom. hundred seconds. According to observations taken on 2 June 2009, VLP pulses occurred 2 9 times per hour. 3. Observation Methods [5] A SO 2 imaging system was used to conduct SO 2 flux measurements from 10:00 to 14:00 on 2 June The system instruments were set up at the Asama Volcano Observatory, located about 4 km east of the crater (Figure 1a). Weather conditions during the measurement period were optimal, with clear skies and little wind. [6] The imaging system consists of two UV cameras [Mori and Burton, 2006] and a compact ultraviolet spectrometer system (COMPUSS) [Mori et al., 2007]. The two UV cameras (Alta U260, Apogee Inc.) have UV band pass filters, which were centered at 313 nm (affected by SO 2 absorption) and 330 nm (where SO 2 does not absorb). The UV cameras were synchronized and obtained one UV image of a volcanic plume every 2 s. The UV images provide SO 2 column distribution in the plume, plume speed, and SO 2 flux. The methodology to process UV images and to calculate the SO 2 flux were taken from Mori and Burton [2006], Bluth et al. [2007] and Kantzas et al. [2010]. The COMPUSS, based on a UV spectrometer (USB2000, Ocean Optics Inc.), was designed for SO 2 measurements by UV spectral analysis. We observed a part of the volcanic plume with the COMPUSS in order to quantify a light dilution effect that was reported by Mori et al. [2006] and examined by Kern et al. [2010]. This effect causes a significant underestimate of SO 2 amount. The UV spectra of the volcanic plume were used for the calibration of the UV cameras. The 4. Analyses [8] SO 2 flux measured by the SO 2 imaging system ranged between kg/s (av kg/s), which was consistent with SO 2 flux of kg/s (av. 9.2 kg/s) measured by the traverse method using the COMPUSS on the same day (JMA; 306_So2emission.htm). [9] Figure 2 shows the SO 2 flux time series and vertical velocity seismogram during the observation period. A clear pattern of SO 2 flux follows each VLP pulse. We detected seven degassing bursts accompanying clear VLP pulses; these bursts originated from the vent and are clearly recognizable by the footage obtained from the inner crater camera. Events with a feeble gas emission, ones whose burst could not be recognized due to poor visibility at the crater bottom and ones with corresponding VLP pulses affected by distant earthquakes were excluded from our analyses. The SO 2 emissions associated with the eruptions were estimated by integrating the SO 2 flux data. The integration ranges were determined by the starting and ending time of each eruption. These times were obtained from the inner crater video footage (Figure 1b) and the SO 2 column footage. When integrating the SO 2 flux, we had to consider the time lag for the gas mass to reach the summit from the crater bottom. When a gas burst occurred at 10:30 (event 2 in Figure 2a), we were able to determine the starting time of the burst from the vent by the inner crater video camera and the time the gas mass reached the top of the summit by the UV footage of the volcanic plume (see auxiliary material). 1 The time lag between the starting time of the burst and the time the gas mass reached the summit was 36 s. The footage of the SO 2 column and the inner crater shows that gas emitted from the vent accumulated inside the crater, and then ascended as a gas mass like a puff (ponding of gas in the crater before buoyant rise; see auxiliary material). In order to consider this effect, we also calculated the expected time lag (25 60 s) by using the depth of the crater (about 250 m) and plume speed calculated by inner crater video images (4 10 m/s). A time lag value of 36 s was considered acceptable, and we applied this value for all of the eruptions. The errors associated with the estimation of SO 2 emission 1 Auxiliary materials are available in the HTML. doi: / 2011GL of5

3 Figure 2. SO 2 flux time series and low pass filtered seismic record for 2 June The crosses show the SO 2 flux time series for the area immediately above the summit. The gray line shows the seismic signals. (a) Overview of entire SO 2 flux time series and seismic signals. The analyzed events are indicated by arrows at upper side. The lack of SO 2 flux data corresponds to the calibration of the SO 2 imaging system. The striking seismic signal starting at around 11:30 is due to a distant earthquake. (b) Close up view of the time series from 13:40 to 13:50 for event 7. SO 2 flux increased after the occurrence of the VLP pulse. The shaded area corresponds to the amount of SO 2 emission during this event. Note that the seismometer s response is not corrected. were also defined by uncertainty of the integration time due to the expected time lag of 25 to 60 s. [10] Since SO 2 flux fluctuated over time due to fumarolic activity around the crater bottom, which is not related to VLP pulses (Figure 1b), we had to eliminate these effects and estimate the SO 2 emissions corresponding only to VLP pulses. Taking into account the SO 2 flux excluding the occurrence of eruptions (e.g., the period from 11:10 to 11:25 in Figure 2a), i.e., the fluctuation cycle (about 300 s), the fluctuation range (relative standard deviation is 49%) and the average level of the flux during the observational period (11.2 kg/s), we assumed that the SO 2 flux caused by the fumaroles could be approximated by the baselines between the SO 2 flux values at the beginning and the ending times of the integration. We obtained SO 2 emission data corresponding only to VLP pulses by subtracting the SO 2 emission from fumaroles from the total SO 2 emission (Figure 2b). [11] In some cases during the observation period, new VLP signals occurred before the previous gas burst stopped (events 2,3 and 4,5 in Figure 2). For the previous burst (events 2 and 4), we defined the ending time of the gas burst as the starting time of the following burst (events 3 and 5). Because of this overlap, the gas emissions of the previous eruptions and the following eruptions are systematically under and overestimated, respectively. The SO 2 emissions of seven events were tons (Figure 3). Events 2 and 3 are shown in the auxiliary material, which includes footage showing the SO 2 column amount and inner crater area, and charts showing the SO 2 amount and a vertical seismogram. [12] Seismic moment for each event was estimated from a waveform inversion. Since the station network on the event date was restricted to only four permanent stations, source location and mechanism were assumed to be the same as representative events in 2008 [Maeda and Takeo, 2011], when 10 additional temporary stations were available. Based on this assumption, source time function of a single component was Figure 3. Plot between SO 2 emission and VLP pulse moment. (a) For seven eruptions: the triangles and circles with arrows correspond to the events that systematically under or overestimated SO 2 emissions, respectively. The symbols * and show the pairing of consecutive events as event 2 3 andevent4 5, respectively. (b) The relationship between volcanic gas emission and moment of VLP pulse where values of consecutive events, which include systematical errors, are summed. The line shows linear regression of the relationship, and R 2 represents the determination coefficient of the regression. For further explanation, see text. 3of5

4 estimated from the waveform inversion, from which the seismic moment was determined. 5. Results and Discussion 5.1. Relation between Volcanic Gas Emission and VLP Pulse Moment [13] The correlation between the SO 2 emission of each gas burst and the moment of VLP pulse is summarized in Figure 3a. Excluding the systematical errors for the four overlapping events, the gas emissions are positively correlated with the VLP pulse moments. The VLP signals of Mt. Asama have been explained as resulting from a synchronized expansion of a tensile crack and a cylinder [Maeda and Takeo, 2011]. The moment release of the VLP signals should be proportional to the stress free volumetric change, which again is proportional to the real change of source volume times pressure [e.g., Kawakatsu and Yamamoto, 2007]. Assuming this relation, we can sum the VLP pulse moments and gas emissions of the pairs of under and overestimated events, and thus cancel out the systematic errors induced by this procedure. The processed data are plotted in Figure 3b. The proportional relationship between a single very long period pulse and the volcanic gas emission is found to be G ¼ 7: M; where M and G describe VLP moment release (Nm) and SO 2 emission (ton), respectively. The coefficient of determination of this relationship is Note that this equation is specific to Mt. Asama, and we must conduct similar observations at other volcanoes in order to obtain the coefficient for each volcano. This is the first report of the relationship between volcanic gas amount and corresponding VLP pulse. By assuming equation (1), we can estimate the gas emission from the amplitude of a VLP pulse without a gas flux observation. On the observation day, VLP pulses occurred a total of 149 times, and the daily total amount of SO 2 emission associated with the VLP pulses was calculated to be about 320 tons. Since volcanic gas flux measurements require UV light from the sun, data collection is fundamentally limited by observational conditions. Measurements cannot be taken at nighttime, for instance, because of the absence of UV light. The volcanic gas estimation procedure using seismic signals that we have developed provides the possibility of the measurement of eruptive volcanic gas emissions all day long Gas Emissions Related to Eruptions and Quiescent Degassing [14] At the time of this study s observations, Mt. Asama was emitting volcanic gas as a persistent degassing from fumaroles in the crater and as gas bursts from the vent. Since the SO 2 imaging system allows us to measure the gas mass for a single eruption, we can estimate the relative contribution of individual eruptive gas emissions compared with the total emission. In the case of ash free eruptions at Mt. Asama, eruptive and quiescent gas emissions are thought to reflect the amount of magma associated with each eruption and a steady magma degassing rate, respectively. The total SO 2 emission during the observation period was about 134 tons, and the SO 2 emissions from the eruptions in this period amounted to about 22 tons. Thus the SO 2 emissions ð1þ corresponding in time to the ash free eruptions contributed about 16% of the total emission. [15] We compared our results with similar studies carried out at other volcanoes and found considerable differences. Mori and Burton [2009] reported that SO 2 emissions corresponding to discrete eruptions at Stromboli, Italy, amounted to kg, contributing 3 8% of the total daily gas emission. Fischer et al. [2002] measured the SO 2 emissions of small ashy eruptions and found they totaled kg, contributing over 90% of the total SO 2 emissions at Karymsky volcano, Kamchatka. [16] These differences reflect the degassing characteristics of each volcano. The eruptive gas emissions at Mt. Asama are larger than those at Stromboli [Mori and Burton, 2009] by one to three orders of magnitude. That is, the magma amount related to each eruption at Mt. Asama is larger than that at Stromboli. The eruptive SO 2 emissions at Mt. Asama considered in this study are comparable to those at Karimsky [Fischer et al., 2002]. However, the ratios between eruptive and quiescent degassing amounts are completely different. Almost all gas was degassed by eruptions without quiescent degassing at Karmysky volcano, that is, this volcano is characterized by feeble passive degassing in that eruptive period. In the case of Mt. Asama, the gas contribution from fumaroles was about 4 times as large as that from eruptions. This fact suggests that large gas pathways exist even when a volcano is in an eruptive period. 6. Concluding Remarks [17] We conducted multiple volcanic observations at Mt. Asama, Japan, and found a linear relationship between VLP seismic moments and volcanic gas emissions. The observational results strongly support the VLP source model and provide a linkage between VLP seismic signals and volcanic gas. This relationship also suggests the possibility of gas mass quantification by seismic analyses of VLP pulses. Based on comparisons between observed volcanic gas emission (total degassing) and calculated gas emission (eruptive degassing), we were able to calculate the ratio between eruptive and quiescent degassing amounts. This new insight implies that a huge persistent degassing mechanism exists even when Mt. Asama is in the eruptive period. [18] Acknowledgments. The authors wish to thank John Tamura of the University of Tokyo s GCOE program for the English proofreading and editing of this paper. We are also very grateful to Phillip Dawson and the anonymous reviewer for their constructive reviews. This study was supported by Grant in Aid for JSPS Fellows and the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan, under its Observation and Research Program for Prediction of Earthquakes and Volcanic Eruptions. [19] The Editor thanks Tobias Fischer and Phillip B. Dawson for their assistance in evaluating this paper. References Aoki, Y., H. Watanabe, E. Koyama, J. Oikawa, and Y. Morita (2005), Ground deformation associated with the unrest of Asama volcano, Japan, Bull. Volcanol. Soc. Jpn., 50, Bluth, G. J. S., J. M. Shannon, I. M. Watson, A. J. Prata, and V. J. Realmuto (2007), Development of an ultra violet digital camera for volcanic SO 2 imaging, J. Volcanol. Geotherm. Res., 161, 47 56, doi: /j. jvolgeores Chouet, B., P. Dawson, and A. Arciniega Ceballos (2005), Source mechanism of vulcanian degassing at Popocatepetl volcano, Mexico, determined 4of5

5 from waveform inversions of very long period signals, J. Geophys. Res., 110, B07301, doi: /2004jb Chouet, B., P. Dawson, M. R. James, and S. J. Lane (2010), Seismic source mechanism of degassing bursts at Kilauea volcano, Hawaii: Results from waveform inversion in the s band, J. Geophys. Res., 115, B09311, doi: /2009jb Dawson, P., M. C. Benitez, B. Chouet, D. Wilson, and P. G. Okubo (2010), Monitoring very long period seismicity at Kilauea volcano, Hawaii, Geophys. Res. Lett., 37, L18306, doi: /2010gl Fischer, T. P., M. M. Morrissey, V. M. L. Calvache, M. D. Gomez, C. R. Torres, J. Stix, and S. N. Williams (1994), Correlation between SO 2 flux and long period seismicity at Galeras volcano, Nature, 368, , doi: /368135a0. Fischer, T. P., K. Roggensack, and P. R. Kyle (2002), Open and almost shut case for explosive eruptions: Vent processes determined by SO 2 emission rates at Karymsky volcano, Kamchatka, Geology, 30(12), , doi: / (2002)030<1059:oaascf>2.0. CO;2. James, M. R., S. J. Lane, B. Chouet, and J. S. Gilbert (2004), Pressure changes associated with the ascent and bursting of gas slugs in liquidfilled vertical and inclined conduits, J. Volcanol. Geotherm. Res., 129, 61 82, doi: /s (03) Kantzas, E., A. McGonigle, G. Tamburello, A. Aiuppa, and R. Bryant (2010), Protocols for UV camera volcanic SO 2 measurements, J. Volcanol. Geotherm. Res., 194, 55 60, doi: /j.jvolgeores Kawakatsu, H., and M. Yamamoto (2007), Volcano seismology, in Treatise on Geophysics, vol. 4, Earthquake Seismology, edited by H. Kanamori and G. Schubert, pp , Elsevier, Amsterdam. Kazahaya, R., T. Mori, and T. Mori (2010), Degassing activity fluctuation before eruptions at Sakurajima volcano, Japan, paper presented at 6th Cities on volcanoes, Cons. de Obras Publicas y Transp., SEMR and ISRM, Tenerife, Spain. Kern, C., T. Deutschmann, L. Vogel, M. Wöhrbach, T. Wagner, and U. Platt (2010), Radiative transfer corrections for accurate spectroscopic measurements of volcanic gas emissions, Bull. Volcanol., 72, , doi: /s Maeda, Y., and M. Takeo (2011), Very Long Period pulses at Asama volcano, central Japan, inferred from dense seismic observations, Geophys. J. Int., 185, , doi: /j x x. Mori, T., and M. Burton (2006), The SO 2 camera: A simple, fast and cheap method for ground based imaging of SO 2 in volcanic plumes, Geophys. Res. Lett., 33, L24804, doi: /2006gl Mori, T., and M. Burton (2009), Quantification of the gas mass emitted during single explosions on Stromboli with the SO 2 imaging camera, J. Volcanol. Geotherm. Res., 188, , doi: /j. jvolgeores Mori, T., T. Mori, K. Kazahaya, M. Ohwada, J. Hirabayashi, and S. Yoshikawa (2006), Effect of UV scattering on SO 2 emission rate measurements, Geophys. Res. Lett., 33, L17315, doi: /2006gl Mori, T., J. Hirabayashi, K. Kazahaya, T. Mori, M. Ohwada, M. Miyashita, H. Iino, and Y. Nakahori (2007), A Compact Ultraviolet Spectrometer System (COMPUSS) for monitoring volcanic SO 2 emission: Validation and preliminary observation, Bull. Volcanol. Soc. Jpn., 52, Nadeau, P. A., J. L. Palma, and G. P. Waite (2011), Linking volcanic tremor, degassing, and eruption dynamics via SO 2 imaging, Geophys. Res. Lett., 38, L01304, doi: /2010gl O Brien, G. S., and C. J. Bean (2008), Seismicity on volcanoes generated by gas slug ascent, Geophys. Res. Lett., 35, L16308, doi: / 2008GL Ripepe, M., E. Marchetti, G. Ulivieri, A. Harris, J. Dehn, M. Burton, T. Caltabiano, and G. Salerno (2005), Effusive to explosive transition during the 2003 eruption of Stromboli volcano, Geology, 33, , doi: /g Yamamoto, M., et al. (2005), A unique earthquake activity preceding the eruption at Asama volcano in 2004, Bull. Volcanol. Soc. Jpn., 50, R. Kazahaya and T. Mori, Geochemical Research Center, University of Tokyo, 7 3 1, Hongo, Bunkyo ku, Tokyo , Japan. (k ryu@ egchem.s.u tokyo.ac.jp) Y. Maeda, National Research Institute for Earth Science and Disaster Prevention, 3 1, Tennodai, Tsukuba, Ibaraki , Japan. T. Ohminato, M. Takeo, and T. Urabe, Earthquake Research Institute, University of Tokyo, 1 1 1, Yayoi, Bunkyo ku, Tokyo , Japan. 5of5