Splitting of the Philippine Sea Plate and a magma chamber beneath Mt. Fuji



[1] Mt. Fuji is located in a tectonically unique area, but various aspects of the structure have not been fully explained. Here we show the results from a magnetotelluric survey across Mt. Fuji along a 70 km observation line. The profile shows that a conductive body is located between two resistive bodies at depths greater than 15 km. Low frequency earthquakes occur above the conductor. We interpret these results in a model where beneath Mt. Fuji, the subducting Philippine Sea Plate is split into two parts, and a magma chamber is located in the gap. Due to this unique structure, Mt. Fuji may be able to sustain a high basalt magma flux throughout its entire history.

1. Introduction

[2] Mt. Fuji volcano (called Fuji-san) is the highest and largest stratovolcano in Japan, and its eruption rate (the order of 1000 km3/100,000years) is larger than most other island-arc volcanoes by one order of magnitude [Fujii, 2001]. The erupted products are almost all basaltic [Tsuya, 1968] for its entire history of about 100,000 years, despite the fact that Mt. Fuji is an island-arc type volcano. It has been suspected that these unique features may be due to the special tectonic setting of the volcano.

[3] The tectonics of the region around Mt. Fuji are very complex with three converging Plates, the Eurasia (EUR), the North American (NAM), and the Philippine Sea (PHS) Plates, as shown in Figure 1a. Beneath Mt. Fuji, the PHS Plate is subducting along the Sagami and Suruga troughs toward the northeast, and the Pacific Plate is subducting along the Japan Trench toward the west. This region also can be regarded as a crossover area between the volcanic front and plate boundary regions. Furthermore, this region is a zone of collision where the Izu block has been colliding with the Honshu block, pushing north-northwestward since the early Quaternary (2 Ma to present) [Matsuda, 1978]. Many Quaternary volcanoes, such as Hakone caldera (0.5 Ma to present), Ashitaka (4 Ma-0.1 Ma), Izu polygenetic volcanoes (1 Ma-0.2 Ma), Higashi-Izu monogenetic volcano (0.15 Ma to present), and Izu-Oshima (0.04 Ma to present), exist in this region. Mt. Fuji is also formed on the foundation of an old Quaternary volcano, Komitake (0.5 Ma-0.2 Ma). Naturally, this region is characterized by a highly deformed zone with many historic earthquakes, but a remarkable lack of seismicity is recognized around Mt. Fuji, so many previous studies have disagreed on the exact location of the plate boundary and the detailed configuration of the PHS Plate [e.g., Sugimura, 1972; Nakamura and Shimazaki, 1981; Seno, 1986; Iidaka et al., 1990; Ishida, 1992]. Whether or not the PHS Plate exists in this low seismicity area is also not clear. At depths of 15 km in an area beneath Mt. Fuji, low frequency earthquakes are regularly seen [Ukawa, 2001]. Low frequency earthquake activity became vigorous in October and November 2000 and April and May 2001, but no other associated volcanic activity was observed at that time.

Figure 1.

(a) Plate boundaries around Japan. Thick solid lines indicate plate boundaries. EUR, NAM, PAC, PHS represents the Eurasian Plate, the Northern American Plate, the Pacific Plate, and the Philippine Sea Plate, respectively. Shaded thick line and blue box indicate the volcanic front and the region shown in (b), respectively. (b) Topographic map around Mt. Fuji with observation sites (red triangles) of the magnetotelluric (MT) survey. Topographic contour intervals are 500 m. Quaternary volcano are shown with characters: FJI is Mt. Fuji, ATK is Ashitaka volcano, HKN is Hakone caldera, HIZ is east Izu monogenetic volcano field, OSM is Izu-Oshima volcano and IZU is area of polygenetic volcanoes on the Izu peninsula.

2. Magnetotelluric Measurement and 2-D Modeling

[4] In order to clarify the subsurface structure, and understand the relation between tectonic processes and the unique features of Mt. Fuji, we carried out wide-band magnetotelluric (MT) soundings on a survey line trending northeast to southwest (Figure 1b). This survey is one of the first MT studies on Mt. Fuji. The data were acquired during 9 September to 19 September 2002 and 8 May to 22 May 2003. The typical recording duration for one site was 7 to 10 days. Natural electric and magnetic fields in the period range of 0.003 to 2,000s were measured at the surface using Phoenix MTU5 and MTU2E systems. Because leakage electric current from railways seriously contaminated the data, we used a remote reference processing [Gamble et al., 1978] to calculate the noise-free impedance with geomagnetic data which were recorded in northern Hokkaido (1000 km north of Mt. Fuji) and the Tohoku region (600 km north of Mt. Fuji).

[5] Induction vectors show a complicated pattern, and point to the coastline in the period range beyond 100s. However, parasitic cones of Mt. Fuji which are mainly aligned along a northwest to southeast direction [Tsuya, 1968], the pattern of gravity anomalies [Komazawa, 1987], and seismicity in this region [Tsuruoka, 1997], all suggest that there is a NW-SE trend of the physical structures in the region. The principal axis [Swift, 1967] of the impedance tensor and relatively small diagonal elements of impedances (impedance skewness are 0.1 to 0.4) also suggest NW-SE trending 2D structures. Therefore in this study, we determined the resistivity structure by inverting the data using a two dimensional code [Ogawa and Uchida, 1996] assuming a regional strike of N40W (which is parallel to the structure).

[6] The inversion was carried out in two steps. First, a 100 ohm-m uniform 2D model, including the topography, which extends 1000 km horizontally and vertically, was used as a starting model. The inversion was carried out to fit the TM data. Next, the inversion was started using the model obtained in the first step, and included TE mode phase data in addition to the TM mode data. To avoid the effects of 3D subsurface structure and topography as much as possible, we did not used the TE mode apparent resistivity. At the site NE of the profile, the topography significantly varies in a NW-SE direction (electric current flows on TE mode direction). The vertical component of the geomagnetic data is not used on the inversion, and the ocean is not included in the model. We also carried out the inversion assuming regional strikes of N20W and N60W for the structure, and the results showed similar resisitivity models. The inversion, which constrained a strong conductor 40 km outside of the profile, does not affect the structure beneath the profile significantly.Auxiliary Material Although the data qualities are not of the highest quality at all of the observation sites, the data are found to be sufficiently sensitive to the main features of the model from numerical tests. Figure 2 shows comparisons between observed data and calculated data.

Figure 2.

Pseudo-sections for the observed and calculated data; Upper and Lower rows are the observed and calculated pseudo-sections, respectively. From left to right, plotted are apparent resistivity for TM mode and phase for TM and TE modes. Calculated apparent resistivity includes static shifts.

3. Interpretation of the Resistivity Structure

[7] The best fit resistivity model shown in Figure 3 exhibits two significant resistive bodies over 2000 Ωm, located at −40 km to −20 km (R2) and at 0 km to 40 km (R1) in Figure 3. Although at −20 km to 20 km the seismicity is very low, as stated in the previous section, the upper surfaces of the two resistive bodies correspond to the zones of tectonic earthquakes. In light of previous magnetotelluric studies which showed that the subducting plate (slab) is highly resistive [e.g., Wannamaker et al., 1989; Satoh et al., 2001], these two resistive bodies are interpreted as parts of the PHS Plate. At 0 to 20 km R1 is interpreted as a low seismicity part of the subducting PHS Plate. At 10 to 20 km above R1 there is a remarkable conductor. Since this conductor locates beneath the Fujigawa fault, which is a reverse fault on the north extension of the Suruga trough, it may image the fluids trapped in the deep fault. This conductor is suspected to affect the slab seismicity, but its mechanism is not known.

Figure 3.

Best fit resistivity model obtained by 2-D inversion modeling of MT data. 2D regional strike direction (parallel to the structure) is assumed to be N40W. Inverted triangles indicate the measurement sites. Seismicity for 5 years (1998–2002) in a swath of 20 km wide is also plotted. Earthquake magnitude is shown on right. Solid circles indicate tectonic earthquakes [Tsuruoka, 1997]. Stars show locations of 76 low-frequency earthquakes which were recorded with good signal to noise ratios [Nakamichi et al., 2003a]. R1, R2, and C1 refer to features discussed in the text. Note that the Philippine Sea Plate is subducting in a northwest direction.

[8] At depths deeper than 20 km in Figure 2, there is a large conductor (C1) in the region above which low frequency earthquakes [Nakamichi et al., 2003a] are occurring. Although the mechanism of low-frequency earthquakes is not fully understood, it is thought that they are related to magma activity [e.g., Ukawa and Ohtake, 1987; Nakamichi et al., 2003b]. We interpret this conductor as a magma chamber below Mt. Fuji. Seismic tomography which suggests that a large low velocity anomaly exists beneath Mt. Fuji at depths greater than 20 km [Lee and Ukawa, 1992] supports this interpretation.

[9] Combining the above-mentioned results and interpretation, we propose a tectonic model that the PHS Plate is split into two parts and magma ascends through the gap to form the large magma chamber. The low-frequency earthquakes might be caused by magma ascending from the magma chamber.

4. Discussion and Conclusion

[10] On the speculation that the PHS Plate is split in the low seismicity region, which is directly beneath the summit area, Takahashi [2000] explained the large eruption rates and the basaltic composition of Mt. Fuji by the subsurface stress field. Although our results differ from his speculation in detail, our conclusions are rather similar. The splitting of the PHS Plate causes a tensile stress field in the lower and middle crust beneath Mt. Fuji allowing mantle-derived basaltic magma to ascend easily. This explanation is similar to the way basaltic magma rises at mid-ocean ridges. In fact, it has been pointed out that the composition of Mt. Fuji is similar to that of mid-ocean ridges [Arculus et al., 1991].

[11] Takahasi's model needs improvement because of differences between his speculation and our results. The slab with low seismicity southwest of the summit, shown in Figure 3, is most striking. Because of this slab, the location of the summit does not coincide with the center of the PHS Plate gap, but is shifted westward. East of Mt. Fuji, it is suggested that the PHS Plate is buoyantly subducting and clinging to the bottom of the old PHS Plate fragment (Tanzawa Mountains), which was intensely uplifted in the Quaternary [Huchon and Kitazono, 1984; Ishibashi, 1986]. We suggest that the eastern zone in the gap is under the influence of a compressive stress field because of the collision and buoyant subduction of the Izu block, so magma cannot migrate to the east. Although this interpretation may appear to conflict with the fact that Hakone caldera exists near the collision zone, the present Hakone caldera may also be influenced by a compressive stress field and there is very little new magma rising under the volcano. This idea is supported by the velocity structure in which the region around Hakone shows distinctively high velocities [Lee and Ukawa, 1992]. The inferred tectonics around Mt. Fuji are schematically illustrated in Figure 4. Geodetic data [Sagiya, 1999] which suggest that the Izu Peninsula moves westward relative to other areas may be explained by the splitting of the PHS Plate without invoking microplate tectonics.

Figure 4.

Tectonic implications around Mt. Fuji inferred from the 2-D resistivity structure. Thick broken lines are boundaries between the region in which the subducting Philippine Sea Plate exists and the region in which it does not exist. Thick blue arrows indicate motions of the subducting Philippine Sea Plate inferred from geodetic data [Sagiya, 1999]. Red body indicates the magma chamber under Mt. Fuji located at a depth greater than 20 km. The subducting Philippine Sea Plate is split into two parts and a large magma chamber is located in the gap.

[12] On the basis of seismicity and the velocity structure, Ishida [1992] suggested that the PHS Plate is apt to lead to tears, cracks, or weakening in the wedge-shaped area which extends from the northern part of the Izu peninsula to the northwest, including the area of Mt. Fuji. However recent seismic tomography revealed an aseismic slab at a depth of 70–80 km northwest of Mt. Fuji [Sekiguchi, 2001], therefore it is concluded that the gap of the PHS Plate does not extend to the northwest but exists only around the area of Mt. Fuji. This conclusion suggests that Mt. Fuji was created by the PHS Plate splitting. We propose the following scenario to explain this process. A fracture in the subducting PHS Plate was created somewhere southeast of Mt. Fuji. It expands to the northwest as the PHS Plate subducts, and then reached the area of Mt. Fuji. The split in the PHS Plate was large and deep enough to cause a tensile stress field in the lower and middle crust, allowing high rates of magma (the order of 0.01 km3/year) to ascend, forming a huge symmetric volcanic cone on the surface, which is called Mt. Fuji. Collision of the Izu block with the Honshu block about 2 million years ago may be a cause for the PHS Plate splitting.


[13] We thank to Y. Ogawa, S. B. Tank, W. Kanda, T. Hashimoto, S. Sakanaka, Y. Furukawa, M. Uyesima, T. Ogawa, S. Koyama, I. Shiozaki, T. Uto, A. W. Hurst, M. Yoshimura, K. Yoshimoto, K. Yamazaki, S. Nakao, and T. Kagiyama for their help in the field. T. Mogi and Geographical Survey Institute allowed us to use the MT data as remote-reference processing in this study. Seismic data was provided by Japan Meteorological Agency. Helpful review of manuscript was provided by J. Mori. Comments from Paul A Bedrosian and another anonymous reviewer helped us in improving the manuscript. This research was partly supported by special coordination funds for promoting science and technology.