- Email: [email protected]
Response of a hydrothermal system to magmatic heat inferred from temporal variations in the complex frequencies of long-period events at Kusatsu-Shirane Volcano, Japan
Journal of Volcanology and Geothermal Research 147 (2005) 233 – 244 www.elsevier.com/locate/jvolgeores
Response of a hydrothermal system to magmatic heat inferred from temporal variations in the complex frequencies of long-period events at Kusatsu-Shirane Volcano, Japan Masaru Nakanoa,*, Hiroyuki Kumagaib a
Graduate School of Environmental Studies, Nagoya University, Nagoya, 464-8601, Japan (Now at US Geological Survey, Menlo Park, CA, USA) b National Research Institute for Earth Science and Disaster Prevention, Tsukuba, Japan Received 4 June 2004; received in revised form 30 March 2005; accepted 4 April 2005
Abstract We investigate temporal variations in the complex frequencies (frequency and quality factor Q) of long-period (LP) events that occurred at Kusatsu-Shirane Volcano, central Japan. We analyze LP waveforms observed at this volcano in the period between 1988 and 1995, which covers a seismically active period between 1989 and 1993. Systematic temporal variations in the complex frequencies are observed in October–November 1989, July–October 1991, and September 1992–January 1993. We use acoustic properties of a crack filled with hydrothermal fluids to interpret the observed temporal variations in the complex frequencies. The temporal variations in October–November 1989 can be divided into two periods, which are explained by a gradual decrease and increase of a gas-volume fraction in a water–steam mixture in a crack, respectively. The temporal variations in July–October 1991 can be also divided into two periods. These variations in the first and second periods are similar to those observed in November 1989 and in September–November 1992, respectively, and are interpreted as drying of a water– steam mixture and misty gas in a crack, respectively. The repeated nature of the temporal variations observed in similar seasons between July and November suggests the existence of seasonality in the occurrence of LP events. This may be caused by a seasonally variable meteoritic water supply to a hydrothermal system, which may have been heated by the flux of volcanic gases from magma beneath this volcano. D 2005 Elsevier B.V. All rights reserved. Keywords: long-period (LP) events; complex frequency; temporal variation; fluid-filled crack; hydrothermal system
1. Introduction
* Corresponding author. Fax: +1 650 329 5203. E-mail address: [email protected] (M. Nakano). 0377-0273/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.jvolgeores.2005.04.003
Various types of seismic signals such as volcanotectonic (VT) earthquakes, long-period (LP) and verylong-period (VLP) events, and tremor are observed at
234
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
active volcanoes. These signals may be caused by internal volcanic activities associated with fluid movement, heat and gas supply from magma, and interactions between magma and underground water. A
swarm of VT earthquakes is often the first sign of renewed volcanic activity. VT earthquakes are indistinguishable from ordinary double-couple tectonic earthquakes, and represent the brittle response of the
Fig. 1. (a) Locations of seismic stations operated by the Earthquake Research Institute of University of Tokyo at Kusatsu-Shirane volcano. Solid and open triangles indicate stations with three-component seismometers and vertical-component seismometers, respectively. A cross between Yugama and Mizugama marks the epicenter region of LP events determined by Kumagai et al. (2002b) and Nakano et al. (2003a). The inset shows the location of Kusatsu-Shirane volcano in central Japan. (b) The monthly number of VT events that occurred at Kusatsu-Shirane Volcano in 1988–1995 (based on data of Japan Meteorological Agency, 1995).
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
volcanic rock associated with fluid movement beneath a volcano (e.g., Chouet, 1996a). LP events with typical oscillation periods between 0.2 and 2 s are considered as acoustic vibrations of a fluid-filled resonator in magmatic and hydrothermal systems, which may be triggered by pressure fluctuations associated with mass transport and/or thermodynamic processes of fluids in the resonator (e.g., Chouet, 1986, 1996a,b). The similarity of LP waveform signatures in individual LP activities indicates the nondestructive and repeated activation of a common source for LP events (Neuberg et al., 1998; Nakano et al., 2003a). Tremor has similar typical oscillation periods to those of LP events, but longer durations compared to LP events (Neuberg et al., 1998; Arciniega-Ceballos et al., 2000). This feature suggests that tremor may have the same source as LP events but with a sustained excitation. VLP signals are often accompanied by surface explosions and deformation (Neuberg et al., 1994; Rowe et al., 1998; Dawson et al., 1998; Chouet et al., 1999; Arciniega-Ceballos et al., 1999) and may reflect the movement of magma and/or hydrothermal fluids (e.g., Ohminato et al., 1998; Kawakatsu et al., 2000; Kumagai et al., 2001; Yamamoto et al., 2002; Chouet et al., 2003). Among these seismic signals, LP events have a great potential to diagnose the state of volcanic fluids inside the resonators (Kumagai and Chouet, 1999, 2000). The LP event is characterized by its simple waveform and consists of a superposition of decaying harmonic oscillations, which may represent the impulsive response of the resonator system. Therefore, characteristic properties of the resonator system can be determined from the complex frequencies (frequency f and quality factor Q) of the decaying harmonic oscillations. It has been recognized that the complex frequencies of LP events show temporal variations (Lesage and Surono, 1995; Gil Cruz and Chouet, 1997; Nakano et al., 1998; Kumagai and Chouet, 1999; Aoyama and Takeo, 2001; Kumagai et al., 2002a; Molina et al., 2004), which are of particular importance for diagnosing the state of fluids inside the resonators. For example, Molina et al. (2004) showed that the temporal variations in the complex frequencies of LP events observed at Tungurahua Volcano, Ecuador, can be explained by increasing the ash content within a crack-like resonator, which may be caused by repetitive injec-
235
tions of an ash-laden gas into a pre-existing crack in the volcanic conduit. Kusatsu-Shirane Volcano is an andesitic composite volcano located in central Japan (Fig. 1a). Re-
Fig. 2. Vertical velocity seismograms of LP events observed at station JIE. The amplitude scales are indicated by vertical bars at the left of the seismograms, where the length of each bar indicates an amplitude of 1 Am/s. Event dates (local time) are indicated at the upper right of each waveform.
236
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
cent volcanic activity of Kusatsu-Shirane was characterized by phreatic eruptions from the summit crater lakes, Mizugama in 1976 and Yugama in 1982 and 1983(e.g., Ossaka et al., 1997). Renewed seismic activity of VT earthquakes and LP events began in October 1989 and continued for about 3 yr (Fig. 1b). Eruptive activity in 1989–1993 was limited to a small eruption that occurred at the bottom of the Yugama crater lake on 6 January 1989, immediately before the start of the renewed activity. Kumagai et al. (2002a) analyzed LP waveforms observed at this volcano in the period between 1992 and 1993, which represents the late stage of the renewed activity. They found systematic temporal variations in the complex frequencies of LP events, and interpreted the observed temporal variations as crack growth, drying of a misty gas in the crack, and crack collapse in a hydrothermal system in response to magmatic heat beneath this volcano.
Since the analyses of Kumagai et al. (2002a) were limited to the LP events in the late stage of the renewed activity, systematic analyses of LP events that occurred in the entire period of the renewed activity are required for a comprehensive interpretation of the LP activity beneath this volcano. In this paper, we analyze the waveform data from LP events observed at this volcano during the period 1988 to 1995, which covers the period of the renewed activity between 1989 and 1993. We perform detailed spectral analyses of these events, and discuss the implications for the relation between LP events and magmatic and hydrothermal activities beneath this volcano.
2. Data Kusatsu-Shirane Volcano is monitored by a network of seven seismic stations operated by the Earthquake Research Institute of Tokyo University since
Fig. 3. Temporal variations in frequency determined for the dominant spectral peaks of LP events during 1989–1994 by spectral analyses based on the fast Fourier transform. Shaded areas indicate periods without seismic data most probably due to lack of data acquisition. Three periods (October–November 1989, July–October 1991, and September 1992–January 1993), in which the complex frequencies of LP events show systematic temporal variations, are indicated by horizontal bars.
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
1988 (Fig. 1a). Four stations (JIE, YNE, YGW, and AIM) feature three-component seismometers and three stations (JIW, MZW, and YNW) feature vertical-component seismometers. The seismometers have a natural frequency of 1 Hz and are critically damped. All the records are sampled at 120 samples/s/channel, and are acquired by an event-trigger system (Ida et al., 1989). LP events have been frequently observed at this volcano (Hamada et al., 1976; Fujita et al., 1995). The source of LP events estimated by waveform inversions (Kumagai et al., 2002b; Nakano et al., 2003a) is located at a depth of about 200 m between the Yugama and Mizugama crater lakes (Fig. 1a). We use seismic data observed by this network in the period from June 1988 through December 1995 (Nakano et al., 2003b), which covers the seismically active period in 1989–1993 (Fig. 1b). Examples of LP waveforms are shown in Fig. 2. Variations in the dominant frequency and decay characteristics are apparent in these waveforms. There are, however, peri-
237
ods without any seismic data for more than 1 week in the triggered waveform data from the Kusatsu seismic network, although more than one local and/or regional earthquakes are usually observed by this network within 1 week even in seismically inactive periods. The lack of seismic data, therefore, may be attributed to lack of acquisition in the observational system. In Fig. 3, we show these periods without seismic data for more than 10 days as shaded areas. We find more than 350 LP events in the existing seismic data between 1988 and 1995. Approximately 90% of them are recorded during the seismically active period, and are positively correlated with the VT seismicity.
3. Temporal variations in the complex frequencies of LP events Fig. 3 shows the lowest dominant peak frequencies of LP events determined by the spectral analyses
Fig. 4. Amplitude spectra (a–c) and plots of the complex frequencies of individual wave elements (d–f) for all the trial AR orders (4–60) estimated for the LP events that occurred in 1991, of which waveforms are shown in Fig. 2. The clusters of points represent clear signal, and the scattered points represent incoherent noise. The solid lines represent lines along which Q is constant. Signals within the ellipses are used to quantify the temporal variations in the complex frequencies.
238
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
based on the fast Fourier transform (Cooley and Tukey, 1965). Except for the periods when the acquisition system may be down, LP events are commonly found between 1989 and 1994. In 1988 and 1995, on the other hand, only a few LP events were recorded, and are not plotted in Fig. 3. Systematic temporal variations in frequency are found in October–November 1989, July–October 1991, and September 1992– January 1993, indicated by horizontal bars in Fig. 3. It should be noted that the temporal variations in September 1992–January 1993 have been analyzed by Kumagai et al. (2002a). No systematic temporal variations are observed in the frequencies of the LP events in 1990, although the LP activity was high during this year. We use the Sompi method (Kumazawa et al., 1990) to determine the complex frequencies of the LP events during the systematic temporal variations in frequency observed in 1989 and 1991. Sompi is a high-resolution spectral analysis method based on a homogeneous autoregressive (AR) model. The complex
frequency is defined as f–ig, where pffiffiffiffiffiffiffi f is the frequency, g is the growth rate, and i = 1 . Accordingly, the quality factor, Q, is defined as Q = f/2. The results of Sompi analyses for the waveforms in Fig. 2 are shown in Fig. 4 in the form of frequency–growth rate ( f–g) diagrams, where the complex frequencies of wave elements for all the trial AR orders (4–60) are shown. Densely populated regions in the f–g diagrams represent signals for which the complex frequencies are stably determined for different AR orders. The scattered points, on the other hand, represent incoherent noise. The complex frequency and associated errors of each waveform were conventionally determined by taking the mean value and variance of the estimated signals of all the trial AR orders, respectively. The temporal variations in the complex frequencies determined for the dominant spectral peaks of the LP events in October–November 1989 and July–October 1991 are shown in Fig. 5. We also show the temporal variations in the complex frequencies of the LP events in September 1992–January
Fig. 5. Temporal variations in frequency and Q determined for the dominant spectral peaks of LP events in three periods by Sompi spectral analyses: (a, b) October–November 1989, (c, d) July–October 1991, and (e, f) September 1992–January 1993. The frequency and Q are plotted as circles with error bars. The dashed and dotted lines show best fits of r and Q r calculated for a crack containing a water–steam mixture and misty gas, respectively. See text for details.
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
1993, which were determined by Sompi analyses of Kumagai et al. (2002a). The observed temporal variations in 1989 (Fig. 5a and b) can be divided into two periods. During the first period, before 1 November, the frequency decreases from 3.2 Hz to 1.7 Hz, while Q increases from about 40 to 100 and then decreases down to about 20, although the values of Q are highly scattered. During the second period, the frequency gradually increases up to 4.0 Hz, while Q again increases up to about 100 and then decreases down to about 30. The observed temporal variations in 1991 (Fig. 5c and d) can also be divided into two periods. During the first period, before the middle of August, the frequency increases from about 2.5 Hz to 4.5 Hz, while Q increases from about 40 to 100 and then decreases down to about 60. At the end of the first period, the frequency suddenly decreases down to 1.4 Hz, and then gradually increases up to 2 Hz during the second period. The Q factor gradually decreases down to about 20 during the second period. These temporal variations during the second period in 1991 are similar to those observed during
239
the second period in 1992–1993 (Fig. 5e and f) (Kumagai et al., 2002a).
4. The crack model We use acoustic properties of a crack containing hydrothermal fluids to interpret the temporal variations in the complex frequencies observed in 1989 and 1991 following the study of Kumagai et al. (2002a). Recent studies based on waveform inversions (Kumagai et al., 2002b; Nakano et al., 2003a) indicate that LP events at Kusatsu-Shirane Volcano originate in the resonance of a subhorizontal crack located at a depth near 200 m below the summit crater lakes, which justifies our assumption of the crack geometry for the source of LP events at this volcano. In light of the shallowness of the source, it may be reasonable to assume that the fluids in the resonator are hydrothermal fluids. Fig. 6 shows the dimensionless frequency v and Q r for a crack containing three types of hydrothermal fluids: a mixture of H2O–CO2 gases, a water–steam
Fig. 6. The quality factor due to the radiation loss (Qr ) and dimensionless frequency (r) for a crack containing different types of hydrothermal fluids: (a, d) gas mixture of H2O–CO2, (b, e) water–steam mixture, and (c, f) misty gas (after Kumagai et al., 2002a).
240
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
mixture in the form of either bubbly water or foam, and a water droplet-steam mixture (misty gas) (Kumagai et al., 2002a). Qr is the quality factor due to the radiation loss, and the quality factor due to intrinsic losses in the fluids is not included in the crack model simulations. To obtain r and Qr, the transverse mode with wavelength 2W/3 is assumed and the following parameter values are used: the rock density and Pwave velocity (a) of 2300 kg/m3 and 2500 m/s (Ida et al., 1989), respectively, and the crack geometrical parameters W/L = 0.5 and L/d = 104, where L, W, and d are crack length, width, and aperture, respectively. See Kumagai et al. (2002a) and Kumagai and Chouet (2000) for details of the procedure. For a crack containing a gas mixture of H2O–CO2, Qr ranges from 26 to 37, and r ranges from 0.25 to 0.28 depending on a volume fraction of H2O (Fig. 6a and d). For a crack containing a water–steam mixture, Qr is around a few to several tens at gas-volume fraction (GVF) near 0 and 1, and reaches a maximum value of about 100 in the vicinity of GVF of 0.5 (Fig. 6b). On the other hand, r gradually increases with increasing GVF (Fig. 6e). For a crack containing a misty gas, Qr decreases from 180 to 40 with increasing gas-weight fraction (GWF), and r increases with increasing GWF (Fig. 6c and f). We find that the observed temporal variations in f and Q during the first period in 1989 can be explained by a decrease in GVF in a water–steam mixture from almost 1 to 0 (Fig. 5a and b). The fits of r and Qr calculated for a crack containing a water–steam mixture to the observed temporal variations are shown in Fig. 5a and b. The fits are obtained by trial and error assuming a proportionality between the GVF and time. To fit the observed temporal variation in frequency, we used L = 70 m in the relation f = ra/L. The observed temporal variations in f and Q during the second period in 1989 are symmetrical to those during the first period, and therefore can be explained by an increase in GVF in a water–steam mixture from 0 to almost 1 as shown in Fig. 5a and b. The fits are again obtained by trial and error assuming a proportionality between the GVF and time, in which we used L = 70 m. These variations suggest that a water–steam mixture in the crack first gradually gets wet and then dries out over a period of 1 month. The observed temporal variations in f and Q during the first period in 1991 show similar trends to those
observed during the second period in 1989, and therefore can be explained by an increase in GVF in a water–steam mixture with a crack length L = 50 m (Fig. 5c and d). As mentioned before, the observed temporal variations during the second period in 1991 are similar to those observed during the second period in 1992–1993, and are explained by an increase in GWF in a misty gas from a few percent to almost 100% with a crack length L = 310 m (Fig. 5c and d). These results indicate that the source region of the LP events during the temporal variations in 1991 starts fairly wet as a bubbly water, then gradually dries as a foam and misty gas, and ends up completely dry, suggesting an overall drying trend. The observed temporal variations in f and Q in 1992–1993 were studied by Kumagai et al. (2002a) and we summarize their results here. The temporal variations are divided into three periods (Fig. 5e and f), and those during the second period are explained by an increase in GWF in a misty gas with a crack length L = 290 m. The variations during the first and third periods are characterized by an almost constant Q with a rapid decrease and increase of f, respectively, which are interpreted as a growth and collapse of the crack, respectively.
5. Discussion We have analyzed LP waveforms observed by the Kusatsu seismic network in the period between 1988 and 1995, in which we have found systematic temporal variations in the complex frequencies of the LP events that occurred in October–November 1989, July–October 1991, and September 1992–January 1993 based on Sompi spectral analyses. Using acoustic properties of a crack containing hydrothermal fluids, we have consistently explained these observed temporal variations, which suggest that a drying process in a crack filled with a water–steam mixture or misty gas is the cause of most of the observed temporal variations. In our interpretations based on the crack model, we have used the transverse mode with wavelength of 2W/3. Kumagai et al. (2002b) showed from waveform inversions that the dominant spectral peak of an LP event at Kusatsu-Shirane Volcano is interpreted as one of the odd modes with wavelength 2L/n, 2W/n, n = 3, 5, 7, . . . . The mode 2W/3 is the fundamental trans-
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
verse odd mode, and is the lowest dominant mode for a crack excitation triggered by a step increase in pressure applied at the center of the crack (Kumagai and Chouet, 2000, 2001). We have no clear justification for the use of the mode 2W/3 except for these points. However, the choice of the mode does not affect our essential results, but does affect the estimates of the crack dimensions, since basic features of r and Qr for a crack containing different types of hydrothermal fluids as shown in Fig. 6 are not strongly affected by the choice of mode except for the ranges of r and Qr (Kumagai and Chouet, 2001). Using the mode 2W/3, we have obtained the crack length L = 70 m in 1989, L = 50 m and 300 m in the first and second periods in 1991, respectively, and L = 290 m in the second period in 1992–1993. In 1991, the frequency rapidly decreases from 4.5 to 1.4 Hz between the first and second periods, and increases rapidly from 2 to 4.8 Hz after a gradual increase from 1.4 to 2 Hz during the second period (Fig. 5c). These variations are similar to those observed during the first, second, and third periods in 1992–1993 (Fig. 5e), which were interpreted as a growth, drying, and collapse of a crack, respectively (Kumagai et al., 2002a). These processes may have repeatedly occurred in 1991, and the difference in the crack length between the first and second periods in 1991 may be attributed to the growth of a crack in response to a pressure increase in the crack caused by a change of the fluid in the crack from a water foam to a misty gas. In view of the similar crack lengths between 1989 and the first period in 1991 and between the second periods in 1991 and 1992–1993, it may be reasonable to assume that repeated triggering of a single crack was the source of the LP events in the entire period of the renewed activity. The observed values of Q show considerable scatter especially during the first period in 1989 (Fig. 5b), while the overall trends in this period are interpreted as a decrease in GVF in a water–steam mixture from almost 100% (steam) down to a few percent (bubbly water). The observed quality factor Q of the LP event may be expressed as 1 Q1 ¼ Q1 r þ Qi ;
ð1Þ
where Qi is the quality factor due to intrinsic losses in the fluid, which are not included in the crack model
241
simulations. In case of a bubbly water, Qi depends on GVF, bubble radius, and frequency of the traveling wave (Chouet, 1996b; Nakano et al., 1998; Kumagai and Chouet, 2000). For example, Q1 becomes comi parable to or larger than Q r1 in a bubbly water containing bubbles whose radii are larger than 1 mm (Kumagai and Chouet, 2000). Although there are no reliable estimates of Qi in a water–steam mixture whose GVF is larger than 0.1 (foam), Q1 may bei come larger Q r1 in a water foam containing larger bubbles. Intrinsic losses, therefore, may cause scatter in the observed Q of the LP events. It should be also noted that scatter in the observed Q may be caused by errors and instabilities that stem from fitting exponentially decaying oscillations in the spectral analysis. The observed scatter in Q may originate in the combined effect of these two mechanisms. Most of the temporal variations analyzed in our study suggest drying of the hydrothermal fluid in a crack for LP events between July and November, in which similar patterns in the temporal variations are found between the second period in 1989 and the first period in 1991 and between the second period in 1991 and the second period in 1992–1993. The repeated nature observed in similar seasons strongly suggests the existence of a seasonal variation in the occurrence of LP events, although no seismic data exist in the first halves of 1991 and 1992. One of the sources of hydrothermal fluids beneath this volcano is meteoritic water from rain and snow (Hirabayashi, 1999; Ohba et al., 2000). Most precipitation occurs between June and October in a usual year in this region, and snow on Kusatsu-Shirane melts between March and May. It may be possible that the precipitated and melted water enhanced in March–October affects the occurrence of LP events. The seepage of water into the source region of LP events, although the time required for water at the surface to reach the source region is unknown, may trigger more LP events in response to a pressure increase due to bubble and gas formation in water caused by heat transferred by the flux of volcanic gases from the magma beneath this volcano. As the seepage of water subsides, heat gradually changes the fluid in the crack from a wet water– steam mixture or misty gas to a dry gas. After the fluid in the crack completely dries out, the LP seismicity wanes due to the lack of hydrothermal water to trigger LP events.
242
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
We thus interpret the overall LP and VT activities based on the results obtained above as follows. The VT earthquakes represent the brittle response of the volcanic rock associated with the movement of volcanic gases from the magma, and therefore may be
viewed as a qualitative measure of the gas movement and thus heat supply to the hydrothermal system. The heightened activity of both LP events and VT earthquakes began in October 1989 (Figs. 1b and 7a), suggesting an increased heat supply from the
Fig. 7. The daily and monthly number of LP events observed at Kusatsu-Shirane Volcano. Vertical bars and circles with dotted lines indicate daily and monthly number, respectively. Shaded areas indicate periods without seismic data.
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
magma. At this time, the hydrothermal system was fairly wet due to the seepage of precipitated and melted water, resulting in a wet hydrothermal crack at the source of LP events during the first period of this activity. Further heating of the hydrothermal system gradually changed the fluid in the crack from a wet water–steam mixture to a dry gas during the second period. As the fluid in the crack completely dried out, the LP activity gradually waned (Fig. 7a). In 1990, the VT activity was the highest during this renewed activity (Fig. 1b), whereas the LP activity in the first half of this year was relatively low (Fig. 7b), implying that the hydrothermal system remained dry during this period in spite of a heightened heat supply from the magma. The LP activity heightened after September in response to a water supply to the heated hydrothermal system, although no systematic temporal variations in the complex frequencies of LP events were observed. The lack of systematic temporal variations may be attributed to unstable conditions of the source region due to the high flux of volcanic gases from the magma. In 1991–1993, the VT activity gradually decreased (Fig. 1b), while the LP activity remained high especially in August 1991 and September in 1992 (Fig. 7c and d), which may be caused by the seepage of water into the heated hydrothermal system. After 1993, both the VT and LP activity waned, and heating of the hydrothermal system subsided.
6. Conclusions We have analyzed the complex frequencies of the LP events observed at Kusatsu-Shirane Volcano, and found systematic temporal variations in the complex frequencies of LP events that occurred in 1989, 1991, and 1992–1993. The temporal variations can be consistently explained by resonance of a crack filled with hydrothermal fluids. Most of the temporal variations are interpreted as gradual drying of the fluid in the crack caused by a seasonally variable water supply to the hydrothermal system, which may have been heated by the flux of volcanic gases from the magma. Our study further demonstrates that the complex frequencies of LP events are useful to investigate internal physical processes and to monitor hydrothermal and magmatic systems beneath volcanoes.
243
Acknowledgments We thank Yoshiaki Ida, Jun Oikawa, and Keigo Yamamoto for sharing the data recorded by the Earthquake Research Institute of the University of Tokyo. We also thank Kenji Nogami for the useful discussion. We gratefully appreciate valuable comments from Phillip Dawson. This work was supported by the Japanese Ministry of Education, Culture, Sorts, Science and Technology under grant 14740265 to M.N.
References Aoyama, H., Takeo, M., 2001. Wave properties and focal mechanism of N-type earthquakes at Asama volcano. J. Volcanol. Geotherm. Res. 105, 163 – 182. Arciniega-Ceballos, A., Chouet, B.A., Dawson, P., 1999. Very longperiod signals associated with vulcanian explosions at Popocatepetl volcano, Mexico. Geophys. Res. Lett. 26, 3013 – 3016. Arciniega-Ceballos, A., Valdes-Gonzalez, C., Dawson, P., 2000. Temporal and spectral characteristics of seismicity observed at Popocatepetl volcano, central Mexico. J. Volcanol. Geotherm. Res. 102, 207 – 216. Chouet, B., 1986. Dynamics of a fluid-driven crack in three dimensions by the finite difference method. J. Geophys. Res. 91, 13967 – 13992. Chouet, B.A., 1996a. Long-period volcano seismicity: its source and use in eruption forecasting. Nature 380, 309 – 316. Chouet, B., 1996b. New methods and future trends in seismological volcano monitoring. In: Scarpa, R., Tilling, R. (Eds.), Monitoring and Mitigation of Volcano Hazards. Springer-Verlag, New York, pp. 23 – 97. Chouet, B., Saccorotti, G., Dawson, P., Martini, M., Scarpa, R., De Luca, G., Milana, G., Cattaneo, M., 1999. Broadband measurements of the sources of explosions at Stromboli Volcano, Italy. Geophys. Res. Lett. 26, 1937 – 1940. Chouet, B., Dawson, P., Ohminato, T., Martini, M., Saccorotti, G., Giudicepietro, F., De Luca, G., Milana, G., Scarpa, R., 2003. Source mechanisms of explosions at Stromboli Volcano, Italy, determined from moment-tensor inversions of very-long-period data. J. Geophys. Res. 108, 2019. doi:10.1029/2002JB001919. Cooley, J.W., Tukey, J.W., 1965. An algorithm for machine calculation of complex Fourier series. Math. Comput. 19, 297 – 301. Dawson, P.B., Dietel, C., Chouet, B.A., Honma, K., Ohminato, T., Okubo, P., 1998. A digitally telemeterd broadband seismic network at Kilauea Volcano, Hawaii. Open-File Rep. (U.S. Geol. Surv.) 98–108, 1 – 121. Fujita, E., Ida, Y., Oikawa, J., 1995. Eigen oscillation of a fluid sphere and source mechanism of harmonic volcanic tremor. J. Volcanol. Geotherm. Res. 69, 365 – 378. Gil Cruz, F., Chouet, B.A., 1997. Long-period events, the most characteristic seismicity accompanying the emplacement and extrusion of a lava dome in Galeras Volcano, Colombia, in 1991. J. Volcanol. Geotherm. Res. 77, 121 – 158.
244
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
Hamada, N., Jingu, H., Ikumoto, K., 1976. On the volcanic earthquake with slowly decaying coda wave. Bull. Volcanol. Soc. Jpn. 21, 167 – 183 (in Japanese with English abstract). Hirabayashi, J., 1999. Formation of volcanic fluid reservoir and volcanic activity. J. Balneol. Soc. Jpn. 49, 99 – 105. Ida, Y., Osada, N., Sawada, M., Koyama, E., Kagiyama, T., 1989. Seismological study based on recently installed permanent stations and a small eruptive event on January 6, 1989 at Kusatsu-Shirane volcano. Bull. Earthq. Res. Inst. Univ. Tokyo 64, 325 – 345 (in Japanese with English abstract). Japan Meteorological Agency (JMA), 1995. Report of coordinating committee for prediction of volcanic eruption. Rep. 61, Tokyo, 97. Kawakatsu, H., Kaneshima, S., Matsubayashi, H., Ohminato, T., Sudo, Y., Tsutsui, T., Uhira, K., Yamasato, H., Ito, H., Legrand, D., 2000. Aso94: Aso seismic observation with broadband instruments. J. Volcanol. Geotherm. Res. 101, 129 – 154. Kumagai, H., Chouet, B.A., 1999. The complex frequencies of long-period seismic events as probes of fluid composition beneath volcanoes. Geophys. J. Int. 138, F7 – F12. Kumagai, H., Chouet, B.A., 2000. Acoustic properties of a crack containing magmatic or hydrothermal fluids. J. Geophys. Res. 105, 25493 – 25512. Kumagai, H., Chouet, B.A., 2001. The dependence of acoustic properties of a crack on the resonance mode and geometry. Geophys. Res. Lett. 28, 3325 – 3328. Kumagai, H., Ohminato, T., Nakano, M., Ooi, M., Kubo, A., Inoue, H., Oikawa, J., 2001. Very-long-period seismic signals and caldera formation at Miyake Island Japan. Science 293, 687 – 690. Kumagai, H., Chouet, B.A., Nakano, M., 2002a. Temporal evolution of a hydrothermal system in Kusatsu-Shirane Volcano, Japan, inferred from the complex frequencies of long-period events. J. Geophys. Res. 107, 2236. doi:10.1029/2001JB000653. Kumagai, H., Chouet, B.A., Nakano, M., 2002b. Waveform inversion of oscillatory signatures in long-period events beneath volcanoes. J. Geophys. Res. 107, 2301. doi:10.1029/2001JB001704. Kumazawa, M., Imanishi, Y., Fukao, Y., Furumoto, M., Yamamoto, A., 1990. A theory of spectral analysis based on the characteristic property of a linear dynamic system. Geophys. J. Int. 101, 613 – 630. Lesage, P., Surono,, 1995. Seismic precursors of the February 10, 1990 eruption of Kelut volcano, Java. J. Volcanol. Geotherm. Res. 65, 135 – 146.
Molina, I., Kumagai, H., Yepes, H., 2004. Resonances of a volcanic conduit triggered by repetitive injections of an ash-laden gas. Geophys. Res. Lett. 31. doi:10.1029/2003GL018934. Nakano, M., Kumagai, H., Kumazawa, M., Yamaoka, K., Chouet, B.A., 1998. The excitation and characteristic frequency of the long-period volcanic event: an approach based on an inhomogeneous autoregressive model of a linear dynamic system. J. Geophys. Res. 103, 10031 – 10046. Nakano, M., Kumagai, H., Chouet, B.A., 2003a. Source mechanism of long-period events at Kusatsu-Shirane Volcano, Japan, inferred from waveform inversion of the effective excitation functions. J. Volcanol. Geotherm. Res. 122, 149 – 164. Nakano, M., Kumagai, H., Oikawa, J., 2003b. Retrieval of seismic data recorded on open-reel-type magnetic tapes (MT) by using existing devices. Tech. Res Rep.-Univ. Tokyo. Earthq. Res. Inst. 9, 1 – 7 (in Japanese with English abstract). Neuberg, J., Luckett, R., Ripepe, M., Braun, T., 1994. Highlights from a seismic broadband array on Stromboli volcano. Geophys. Res. Lett. 21, 749 – 752. Neuberg, J., Baptie, B., Luckett, R., Stewart, R., 1998. Results from the broad-band seismic network on Montserrat. Geophys. Res. Lett. 25, 3661 – 3664. Ohba, T., Hirabayashi, J., Nogami, K., 2000. D/H and 18O/16O ratios of water in the crater lake at Kusatsu-Shirane volcano Japan. J. Volcanol. Geotherm. Res. 97, 329 – 346. Ohminato, T., Chouet, B.A., Dawson, P., Kedar, S., 1998. Waveform inversion of very long period impulsive signals associated with magmatic injection beneath Kilauea Volcano, Hawaii. J. Geophys. Res. 103, 23839 – 23862. Ossaka, J., Ossaka, T., Hirabayashi, J., Oi, T., Ohba, T., Nogami, K., Kikawada, Y., Yamano, M., Yui, M., Fukuhara, H., 1997. Volcanic activity of Kusatsu-Shirane volcano, Gumma, and secular change in water quality of crater lake Yugama. Chikyukagaku (Geochemistry) 31, 119 – 128 (in Japanese with English abstract). Rowe, C.A., Aster, R.C., Kyle, P.R., Schlue, J.W., 1998. Broadband recording of Strombolian explosions and associated very-longperiod seismic signals on Mount Erebus volcano, Ross Island, Antarctica. Geophys. Res. Lett. 25, 2297 – 2300. Yamamoto, M., Kawakatsu, H., Yomogida, K., Koyama, J., 2002. Long-period (12 sec) volcanic tremor observed at Usu 2000 eruption: seismological detection of a deep magma plumbing system. Geophys. Res. Lett. 29. doi:10.1029/2001GL013996.
Response of a hydrothermal system to magmatic heat inferred from temporal variations in the complex frequencies of long-period events at Kusatsu-Shirane Volcano, Japan Masaru Nakanoa,*, Hiroyuki Kumagaib a
Graduate School of Environmental Studies, Nagoya University, Nagoya, 464-8601, Japan (Now at US Geological Survey, Menlo Park, CA, USA) b National Research Institute for Earth Science and Disaster Prevention, Tsukuba, Japan Received 4 June 2004; received in revised form 30 March 2005; accepted 4 April 2005
Abstract We investigate temporal variations in the complex frequencies (frequency and quality factor Q) of long-period (LP) events that occurred at Kusatsu-Shirane Volcano, central Japan. We analyze LP waveforms observed at this volcano in the period between 1988 and 1995, which covers a seismically active period between 1989 and 1993. Systematic temporal variations in the complex frequencies are observed in October–November 1989, July–October 1991, and September 1992–January 1993. We use acoustic properties of a crack filled with hydrothermal fluids to interpret the observed temporal variations in the complex frequencies. The temporal variations in October–November 1989 can be divided into two periods, which are explained by a gradual decrease and increase of a gas-volume fraction in a water–steam mixture in a crack, respectively. The temporal variations in July–October 1991 can be also divided into two periods. These variations in the first and second periods are similar to those observed in November 1989 and in September–November 1992, respectively, and are interpreted as drying of a water– steam mixture and misty gas in a crack, respectively. The repeated nature of the temporal variations observed in similar seasons between July and November suggests the existence of seasonality in the occurrence of LP events. This may be caused by a seasonally variable meteoritic water supply to a hydrothermal system, which may have been heated by the flux of volcanic gases from magma beneath this volcano. D 2005 Elsevier B.V. All rights reserved. Keywords: long-period (LP) events; complex frequency; temporal variation; fluid-filled crack; hydrothermal system
1. Introduction
* Corresponding author. Fax: +1 650 329 5203. E-mail address: [email protected] (M. Nakano). 0377-0273/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.jvolgeores.2005.04.003
Various types of seismic signals such as volcanotectonic (VT) earthquakes, long-period (LP) and verylong-period (VLP) events, and tremor are observed at
234
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
active volcanoes. These signals may be caused by internal volcanic activities associated with fluid movement, heat and gas supply from magma, and interactions between magma and underground water. A
swarm of VT earthquakes is often the first sign of renewed volcanic activity. VT earthquakes are indistinguishable from ordinary double-couple tectonic earthquakes, and represent the brittle response of the
Fig. 1. (a) Locations of seismic stations operated by the Earthquake Research Institute of University of Tokyo at Kusatsu-Shirane volcano. Solid and open triangles indicate stations with three-component seismometers and vertical-component seismometers, respectively. A cross between Yugama and Mizugama marks the epicenter region of LP events determined by Kumagai et al. (2002b) and Nakano et al. (2003a). The inset shows the location of Kusatsu-Shirane volcano in central Japan. (b) The monthly number of VT events that occurred at Kusatsu-Shirane Volcano in 1988–1995 (based on data of Japan Meteorological Agency, 1995).
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
volcanic rock associated with fluid movement beneath a volcano (e.g., Chouet, 1996a). LP events with typical oscillation periods between 0.2 and 2 s are considered as acoustic vibrations of a fluid-filled resonator in magmatic and hydrothermal systems, which may be triggered by pressure fluctuations associated with mass transport and/or thermodynamic processes of fluids in the resonator (e.g., Chouet, 1986, 1996a,b). The similarity of LP waveform signatures in individual LP activities indicates the nondestructive and repeated activation of a common source for LP events (Neuberg et al., 1998; Nakano et al., 2003a). Tremor has similar typical oscillation periods to those of LP events, but longer durations compared to LP events (Neuberg et al., 1998; Arciniega-Ceballos et al., 2000). This feature suggests that tremor may have the same source as LP events but with a sustained excitation. VLP signals are often accompanied by surface explosions and deformation (Neuberg et al., 1994; Rowe et al., 1998; Dawson et al., 1998; Chouet et al., 1999; Arciniega-Ceballos et al., 1999) and may reflect the movement of magma and/or hydrothermal fluids (e.g., Ohminato et al., 1998; Kawakatsu et al., 2000; Kumagai et al., 2001; Yamamoto et al., 2002; Chouet et al., 2003). Among these seismic signals, LP events have a great potential to diagnose the state of volcanic fluids inside the resonators (Kumagai and Chouet, 1999, 2000). The LP event is characterized by its simple waveform and consists of a superposition of decaying harmonic oscillations, which may represent the impulsive response of the resonator system. Therefore, characteristic properties of the resonator system can be determined from the complex frequencies (frequency f and quality factor Q) of the decaying harmonic oscillations. It has been recognized that the complex frequencies of LP events show temporal variations (Lesage and Surono, 1995; Gil Cruz and Chouet, 1997; Nakano et al., 1998; Kumagai and Chouet, 1999; Aoyama and Takeo, 2001; Kumagai et al., 2002a; Molina et al., 2004), which are of particular importance for diagnosing the state of fluids inside the resonators. For example, Molina et al. (2004) showed that the temporal variations in the complex frequencies of LP events observed at Tungurahua Volcano, Ecuador, can be explained by increasing the ash content within a crack-like resonator, which may be caused by repetitive injec-
235
tions of an ash-laden gas into a pre-existing crack in the volcanic conduit. Kusatsu-Shirane Volcano is an andesitic composite volcano located in central Japan (Fig. 1a). Re-
Fig. 2. Vertical velocity seismograms of LP events observed at station JIE. The amplitude scales are indicated by vertical bars at the left of the seismograms, where the length of each bar indicates an amplitude of 1 Am/s. Event dates (local time) are indicated at the upper right of each waveform.
236
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
cent volcanic activity of Kusatsu-Shirane was characterized by phreatic eruptions from the summit crater lakes, Mizugama in 1976 and Yugama in 1982 and 1983(e.g., Ossaka et al., 1997). Renewed seismic activity of VT earthquakes and LP events began in October 1989 and continued for about 3 yr (Fig. 1b). Eruptive activity in 1989–1993 was limited to a small eruption that occurred at the bottom of the Yugama crater lake on 6 January 1989, immediately before the start of the renewed activity. Kumagai et al. (2002a) analyzed LP waveforms observed at this volcano in the period between 1992 and 1993, which represents the late stage of the renewed activity. They found systematic temporal variations in the complex frequencies of LP events, and interpreted the observed temporal variations as crack growth, drying of a misty gas in the crack, and crack collapse in a hydrothermal system in response to magmatic heat beneath this volcano.
Since the analyses of Kumagai et al. (2002a) were limited to the LP events in the late stage of the renewed activity, systematic analyses of LP events that occurred in the entire period of the renewed activity are required for a comprehensive interpretation of the LP activity beneath this volcano. In this paper, we analyze the waveform data from LP events observed at this volcano during the period 1988 to 1995, which covers the period of the renewed activity between 1989 and 1993. We perform detailed spectral analyses of these events, and discuss the implications for the relation between LP events and magmatic and hydrothermal activities beneath this volcano.
2. Data Kusatsu-Shirane Volcano is monitored by a network of seven seismic stations operated by the Earthquake Research Institute of Tokyo University since
Fig. 3. Temporal variations in frequency determined for the dominant spectral peaks of LP events during 1989–1994 by spectral analyses based on the fast Fourier transform. Shaded areas indicate periods without seismic data most probably due to lack of data acquisition. Three periods (October–November 1989, July–October 1991, and September 1992–January 1993), in which the complex frequencies of LP events show systematic temporal variations, are indicated by horizontal bars.
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
1988 (Fig. 1a). Four stations (JIE, YNE, YGW, and AIM) feature three-component seismometers and three stations (JIW, MZW, and YNW) feature vertical-component seismometers. The seismometers have a natural frequency of 1 Hz and are critically damped. All the records are sampled at 120 samples/s/channel, and are acquired by an event-trigger system (Ida et al., 1989). LP events have been frequently observed at this volcano (Hamada et al., 1976; Fujita et al., 1995). The source of LP events estimated by waveform inversions (Kumagai et al., 2002b; Nakano et al., 2003a) is located at a depth of about 200 m between the Yugama and Mizugama crater lakes (Fig. 1a). We use seismic data observed by this network in the period from June 1988 through December 1995 (Nakano et al., 2003b), which covers the seismically active period in 1989–1993 (Fig. 1b). Examples of LP waveforms are shown in Fig. 2. Variations in the dominant frequency and decay characteristics are apparent in these waveforms. There are, however, peri-
237
ods without any seismic data for more than 1 week in the triggered waveform data from the Kusatsu seismic network, although more than one local and/or regional earthquakes are usually observed by this network within 1 week even in seismically inactive periods. The lack of seismic data, therefore, may be attributed to lack of acquisition in the observational system. In Fig. 3, we show these periods without seismic data for more than 10 days as shaded areas. We find more than 350 LP events in the existing seismic data between 1988 and 1995. Approximately 90% of them are recorded during the seismically active period, and are positively correlated with the VT seismicity.
3. Temporal variations in the complex frequencies of LP events Fig. 3 shows the lowest dominant peak frequencies of LP events determined by the spectral analyses
Fig. 4. Amplitude spectra (a–c) and plots of the complex frequencies of individual wave elements (d–f) for all the trial AR orders (4–60) estimated for the LP events that occurred in 1991, of which waveforms are shown in Fig. 2. The clusters of points represent clear signal, and the scattered points represent incoherent noise. The solid lines represent lines along which Q is constant. Signals within the ellipses are used to quantify the temporal variations in the complex frequencies.
238
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
based on the fast Fourier transform (Cooley and Tukey, 1965). Except for the periods when the acquisition system may be down, LP events are commonly found between 1989 and 1994. In 1988 and 1995, on the other hand, only a few LP events were recorded, and are not plotted in Fig. 3. Systematic temporal variations in frequency are found in October–November 1989, July–October 1991, and September 1992– January 1993, indicated by horizontal bars in Fig. 3. It should be noted that the temporal variations in September 1992–January 1993 have been analyzed by Kumagai et al. (2002a). No systematic temporal variations are observed in the frequencies of the LP events in 1990, although the LP activity was high during this year. We use the Sompi method (Kumazawa et al., 1990) to determine the complex frequencies of the LP events during the systematic temporal variations in frequency observed in 1989 and 1991. Sompi is a high-resolution spectral analysis method based on a homogeneous autoregressive (AR) model. The complex
frequency is defined as f–ig, where pffiffiffiffiffiffiffi f is the frequency, g is the growth rate, and i = 1 . Accordingly, the quality factor, Q, is defined as Q = f/2. The results of Sompi analyses for the waveforms in Fig. 2 are shown in Fig. 4 in the form of frequency–growth rate ( f–g) diagrams, where the complex frequencies of wave elements for all the trial AR orders (4–60) are shown. Densely populated regions in the f–g diagrams represent signals for which the complex frequencies are stably determined for different AR orders. The scattered points, on the other hand, represent incoherent noise. The complex frequency and associated errors of each waveform were conventionally determined by taking the mean value and variance of the estimated signals of all the trial AR orders, respectively. The temporal variations in the complex frequencies determined for the dominant spectral peaks of the LP events in October–November 1989 and July–October 1991 are shown in Fig. 5. We also show the temporal variations in the complex frequencies of the LP events in September 1992–January
Fig. 5. Temporal variations in frequency and Q determined for the dominant spectral peaks of LP events in three periods by Sompi spectral analyses: (a, b) October–November 1989, (c, d) July–October 1991, and (e, f) September 1992–January 1993. The frequency and Q are plotted as circles with error bars. The dashed and dotted lines show best fits of r and Q r calculated for a crack containing a water–steam mixture and misty gas, respectively. See text for details.
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
1993, which were determined by Sompi analyses of Kumagai et al. (2002a). The observed temporal variations in 1989 (Fig. 5a and b) can be divided into two periods. During the first period, before 1 November, the frequency decreases from 3.2 Hz to 1.7 Hz, while Q increases from about 40 to 100 and then decreases down to about 20, although the values of Q are highly scattered. During the second period, the frequency gradually increases up to 4.0 Hz, while Q again increases up to about 100 and then decreases down to about 30. The observed temporal variations in 1991 (Fig. 5c and d) can also be divided into two periods. During the first period, before the middle of August, the frequency increases from about 2.5 Hz to 4.5 Hz, while Q increases from about 40 to 100 and then decreases down to about 60. At the end of the first period, the frequency suddenly decreases down to 1.4 Hz, and then gradually increases up to 2 Hz during the second period. The Q factor gradually decreases down to about 20 during the second period. These temporal variations during the second period in 1991 are similar to those observed during
239
the second period in 1992–1993 (Fig. 5e and f) (Kumagai et al., 2002a).
4. The crack model We use acoustic properties of a crack containing hydrothermal fluids to interpret the temporal variations in the complex frequencies observed in 1989 and 1991 following the study of Kumagai et al. (2002a). Recent studies based on waveform inversions (Kumagai et al., 2002b; Nakano et al., 2003a) indicate that LP events at Kusatsu-Shirane Volcano originate in the resonance of a subhorizontal crack located at a depth near 200 m below the summit crater lakes, which justifies our assumption of the crack geometry for the source of LP events at this volcano. In light of the shallowness of the source, it may be reasonable to assume that the fluids in the resonator are hydrothermal fluids. Fig. 6 shows the dimensionless frequency v and Q r for a crack containing three types of hydrothermal fluids: a mixture of H2O–CO2 gases, a water–steam
Fig. 6. The quality factor due to the radiation loss (Qr ) and dimensionless frequency (r) for a crack containing different types of hydrothermal fluids: (a, d) gas mixture of H2O–CO2, (b, e) water–steam mixture, and (c, f) misty gas (after Kumagai et al., 2002a).
240
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
mixture in the form of either bubbly water or foam, and a water droplet-steam mixture (misty gas) (Kumagai et al., 2002a). Qr is the quality factor due to the radiation loss, and the quality factor due to intrinsic losses in the fluids is not included in the crack model simulations. To obtain r and Qr, the transverse mode with wavelength 2W/3 is assumed and the following parameter values are used: the rock density and Pwave velocity (a) of 2300 kg/m3 and 2500 m/s (Ida et al., 1989), respectively, and the crack geometrical parameters W/L = 0.5 and L/d = 104, where L, W, and d are crack length, width, and aperture, respectively. See Kumagai et al. (2002a) and Kumagai and Chouet (2000) for details of the procedure. For a crack containing a gas mixture of H2O–CO2, Qr ranges from 26 to 37, and r ranges from 0.25 to 0.28 depending on a volume fraction of H2O (Fig. 6a and d). For a crack containing a water–steam mixture, Qr is around a few to several tens at gas-volume fraction (GVF) near 0 and 1, and reaches a maximum value of about 100 in the vicinity of GVF of 0.5 (Fig. 6b). On the other hand, r gradually increases with increasing GVF (Fig. 6e). For a crack containing a misty gas, Qr decreases from 180 to 40 with increasing gas-weight fraction (GWF), and r increases with increasing GWF (Fig. 6c and f). We find that the observed temporal variations in f and Q during the first period in 1989 can be explained by a decrease in GVF in a water–steam mixture from almost 1 to 0 (Fig. 5a and b). The fits of r and Qr calculated for a crack containing a water–steam mixture to the observed temporal variations are shown in Fig. 5a and b. The fits are obtained by trial and error assuming a proportionality between the GVF and time. To fit the observed temporal variation in frequency, we used L = 70 m in the relation f = ra/L. The observed temporal variations in f and Q during the second period in 1989 are symmetrical to those during the first period, and therefore can be explained by an increase in GVF in a water–steam mixture from 0 to almost 1 as shown in Fig. 5a and b. The fits are again obtained by trial and error assuming a proportionality between the GVF and time, in which we used L = 70 m. These variations suggest that a water–steam mixture in the crack first gradually gets wet and then dries out over a period of 1 month. The observed temporal variations in f and Q during the first period in 1991 show similar trends to those
observed during the second period in 1989, and therefore can be explained by an increase in GVF in a water–steam mixture with a crack length L = 50 m (Fig. 5c and d). As mentioned before, the observed temporal variations during the second period in 1991 are similar to those observed during the second period in 1992–1993, and are explained by an increase in GWF in a misty gas from a few percent to almost 100% with a crack length L = 310 m (Fig. 5c and d). These results indicate that the source region of the LP events during the temporal variations in 1991 starts fairly wet as a bubbly water, then gradually dries as a foam and misty gas, and ends up completely dry, suggesting an overall drying trend. The observed temporal variations in f and Q in 1992–1993 were studied by Kumagai et al. (2002a) and we summarize their results here. The temporal variations are divided into three periods (Fig. 5e and f), and those during the second period are explained by an increase in GWF in a misty gas with a crack length L = 290 m. The variations during the first and third periods are characterized by an almost constant Q with a rapid decrease and increase of f, respectively, which are interpreted as a growth and collapse of the crack, respectively.
5. Discussion We have analyzed LP waveforms observed by the Kusatsu seismic network in the period between 1988 and 1995, in which we have found systematic temporal variations in the complex frequencies of the LP events that occurred in October–November 1989, July–October 1991, and September 1992–January 1993 based on Sompi spectral analyses. Using acoustic properties of a crack containing hydrothermal fluids, we have consistently explained these observed temporal variations, which suggest that a drying process in a crack filled with a water–steam mixture or misty gas is the cause of most of the observed temporal variations. In our interpretations based on the crack model, we have used the transverse mode with wavelength of 2W/3. Kumagai et al. (2002b) showed from waveform inversions that the dominant spectral peak of an LP event at Kusatsu-Shirane Volcano is interpreted as one of the odd modes with wavelength 2L/n, 2W/n, n = 3, 5, 7, . . . . The mode 2W/3 is the fundamental trans-
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
verse odd mode, and is the lowest dominant mode for a crack excitation triggered by a step increase in pressure applied at the center of the crack (Kumagai and Chouet, 2000, 2001). We have no clear justification for the use of the mode 2W/3 except for these points. However, the choice of the mode does not affect our essential results, but does affect the estimates of the crack dimensions, since basic features of r and Qr for a crack containing different types of hydrothermal fluids as shown in Fig. 6 are not strongly affected by the choice of mode except for the ranges of r and Qr (Kumagai and Chouet, 2001). Using the mode 2W/3, we have obtained the crack length L = 70 m in 1989, L = 50 m and 300 m in the first and second periods in 1991, respectively, and L = 290 m in the second period in 1992–1993. In 1991, the frequency rapidly decreases from 4.5 to 1.4 Hz between the first and second periods, and increases rapidly from 2 to 4.8 Hz after a gradual increase from 1.4 to 2 Hz during the second period (Fig. 5c). These variations are similar to those observed during the first, second, and third periods in 1992–1993 (Fig. 5e), which were interpreted as a growth, drying, and collapse of a crack, respectively (Kumagai et al., 2002a). These processes may have repeatedly occurred in 1991, and the difference in the crack length between the first and second periods in 1991 may be attributed to the growth of a crack in response to a pressure increase in the crack caused by a change of the fluid in the crack from a water foam to a misty gas. In view of the similar crack lengths between 1989 and the first period in 1991 and between the second periods in 1991 and 1992–1993, it may be reasonable to assume that repeated triggering of a single crack was the source of the LP events in the entire period of the renewed activity. The observed values of Q show considerable scatter especially during the first period in 1989 (Fig. 5b), while the overall trends in this period are interpreted as a decrease in GVF in a water–steam mixture from almost 100% (steam) down to a few percent (bubbly water). The observed quality factor Q of the LP event may be expressed as 1 Q1 ¼ Q1 r þ Qi ;
ð1Þ
where Qi is the quality factor due to intrinsic losses in the fluid, which are not included in the crack model
241
simulations. In case of a bubbly water, Qi depends on GVF, bubble radius, and frequency of the traveling wave (Chouet, 1996b; Nakano et al., 1998; Kumagai and Chouet, 2000). For example, Q1 becomes comi parable to or larger than Q r1 in a bubbly water containing bubbles whose radii are larger than 1 mm (Kumagai and Chouet, 2000). Although there are no reliable estimates of Qi in a water–steam mixture whose GVF is larger than 0.1 (foam), Q1 may bei come larger Q r1 in a water foam containing larger bubbles. Intrinsic losses, therefore, may cause scatter in the observed Q of the LP events. It should be also noted that scatter in the observed Q may be caused by errors and instabilities that stem from fitting exponentially decaying oscillations in the spectral analysis. The observed scatter in Q may originate in the combined effect of these two mechanisms. Most of the temporal variations analyzed in our study suggest drying of the hydrothermal fluid in a crack for LP events between July and November, in which similar patterns in the temporal variations are found between the second period in 1989 and the first period in 1991 and between the second period in 1991 and the second period in 1992–1993. The repeated nature observed in similar seasons strongly suggests the existence of a seasonal variation in the occurrence of LP events, although no seismic data exist in the first halves of 1991 and 1992. One of the sources of hydrothermal fluids beneath this volcano is meteoritic water from rain and snow (Hirabayashi, 1999; Ohba et al., 2000). Most precipitation occurs between June and October in a usual year in this region, and snow on Kusatsu-Shirane melts between March and May. It may be possible that the precipitated and melted water enhanced in March–October affects the occurrence of LP events. The seepage of water into the source region of LP events, although the time required for water at the surface to reach the source region is unknown, may trigger more LP events in response to a pressure increase due to bubble and gas formation in water caused by heat transferred by the flux of volcanic gases from the magma beneath this volcano. As the seepage of water subsides, heat gradually changes the fluid in the crack from a wet water– steam mixture or misty gas to a dry gas. After the fluid in the crack completely dries out, the LP seismicity wanes due to the lack of hydrothermal water to trigger LP events.
242
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
We thus interpret the overall LP and VT activities based on the results obtained above as follows. The VT earthquakes represent the brittle response of the volcanic rock associated with the movement of volcanic gases from the magma, and therefore may be
viewed as a qualitative measure of the gas movement and thus heat supply to the hydrothermal system. The heightened activity of both LP events and VT earthquakes began in October 1989 (Figs. 1b and 7a), suggesting an increased heat supply from the
Fig. 7. The daily and monthly number of LP events observed at Kusatsu-Shirane Volcano. Vertical bars and circles with dotted lines indicate daily and monthly number, respectively. Shaded areas indicate periods without seismic data.
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
magma. At this time, the hydrothermal system was fairly wet due to the seepage of precipitated and melted water, resulting in a wet hydrothermal crack at the source of LP events during the first period of this activity. Further heating of the hydrothermal system gradually changed the fluid in the crack from a wet water–steam mixture to a dry gas during the second period. As the fluid in the crack completely dried out, the LP activity gradually waned (Fig. 7a). In 1990, the VT activity was the highest during this renewed activity (Fig. 1b), whereas the LP activity in the first half of this year was relatively low (Fig. 7b), implying that the hydrothermal system remained dry during this period in spite of a heightened heat supply from the magma. The LP activity heightened after September in response to a water supply to the heated hydrothermal system, although no systematic temporal variations in the complex frequencies of LP events were observed. The lack of systematic temporal variations may be attributed to unstable conditions of the source region due to the high flux of volcanic gases from the magma. In 1991–1993, the VT activity gradually decreased (Fig. 1b), while the LP activity remained high especially in August 1991 and September in 1992 (Fig. 7c and d), which may be caused by the seepage of water into the heated hydrothermal system. After 1993, both the VT and LP activity waned, and heating of the hydrothermal system subsided.
6. Conclusions We have analyzed the complex frequencies of the LP events observed at Kusatsu-Shirane Volcano, and found systematic temporal variations in the complex frequencies of LP events that occurred in 1989, 1991, and 1992–1993. The temporal variations can be consistently explained by resonance of a crack filled with hydrothermal fluids. Most of the temporal variations are interpreted as gradual drying of the fluid in the crack caused by a seasonally variable water supply to the hydrothermal system, which may have been heated by the flux of volcanic gases from the magma. Our study further demonstrates that the complex frequencies of LP events are useful to investigate internal physical processes and to monitor hydrothermal and magmatic systems beneath volcanoes.
243
Acknowledgments We thank Yoshiaki Ida, Jun Oikawa, and Keigo Yamamoto for sharing the data recorded by the Earthquake Research Institute of the University of Tokyo. We also thank Kenji Nogami for the useful discussion. We gratefully appreciate valuable comments from Phillip Dawson. This work was supported by the Japanese Ministry of Education, Culture, Sorts, Science and Technology under grant 14740265 to M.N.
References Aoyama, H., Takeo, M., 2001. Wave properties and focal mechanism of N-type earthquakes at Asama volcano. J. Volcanol. Geotherm. Res. 105, 163 – 182. Arciniega-Ceballos, A., Chouet, B.A., Dawson, P., 1999. Very longperiod signals associated with vulcanian explosions at Popocatepetl volcano, Mexico. Geophys. Res. Lett. 26, 3013 – 3016. Arciniega-Ceballos, A., Valdes-Gonzalez, C., Dawson, P., 2000. Temporal and spectral characteristics of seismicity observed at Popocatepetl volcano, central Mexico. J. Volcanol. Geotherm. Res. 102, 207 – 216. Chouet, B., 1986. Dynamics of a fluid-driven crack in three dimensions by the finite difference method. J. Geophys. Res. 91, 13967 – 13992. Chouet, B.A., 1996a. Long-period volcano seismicity: its source and use in eruption forecasting. Nature 380, 309 – 316. Chouet, B., 1996b. New methods and future trends in seismological volcano monitoring. In: Scarpa, R., Tilling, R. (Eds.), Monitoring and Mitigation of Volcano Hazards. Springer-Verlag, New York, pp. 23 – 97. Chouet, B., Saccorotti, G., Dawson, P., Martini, M., Scarpa, R., De Luca, G., Milana, G., Cattaneo, M., 1999. Broadband measurements of the sources of explosions at Stromboli Volcano, Italy. Geophys. Res. Lett. 26, 1937 – 1940. Chouet, B., Dawson, P., Ohminato, T., Martini, M., Saccorotti, G., Giudicepietro, F., De Luca, G., Milana, G., Scarpa, R., 2003. Source mechanisms of explosions at Stromboli Volcano, Italy, determined from moment-tensor inversions of very-long-period data. J. Geophys. Res. 108, 2019. doi:10.1029/2002JB001919. Cooley, J.W., Tukey, J.W., 1965. An algorithm for machine calculation of complex Fourier series. Math. Comput. 19, 297 – 301. Dawson, P.B., Dietel, C., Chouet, B.A., Honma, K., Ohminato, T., Okubo, P., 1998. A digitally telemeterd broadband seismic network at Kilauea Volcano, Hawaii. Open-File Rep. (U.S. Geol. Surv.) 98–108, 1 – 121. Fujita, E., Ida, Y., Oikawa, J., 1995. Eigen oscillation of a fluid sphere and source mechanism of harmonic volcanic tremor. J. Volcanol. Geotherm. Res. 69, 365 – 378. Gil Cruz, F., Chouet, B.A., 1997. Long-period events, the most characteristic seismicity accompanying the emplacement and extrusion of a lava dome in Galeras Volcano, Colombia, in 1991. J. Volcanol. Geotherm. Res. 77, 121 – 158.
244
M. Nakano, H. Kumagai / Journal of Volcanology and Geothermal Research 147 (2005) 233–244
Hamada, N., Jingu, H., Ikumoto, K., 1976. On the volcanic earthquake with slowly decaying coda wave. Bull. Volcanol. Soc. Jpn. 21, 167 – 183 (in Japanese with English abstract). Hirabayashi, J., 1999. Formation of volcanic fluid reservoir and volcanic activity. J. Balneol. Soc. Jpn. 49, 99 – 105. Ida, Y., Osada, N., Sawada, M., Koyama, E., Kagiyama, T., 1989. Seismological study based on recently installed permanent stations and a small eruptive event on January 6, 1989 at Kusatsu-Shirane volcano. Bull. Earthq. Res. Inst. Univ. Tokyo 64, 325 – 345 (in Japanese with English abstract). Japan Meteorological Agency (JMA), 1995. Report of coordinating committee for prediction of volcanic eruption. Rep. 61, Tokyo, 97. Kawakatsu, H., Kaneshima, S., Matsubayashi, H., Ohminato, T., Sudo, Y., Tsutsui, T., Uhira, K., Yamasato, H., Ito, H., Legrand, D., 2000. Aso94: Aso seismic observation with broadband instruments. J. Volcanol. Geotherm. Res. 101, 129 – 154. Kumagai, H., Chouet, B.A., 1999. The complex frequencies of long-period seismic events as probes of fluid composition beneath volcanoes. Geophys. J. Int. 138, F7 – F12. Kumagai, H., Chouet, B.A., 2000. Acoustic properties of a crack containing magmatic or hydrothermal fluids. J. Geophys. Res. 105, 25493 – 25512. Kumagai, H., Chouet, B.A., 2001. The dependence of acoustic properties of a crack on the resonance mode and geometry. Geophys. Res. Lett. 28, 3325 – 3328. Kumagai, H., Ohminato, T., Nakano, M., Ooi, M., Kubo, A., Inoue, H., Oikawa, J., 2001. Very-long-period seismic signals and caldera formation at Miyake Island Japan. Science 293, 687 – 690. Kumagai, H., Chouet, B.A., Nakano, M., 2002a. Temporal evolution of a hydrothermal system in Kusatsu-Shirane Volcano, Japan, inferred from the complex frequencies of long-period events. J. Geophys. Res. 107, 2236. doi:10.1029/2001JB000653. Kumagai, H., Chouet, B.A., Nakano, M., 2002b. Waveform inversion of oscillatory signatures in long-period events beneath volcanoes. J. Geophys. Res. 107, 2301. doi:10.1029/2001JB001704. Kumazawa, M., Imanishi, Y., Fukao, Y., Furumoto, M., Yamamoto, A., 1990. A theory of spectral analysis based on the characteristic property of a linear dynamic system. Geophys. J. Int. 101, 613 – 630. Lesage, P., Surono,, 1995. Seismic precursors of the February 10, 1990 eruption of Kelut volcano, Java. J. Volcanol. Geotherm. Res. 65, 135 – 146.
Molina, I., Kumagai, H., Yepes, H., 2004. Resonances of a volcanic conduit triggered by repetitive injections of an ash-laden gas. Geophys. Res. Lett. 31. doi:10.1029/2003GL018934. Nakano, M., Kumagai, H., Kumazawa, M., Yamaoka, K., Chouet, B.A., 1998. The excitation and characteristic frequency of the long-period volcanic event: an approach based on an inhomogeneous autoregressive model of a linear dynamic system. J. Geophys. Res. 103, 10031 – 10046. Nakano, M., Kumagai, H., Chouet, B.A., 2003a. Source mechanism of long-period events at Kusatsu-Shirane Volcano, Japan, inferred from waveform inversion of the effective excitation functions. J. Volcanol. Geotherm. Res. 122, 149 – 164. Nakano, M., Kumagai, H., Oikawa, J., 2003b. Retrieval of seismic data recorded on open-reel-type magnetic tapes (MT) by using existing devices. Tech. Res Rep.-Univ. Tokyo. Earthq. Res. Inst. 9, 1 – 7 (in Japanese with English abstract). Neuberg, J., Luckett, R., Ripepe, M., Braun, T., 1994. Highlights from a seismic broadband array on Stromboli volcano. Geophys. Res. Lett. 21, 749 – 752. Neuberg, J., Baptie, B., Luckett, R., Stewart, R., 1998. Results from the broad-band seismic network on Montserrat. Geophys. Res. Lett. 25, 3661 – 3664. Ohba, T., Hirabayashi, J., Nogami, K., 2000. D/H and 18O/16O ratios of water in the crater lake at Kusatsu-Shirane volcano Japan. J. Volcanol. Geotherm. Res. 97, 329 – 346. Ohminato, T., Chouet, B.A., Dawson, P., Kedar, S., 1998. Waveform inversion of very long period impulsive signals associated with magmatic injection beneath Kilauea Volcano, Hawaii. J. Geophys. Res. 103, 23839 – 23862. Ossaka, J., Ossaka, T., Hirabayashi, J., Oi, T., Ohba, T., Nogami, K., Kikawada, Y., Yamano, M., Yui, M., Fukuhara, H., 1997. Volcanic activity of Kusatsu-Shirane volcano, Gumma, and secular change in water quality of crater lake Yugama. Chikyukagaku (Geochemistry) 31, 119 – 128 (in Japanese with English abstract). Rowe, C.A., Aster, R.C., Kyle, P.R., Schlue, J.W., 1998. Broadband recording of Strombolian explosions and associated very-longperiod seismic signals on Mount Erebus volcano, Ross Island, Antarctica. Geophys. Res. Lett. 25, 2297 – 2300. Yamamoto, M., Kawakatsu, H., Yomogida, K., Koyama, J., 2002. Long-period (12 sec) volcanic tremor observed at Usu 2000 eruption: seismological detection of a deep magma plumbing system. Geophys. Res. Lett. 29. doi:10.1029/2001GL013996.