Based on macro seismic and instrumental observation data covering a period of over hundred years, the characteristics of seismic activity around Miyakejima, Kozushima, and Niijima, the northern Izu islands, were investigated. Seismic activity is distributed along the Zenisu ridge and the Nishi-Shitito ridge to the west and along Niijima, Miyakejima, Mikurajima and Hachijojima islands to the east. Most of the seismic activity in the region is of the swarm type and is often induced by a large earthquake or magmatic activity at Miyakejima and other volcanoes. Comparison of macro-seismic data of large earthquakes in the late 1800s and the early 1900s with a seismic intensity map of recent earthquakes suggested that the maximum size of an earthquake in the region is around a magnitude of 6.5 on the JMA magnitude scale. Linear arrangements of seismicity in the NE-SW direction, which is parallel with the Zenisu ridge, and in the NW-SE direction, which is almost normal to the former and parallel to the direction of the plate motion of the Philippine Sea plate, are often recognized. While the strike slip-type earthquake with a NE-SW tension axis is predominant in the region along Niijima, Miyakejima, Mikurajima, and Hachijojima islands, a strike slip with E-W tension axis is predominant in the region along the Zenisu ridge and the Nishi-Shitito ridge. A complex tectonic setting of the region, back arc spreading, collision of the Philippine Sea plate to the Eurasian plate around the Izu peninsula, deformation of the plate due to the subduction of slab from the Suruga and the Sagami trough, must shape the characteristics of seismic activity in the region such as the upper limit of earthquake scale, swarm-type activity, close relation with volcanic activity, and regional stress pattern.
From June 26, 2000, an intensive earthquake swarm started under Miyake-jima Island, 180 km south off Japan. This swarm was closely related to the eruption of Miyake-jima Island, probably dominated by underground magmatic activity. The swarm spread toward the northwestern ocean region from Miyake-jima Island, in which a huge number of earthquakes (over 100, 000) including five large events of M>6.0 were detected over about two months. This earthquake swarm was the most active since we started seismic observations in the 1970's. Although there are some telemetered observation stations on the Izu volcanic islands, no offshore instruments were operated in the area of this earthquake swarm. To understand both the spatial and temporal changes of this activity, we conducted a series of ocean bottom seismometer observations. According to the variation in the seismic activity with time, we changed the array configuration of OBSs six times. Furthermore, real-time seismic observations were undertaken using a buoy-telemetering OBS system. Combining the OBS data with those of the island stations, very precise earthquake locations were determined. The epicenter distribution obtained strongly indicates a northwest-southeastern lineament. The vertical cross-section of the events shows two characteristic trends. Deeper (7- 13km) events are forming a very thin (2-km thick) plane, while shallower ones (< 7 km) show a much thicker distribution. These distribution patterns will provide important constraints on the physical mechanism for understaning magma migration.
Miyakejima Volcano is located 200 km south of Tokyo, Japan, and is one of the active volcanoes situated on the Izu-Mariana Arc. The main cone of Miyakejima has two nested calderas : outer Kuwanokitaira Caldera, 4 km in diameter, and inner Hatchodaira Caldera, 1.8 km by 1.6 km across. The central cone, Oyama, grew in the Hatchodaira Caldera. A short recurrence time of about 22 year-period and continuous crustal inflation in the recent years suggested that the volcano was proceeding to the next eruption. Activity began on June, 26, 2000, resulting in subsidence to form and the forming of a new caldera 1.6 km in diameter. The new Hatchodaira Caldera almost overlaps the site and size of the former Hatchodaira Caldera. This paper presents the volcano-history of Miyakejima during the last 10000 years. Based on the eruption style, together with whole-rock bulk chemistry, the development history during the last 10000 years is divided into four stages. These are 1) the Ofunato Stage of 10000-4000 y.B.P., 2) the Tsubota Stage of 4000-2500 y.B.P., 3) the Oyama stage, from 2500y.B.P. to the early 15th century, and 4) the Shinmio stage, 1469 AD to the present. Since 1085 AD, at least 14 eruptions are documented. The last four eruptions occurred in 1874, 1940, 1962, and 1983. The Ofunato stage is characterized by porphyritic basalts. These lavas and pyroclastics contributed the growth of the main cone and filling up of the Kuwanokitaira Caldera. The Tsubota Stage, reopened after a 3000-year repose, is distinguished by andesitic products from lateral and central eruptions. The Oyama Stage began with the formation of the Hatchodaira Caldera, which resulted from the discharge of ca. 0.37km3 scoria, explosion breccia, and accretionary lapilli. Subsequent products from central and lateral eruptions filled the caldera. It is noteworthy that phreato-magmatic eruptions from the central vent prevail over dry magmatic eruptions. Overflows of lavas from the rim of the Hatchodaira Caldera occurred in 9th century. At the Shinmio stage, 12 documented eruptions, without exception, took place from lateral fissures, with some accompanied by central eruption. The variation of Mg# (=Mg/ (Mg+Fe) × 100) versus erupted age show an abrupt increase of the ratio 2500, 1300, and 500 y.B.P. with a mild decrease. This pattern suggests that relatively undifferentiated magma was supplied to the magma plumbling system underneath the Miyakejima Volcano, which was slowly differentiating.
The 2000 eruption of Miyakejima volcano started with a submarine eruption of basaltic andesite on the morning of June 27, which occurred following earthquake swarms during the previous night. The main phase of the summit eruption began, being associated by a sudden subsidence of the summit area on July 8. Continuous collapsing of the summit area that had continued until midAugust, resulted in the formation of a caldera with the volume of about 0.6 km3. Phreatic (or phreatomagmatic) eruptions took places during the growth of the caldera, although the total volume of eruptives was about 11 million m3. which is smaller by one magnitude than the caldera volume. Eruptives are enriched with hydrothermally altered materials such as smectite and kaolinite. The manner of the first collapse suggests the existence of a large open space under the summit just before the subsidence. Judging from geophysical observation results, the open space may have ascended in the manner of stoping. Successive formation of open spaces at deeper levels is likely to have caused the continuous collapse of the summit area. These open spaces may have been generated by magma's migration from under Miyakejima to the west. The migration is considered to have continued by August 18. It is likely that an inflow of underground water to the open spaces generated a hydrothermal system, where the open spaces acted as a sort of pressure cooker that built up overpressure of eruptions. The hydrothermal system was broken by the largest eruption on August 18, and the eruption column rose about 15 km above the summit. A boiling-over type of eruption occurred on August 29, whereby sufficient overpressure of steam was not built up, resulting in the generation of low-temperature ash cloud surges moving very slowly.
The National Research Institute for Earth Science and Disaster Prevention had been conducting continuous crustal tilt observations at five stations on Miyakejima, prior to the volcanic activity in the year 2000. During the caldera-forming stage, extraordinary step-like tilt signals were detected with the following time sequence. 1) The initial step signal was detected coincidentally at the exact time of the explosive summit eruption on July 8, 18h41m. 2) The 2nd step signal was detected on July 9, roughly 28 hours following the initial one. Since then, step signals have appeared intermittently, at frequencies of 1-3 steps within 1-2 days. 3) The last step signal was detected at 18h09m of August 18, during the largest summit eruption. In total, 46 step signals were detected and all of the step signals occurred within a very short interval of less than two minutes. Tilting directions of the initial and the last step signals were in the sense of ground-up toward the summit. These were then followed by a reverse tilt. Similarly, tilting directions of the remaining 44 step signals were mostly in the sense of ground-up toward the southern part of the island. Again, they were followed by a reverse tilt. From these observational results we conjectured that the step signals were generated by an explosive source of volumetric expansion. Tilt movements after the steps seem to reflect a relaxation process. Since the initial step signal was synchronized with the eruption, which was considered to be a phreatic explosion, this must be the cause of the step. Although the extent of tilting of the last step was very small compared to the first one, their tilting patterns resembled each other. Accordingly, a similar small explosion may have occurred at the same time when the last step signal was detected. As for the remaining 44 step signals, tilt vectors at the northern three stations were aligned in their own directions toward the southern part of the island, while those at the southeast and southwest stations were rather widely spread. These features suggest that the pressure source is in the southern part of the island. A dike, which was considered to have intruded into this part of the island must be related to them.
The 2000 Miyakejima volcano activities started on Jun. 26 after 17 years of quiescence. The NIED Miyakejima volcano observational network, mainly equipped with seismometers and tiltmeters, successfully and continuously detected the migrations of magma. The volcanic activities consist of four stages : the dike intrusion (Stage 1 : Jun. 26-27), the subsidence of volcanic body (Stage 2 : Jun. 27-Jul. 8), tilt-steps and successive collapses of the summit crater (Stage 3 : Jul. 8-Aug. 18), and the subsidence of volcanic body and episodic eruptions (Stage 4 : Aug. 18-). This paper summarizes characteristic seismic phenomena that occurred at each stage; i.e., VT (volcanotectonic) earthquakes, LP (long-period) earthquakes, and volcanic tremor. Significant phenomena are the 50-second pulse waves associated with the tilt-steps at Stage 3. They reflect subsurface magmatic processes and their temporal and spatial variations.
A sudden collapse of the summit of Miyake-jima occurred on July 8, 2000, together with intermittent eruptions. This collapse generated long-period seismic waves with a dominant period of about 10s. Following this event, very long-period seismic pulses (VLP pulses) with a duration of about 50s were observed a few times a day until they ceased at the largest summit eruption on August 18. We analyzed these seismic pulses using waveform data recorded at several domestic stations for broadband seismographs and strong motion seismometers on Miyake-jima. The July 8 event is well characterized by a single-force directed initially upward and later downward during 12 sec. The single-force is interpreted as an abrupt collapse of massive rock. The total mass is estimated to be about 5 × 1010 kg with fall of about 300 m. On the other hand, VLP pulses are modeled by moment-tensors with an isotropic component. They are located about 1 km southwest from the summit and 2 to 3 km deep. All three principal values are positive. The largest one is horizontal and the smallest one is near vertical. The total volume change due to 39 VLP pulses is 2.6× 108m3, amounting to nearly one half of the total volume of the summit collapse. Based on theresults, we propose a buried geyser model. A large reservoir of hot water was formed just after the summit collapse on July 8. The ground water poured into the reservoir, being rapidly heated by hot rock underneath, and evaporated to form a highly pressurized steam, which pushed a lower conduit piston into the magma reservoir to generate VLP pulses. Non-isotropic expansion of the VLP pulses may be ascribed to the shape of the magma reservoir.
Employing both absolute and relative gravimeters, we carried out a hybrid microgravity survey at Miyakejima volcano, Japan. We detected significant gravity changes, e.g., exceeding 1000 At gal, during the course of our measurements since June 1998. The spatio-temporal gravity changes provide us with strong evidence of preexisting vacant space beneath the island before the summit collapse of the volcano. Also, our data are totally in conflict with the so-called magma drainback model, suggesting that the magma flowed out laterally.
Electric and magnetic field observations have been extensively carried out since 1995. A precursory magnetic anomaly was detected in July 1996, which was ascribed to thermal demagnetization at a depth of several hundreds of meters beneath the southern periphery of the summit Hatcho-taira caldera. Magnetic data revealed that the large depression at the summit associated with the steam explosion on July 8, 2000 had been completed within four minutes. Since the beginning of July, anomalous magnetic changes were observed at several magnetometer sites along the central N-S line of Miyake-jima volcano, which indicated the rise of a demagnetized area from depth to the summit. On July 4, a few days before the steam explosion, an area survey of SP in the summit caldera was conducted, discovering an extremely negative zone around the forthcoming depression, which suggested the intense absorption of ground water. Tilt-step events; i.e., abrupt uplifts around the summit area, were accompanied by electric field variations, which were very similar to the velocity waveform of the ground motion, as well as magnetic variations with step-like changes. An electric field can be interpreted as being due to electric currents generated by the forced injection of steam and/or water from the pressure source (electrokinetic phenomena). Magnetic changes are attributed to the piezomagnetic effect of rocks due to increased stresses. The geomagnetic total intensity showed large variations after the July 8 eruption, the typical feature of which was positive at the east and west sides and negative along the central north-south line of the volcano. They are ascribed to 1) the loss of magnetic mass from the summit and 2) the thermal demagnetization at depth. After the August 18 eruption, which was the largest, the steep changes in total intensity became flat, which suggested that the temperature rise at depth had weakened. At the time of the August 18 eruption, a large increase in self-potential was observed around the southwestern foot of the central cone Oyama : This implies that a definite change occurred in the hydrothermal system of the volcano.
Basaltic volcanoes above oceanic crust or island-arc crusts develop calderas. Upon the formation of this caldera, the collapsed volume was generally far larger than erupted volume. Caldera width, its depth, and caldera horizontal width (CR) / volcano size (VR) depend on the physical properties of an oceanic crust. The CR/VR ratio decreases away from the ridge. At Fernandina volcano, Galapagos, the elongated caldera of 3.5 km × 2.5 km was formed in 1968 during a phase dominated by circumferential fissure eruptions after a phase dominated by radial fissure eruptions. At Kilauea volcano, Hawaii, during a phase dominated by central eruptions, several drain backs of lava lake ccurred from 1800, and, finally, a caldera of 1 km in diameter collapsed in 1924. Caldera collapse seems to be inevitable because accumulated crystals and solidified magma under a volcano increases gravitational instability. According to the gravitational collapse model proposed in this paper, it is difficult to determine when, how wide, and how deep a caldera collapse will occur. The magma plumbing system expands horizontally and vertically during long-term growth. Caldera collapse should contribute to vertical growth. Horizontal growth and vertical growth are governed by physical properties of the crust beneath the volcano; the former process is dominant in Hawaii, and the latter in Galapagos. In the case of Miyakejima volcano, the caldera collapse may be triggered by dike intrusions into a region with a low probability of intrusions or by an increase in the magma supply beneath the magma chamber. At the Miyakejima eruption in 2000, the caldera of 1.5 km in diameter formed in the shallow crust; ductile mass or dense magma with crystal mush may have moved downward or northwestward in the deep crust.
To understand the eruptive mechanism of the 2000 Miyakejima volcanic activity, we conducted intensive geological, petrographic, and mineralogical studies on the pyroclastics of the August 18 eruption. Volcanic ashes, which were rich in accretionary lapilli, covered most of the islands. Cauliflower-shaped bombs and lapilli were ejected along with accidental lava blocks. Black-colored angular scoriaceous particles with abundant vesicles 10 -100 μm in diameter are found among ashes, comprising about 40 wt. % of total constituents. These bombs, lapilli, and black ashes have identical bulk chemical compositions and constituent mineral compositions, suggesting a common origin. Existence of oxidized ashes and accretionary lapilli attached to a large flattened bomb and chemicallyreacted anhydrite particles trapped in the voids of bombs suggest that bombs were still hot and ductile when they were emplaced on the ground. We, therefore, conclude that the August 18 eruption was a phreatomagmatic eruption and cauliflower-shaped bombs and black ashes were essential magmatic materials. Significant SO2 emissions from the volcano after August 18 also suggest convective upwelling of magmas to a shallower level beneath the volcanic edifice. We propose a magma-ascending model in which vesiculating magmas continuously ascend through the wall of subsided piston-like blocks.
To understand degassing processes during the 2000 Miyakejima volcanic activity, we applied the following two methods : 1) repeated analyses of adhered water-soluble gas component such as SO4and Cl ions on ashes produced at eruptions from July 8 and mid-September, and 2) SO2 flux measurements by COSPEC since August 26. The repetitive analyses of soluble component show remarkable changrd in Cl/S. Until August 18, the adhered SO4 concentrations are quite high and Cl concentrations are always low with Cl/S of 0.01-0.05, indicating that a certain mass of groundwater existed in aquifers beneath the summit crater and Cl component selectively dissolved in the groundwater. Since August 29, chlorine concentrations became greater and Cl/S ratios were determined as 0.1-0.14 on August 29 and 0.5-1.5 in September, which coincided with the strong volcanic gas emissions which started in mid-August. Groundwater boiling and establishment of gas conduit are likely to occur to prevent HCl from being absorbed in the groundwater. Sulfur dioxide emission rate has been monitored since August 26. The SO2flux increased in midSeptember from thousands to tens of thousands tons a day. The average SO2 flux after midSeptember to the present is 48 ktons/day. The highest flux was observed on December 7 to be 230 ktons/day. The mass rate of the magma degassing is estimated as 20 Mtons/day and the total volume of the degassed magma is calculated to be 1 km3 so far. The continuous magma degassing without eruptions occurs at a shallow environment by the convective transport of magma from a chamber to a magma head through conduits. A huge degassing rate is likely to be due to a large surface at the magma head, which would be made by a piston-cylinder type of collapse of the volcanic body.