J. Phys. Earth, 45, 397-416, 1997 Quantitative Analysis of Pyroclastic Flows Using Infrasonic and Seismic Data at Unzen Volcano, Japan Hitoshi Yamasato Meteorological Research Institute, Japan Meteorological Agency, Tsukuba 305, Japan The process of the collapse of the dacitic lava dome and the development of pyroclastic flows at Unzen volcano, Japan, were studied using infrasonic, seismic and video records. Characteristic infrasonic and seismic signals were recorded corresponding to the collapse of lava blocks from the dome, the drop of blocks on the slope and the migration of pyroclastic flow on the mountain slope. Small infrasonic and seismic waves are excited when the lava dome starts to collapse. When the lava blocks fall onto the mountain slope and are fragmented, larger waves are excited. This suggests that the seismic waves are generated by the collision of pyroclastics on the mountain slope and that the infrasonic waves are excited by small fractures of the dome and the fragmentation of pyroclastics. Some of the infrasonic signals show an obvious Doppler effect, indicating that the pyroclastic flows emit infrasonic signals during their propagation. The location of dome collapse and the path of pyroclastic flows can be identified and traced by a network of low-frequency microphones. The migrating source of infrasonic signals and probably seismic signals is inferred to be located near the front of pyroclastic flows by comparison with video images. This suggests that the fragmentation of pyroclastics occurs mainly near the front of pyroclastic flows. The speed of pyroclastic flows is estimated as 10-30 m/s from the infrasonic records. The excitation of infrasonic and seismic signals is affected by the topography of the mountain slope. The infrasonic energy is almost the same order as the seismic energy but the ratio of infrasonic to seismic energies increases for larger and more mobile pyroclastic flows. This means that the development of pyroclastic flows is controlled not only by the volume of lava and gravitational force, but also by the explosivity related to the pore gases in the lava. Introduction Unzen volcano is an active dacitic volcano that is located in Beppu-Shimabara graben, Kyushu, Japan. It started eruption in 1990 after 198 years of dormancy. A dacitic lava dome emerged in May of 1991 and its collapse resulted in successive gen- erations of pyroclastic flows. A detailed description of the sequence of the present volcanic activity of Unzen Received April 17, 1997; Accepted December 24, 1997 397 1 398 H. Yamasato was given by Ohta et al. (1992). A phreatic eruption occurred at the summit craters (Jigoku-ato crater and Tsukumo-jima crater) of Fugendake, which is the main peak of Unzen, on 17 November 1990. On 20 May 1991, a dacitic lava dome emerged at Jigoku-ato crater. It grew to collapse and resulted in pyroclastic flows from 24 May. Subsequently, pyroclastic flows frequently occurred until 1995. In particular, the one on 3 June 1991, which traveled 4.6 km, killed 43 people. Many studies have been carried out on pyroclastic flows at Unzen. Several visual observations support the view that most of the dome collapses at Unzen are triggered by gravitational instability of the dome (e.g., Fukui et al., 1991; Ui et al., 1993); that is, pyroclastic flows at Unzen volcano are mostly of the Merapi type. Although pyro- clastic flow is one of the most conspicuous phenomena associated with eruptions at many volcanoes in the world, the study of pyroclastic flow by means of geophysical methods has been scarce. At Unzen volcano, a wealth of geophysical data was acquired from valuable observations. For example, Uhira et al. (1994) studied the seismic signals associated with dome collapse and discussed the force system at the origin of the pyroclastic flows. According to their results, the seismic waves are excited by the collision of lava blocks on the mountain slope, in agreement with visual observations. Yamasato et al. (1993) studied not only seismic signals but also infrasonic signals from pyroclastic flows and obtained some characteristics of these signals. The Japan Meteorological Agency (JMA) operates a low-frequency microphone network and a dense seismic network. Infrasonic observation with low-frequency microphones has been made by some geophysicists at some volcanoes, and infrasonic waves excited by explosive eruptions have been studied (e.g., Tahira, 1981; Iguchi and Ishihara, 1990). Rarely, however, have observations of infrasonic waves by a low- frequency microphone network been associated with pyroclastic flows. Here, the in- frasonic and seismic data obtained by JMA are analyzed more quantitatively and the mechanism of pyroclastic flows is investigated. 2. Data The locality of Unzen volcano and the distribution of observation stations used here are shown in Fig. 1. Short-period seismographs with a natural frequency of 1 Hz are operated at these seismic stations. The Meteorological Research Institute of JMA (MRI) installed a long-period seismograph with a natural frequency of 0.1 Hz (Den et al., 1984) at the observatory (Unzendake Weather Station, O on Fig. 1) in April 1991. Infrasonic observation with four low-frequency microphones started in the spring of 1992. Three of the low-frequency microphones used in this study have a constant pressure response in the frequency band between 0.1 and 10 Hz, the other at station K2 since October 1992, has a constant pressure response between 1 and 40 Hz. For waveform analysis, these records were transformed by filtering so that the records are comparable to each other. The filtering coefficients were determined using the relative amplitude and phase responses between the two types of microphones, which were obtained from a comparative observation. The seismic and infrasonic signals are telemetered to the observatory and are J. Phys. Earth Quantitative Analysis of Pyroclastic Flows 399 Fig. 1. Location of Unzen volcano and observation stations that produced data analyzed in this paper. Closed circles indicate seismic stations; triangles, infrasonic stations; lozenges, video stations; and the square, the observatory of JMA (Unzendake Weather Station). The hatched area shows the pyroclastic flow deposit. recorded continuously on digital tape recorders with a sampling frequency of 50 Hz. Visual data were collected by a time-lapse video recorder at station T1 and by temporary observation with a portable 8 mm video camera at station T2. At the be- ginning of the recordings, the time codes of the video recorders were synchronized at the observatory to the time of the clock with an accuracy of 0.5 s. An example of the record of a low-frequency microphone is shown with the corresponding seismogram in Fig. 2. On the seismic record, two types of seismic signal appear; one is that associated with a pyroclastic flow and the other is a low-frequency earthquake. The same events are recorded in the infrasonic signal with amplitudes of 1 to 2 Pa. The infrasonic signal due to the pyroclastic flow has a long duration, and that from the low-frequency earthquake is impulsive. The impulsive infrasonic signal from the low frequency earthquake is investigated in the another paper (Yamasato, 1998). The infrasonic record is often disturbed by wind noise, as appears in the lower part of the record in Fig. 2. Vol. 45, No. 6, 1997 400 H. Yamasato Fig. 2. Examples of the (a) seismic and (b) infrasonic records obtained by a vertical seismograph and a low-frequency microphone at station E1. PF and LF indicate records of a pyroclastic flow and a low-frequency earthquake, respectively. In this paper, the data associated with pyroclastic flows during the period from June 1992 to November 1993 are analyzed. During this period, dome collapses suc- cessively occurred and pyroclastic flows ran in three directions as shown in Fig. 1. 3. Seismic and Infrasonic Signals Associated with Dome Collapse If we examine the waveforms of the initial phases of seismic signals related to pyroclastic flows, we may get more detailed information on the process of generation of pyroclastic flows. Yamasato et al. (1993) investigated video and seismological data from some dome collapses at Unzen and clarified the time sequence of the dome collapse and the excitation of the corresponding seismic waves. Figure 3 shows an example of the time sequence of a dome collapse, which was obtained from video data obtained at station T2, and the corresponding seismic signal. In this case, an unstable tip of the dome collapsed and generated a pyroclastic flow. From photographs, the volume of the collapsed lava was estimated as 5.7•~104m3. The flow distance from the dome collapse was 2.5 km as estimated from the video record. By combining the seismic data with the visual data, the following sequence was identified. J. Phys. Earth Quantitative Analysis of Pyroclastic Flows 401 Fig. 3. Illustrations of the dome collapse sequence (left) and the corresponding seismic signals (right). A sketch of the dome from station T2 is shown based on a video record taken by a portable 8 mm video recorder. SPs are seismograms obtained by short-period seismographs; LP is by a long-period seismograph. The records are arranged according to the horizontal distance from the source. Seismic records for the vertical component are shown here. A) Seismic waves with small amplitude and predominant frequency of 2-3 Hz were excited almost simultaneously with the collapse of the lava dome. B) The amplitude became larger a few seconds later as lava blocks fell onto the slope. The seismic waves with large amplitude contained a low-frequency component (about 0.5 Hz). C) The fragmented pyroclastics started to flow generating high-frequency seismic waves (more than 2 Hz). In Fig. 4, an example of a clear infrasonic signal from a dome collapse is shown.