SEARCH

SEARCH BY CITATION

Keywords:

  • velocity contrast;
  • crust-mantle transition layer;
  • Izu-Bonin island arc;
  • crustal growth;
  • ocean bottom seismograph

Abstract

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Observation and Data
  5. 3. Analysis
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusion
  9. Acknowledgments
  10. References

The Izu-Bonin island arc is a typical oceanic island arc formed by subduction of the Pacific plate beneath the Philippine Sea plate. Although some crustal structures of this island arc have been identified, detailed information about the region between the deeper crust and uppermost mantle, especially the crust-mantle transition layer, has been inadequate to elucidate a crustal growth model. Comparison of the observed and synthetic waveforms of wide-angle seismic data yielded velocity contrast values of the top and bottom of this transition layer beneath the volcanic front along the northern Izu-Bonin island arc. Using those data, we clarify the nature of the crust-mantle transition layer. The top of the crust-mantle transition layer has a velocity contrast value of about 0.4 km/s in the southern area, extending from Kurose hall to Tori-shima along the northern Izu-Bonin island arc, except for the northern area, extending from Nii-jima to Kurose hall (0.25 km/s). The velocity contrast value at the bottom of this transition layer is large (>0.4 km/s) between Nii-jima and Kurose hall and small (0.2 km/s) from Kurose hall southward. These results and the average P wave velocity in the crust-mantle transition layer indicate that this transition layer beneath the volcanic front along the northern Izu-Bonin island arc is a mixture of mafic residues and olivine cumulates formed during crustal growth. Furthermore, the variation of the velocity contrast at the top and bottom of this transition layer and the average P wave velocity in this layer along this arc imply that this layer in the southern area, extending from Kurose hall southward, has a larger ratio of mantle materials (olivine cumulates) to mafic residues than that in the northern area, extending from Nii-jima to Kurose hall. The Moho along the volcanic front in the oceanic island arc probably has a complex signature, as suggested by the crust-mantle transition layer comprising a mixture of mafic residues and olivine cumulates.

1. Introduction

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Observation and Data
  5. 3. Analysis
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusion
  9. Acknowledgments
  10. References

Crustal growth is occurring now in oceanic island arcs [Rudnick, 1995]. For the last decade, to clarify processes of crustal growth, seismic surveys have been conducted at different oceanic island arcs: the northern Izu-Bonin [e.g., Suyehiro et al., 1996], the Mariana [e.g., Takahashi et al., 2007], the Aleutian island arcs [e.g., Holbrook et al., 1999], and the Tonga Ridge [Crawford et al., 2003]. Crustal structures of each oceanic island arc differ: the northern Izu-Bonin island arc has thick tonalitic middle crust with P wave velocity of 6.0 km/s [e.g., Suyehiro et al., 1996]; the Aleutian island arc has little or no middle crust [e.g., Holbrook et al., 1999]. Such differences might reflect different stages of crustal growth.

The Izu-Bonin island arc is a typical oceanic island arc formed by subduction of the Pacific plate beneath the Philippine Sea plate [e.g., Karig and Moore, 1975; Taira et al., 1998]; it extends from Japan to the Mariana Islands parallel to the Izu-Bonin Trench (Figure 1). A seismic survey conducted in the transect of the northern Izu-Bonin island arc revealed that this arc has a middle crust with about 6 km/s of P wave velocity and thickness of about 4.5 km, overlying the lower crust [Suyehiro et al., 1996; Takahashi et al., 1998]. On the basis of results of petrologic studies of the land area where the Izu-Bonin island arc collides with the Japan islands, Suyehiro et al. [1996] and Takahashi et al. [1998] suggested that this middle crust has a granitic-to-tonalitic composition. The salient implication is that the SiO2 composition of the whole and upper-middle crust of the northern Izu-Bonin island arc is comparable to that of a typical average continental crust [e.g., Taira et al., 1998].

image

Figure 1. Map of the local bathymetry around this survey area of the northern Izu-Bonin island arc. The black line shows the air gun shooting line. Positions of deployed ocean bottom seismographs (OBSs) are shown as black and white circles. The black circles show the positions of OBSs used for this study. The inset depicts a topographic map of the area around the Izu-Bonin island arc.

Download figure to PowerPoint

Beneath the volcanic front along the northern Izu-Bonin island arc, this arc has been shown by results of seismic surveys to have not only a middle and lower crust but also mafic to ultramafic cumulates between the lower crust and the uppermost mantle [Kodaira et al., 2007a, 2007b]. Hereinafter, we refer to such mafic to ultramafic cumulates as a “crust-mantle transition layer.” However, because the structural model estimated by Kodaira et al. [2007a, 2007b] was obtained by seismic refraction tomography using first arrival traveltime data and reflection traveltime mapping, little is known about other seismic characteristics such as the seismic reflectivity below the lower crust, including the crust-mantle transition layer.

The seismic reflectivity below the lower crust, especially that between the lower crust and the uppermost mantle, in addition to the distribution of this reflectivity, provide information about the interaction between the crust and mantle in accordance with various tectonic stages. For instance, seismic imaging of the active continental arcs has revealed that a crust-mantle boundary is gradational rather than sharp because of its lack of distinct reflection from this boundary [e.g., ANCORP Working Group, 2003]. This gradational boundary might result from an active process such as the hydration of mantle rocks, magmatic underplating, partial melting, and intraplating under and into the lowermost crust [ANCORP Working Group, 2003]. Moreover, as inferred from the crustal structure of the Bohemian Massif of central Europe, variation of seismic reflectivity from the lower crust to the uppermost mantle is known to correspond to tectonic stages in the lower crust [Hrubcová et al., 2005]. As described above, to understand the process between the crust and mantle at one tectonic stage, it is important to clarify the seismic reflectivity between the lower crust and the uppermost mantle, especially the crust-mantle transition layer, at the tectonic stage.

For this study, to assess the seismic reflectivity between the lower crust and the uppermost mantle, we present the velocity contrast value at the top and bottom of the crust-mantle transition layer and interfaces in the uppermost mantle using a comparison of the synthetic and observed waveforms which is the same data set used by Kodaira et al. [2007a]. Using this data set, we can identify numerous wide-angle reflections and can interpret many of them, depending on their source-receiver offset and traveltime, to be from the top and bottom of a laterally continuous crust-mantle transition layer. Moreover, other wide-angle reflections exist at more distant offsets. We infer that those are reflections from interfaces within the uppermost mantle. We also discuss the characteristics of this transition layer along the northern Izu-Bonin island arc, where the middle crust is produced.

2. Observation and Data

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Observation and Data
  5. 3. Analysis
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusion
  9. Acknowledgments
  10. References

In 2004, a seismic refraction/reflection survey using ocean bottom seismographs (OBSs) and a controlled source was conducted along the northern Izu-Bonin island arc from Sagami Bay to Tori-shima beneath the volcanic front (Figure 1). This survey line is about 550 km long. At approximately 5 km intervals along this line, 103 OBSs were deployed and later recovered by the R/V Kaiyo of the Japan Agency for Marine-Earth Science and Technology (JAMSTEC). The controlled source was an air gun array with a total volume of 12,000 cubic inches (197 L) from the R/V Kaiyo. Details of this cruise are described by Takahashi et al. [2005] and Kodaira et al. [2007a].

In observed waveforms on the vertical component of several OBSs, not only the first arrival phases but also later phases reflected from interfaces in the crust and uppermost mantle are visible (Figures 2a–2h and Figures 3a and 3c through Figures 9a and 9c). As examples, Figures 2a–2h depict observed waveforms to which a predictive deconvolution filter had been applied on the vertical component for OBS006, OBS017, OBS032, OBS034, OBS056, OBS063, OBS078, and OBS095, respectively, along the survey line. For these observed waveforms, digital band-pass filtering of 3.0–10.0 Hz, Wiener predictive deconvolution filtering with a 0.06-s gap and 0.2-s operator length, and 3-s automatic gain control were conducted (Figures 2a–2h). Figures 3a and 3c through Figures 9a and 9c depict observed waveforms to which digital band-pass filtering of 3.0–12.0 Hz had been applied, with no deconvolution filter. In these observed waveforms, some later phases are identified as reflected ones from the top and bottom of the crust-mantle transition layer and from the interfaces in the uppermost mantle based on traveltimes of these reflections from the top and bottom of the mafic to ultramafic cumulates, and the upper mantle reflector in the velocity model [Kodaira et al., 2007a], which was derived using grid-based travel time calculation [Fujie et al., 2003] (Figures 2a–2h and Figures 3a and 3c through Figures 9a and 9c).

image

Figure 2. Observed waveforms with application of a predictive deconvolution filter for (a) OBS006, (b) OBS017, (c) OBS032, (d) OBS034, (e) OBS056, (f) OBS063, (g) OBS078, and (h) OBS095. Observed waveforms were recorded on the vertical component. For observed waveforms, digital band-pass filtering of 3.0–10.0 Hz, Wiener predictive deconvolution filter with 0.06-s gap and 0.2-s operator length, and 3-s automatic gain control was conducted. Vertical and horizontal axes show the travel time reduced by 8 km/s and the offset in kilometers between the OBS and shots, respectively. The red, blue, and green color arrows correspond to a reflected arrival phases as inferred from interpreted to be from the top of the crust-mantle transition layer, the bottom of this layer, and from the interface in the uppermost mantle, respectively.

Download figure to PowerPoint

image

Figure 2. (continued)

Download figure to PowerPoint

image

Figure 2. (continued)

Download figure to PowerPoint

image

Figure 2. (continued)

Download figure to PowerPoint

image

Figure 2. (continued)

Download figure to PowerPoint

image

Figure 2. (continued)

Download figure to PowerPoint

image

Figure 2. (continued)

Download figure to PowerPoint

image

Figure 2. (continued)

Download figure to PowerPoint

image

Figure 3. Observed and synthetic waveforms for (a, b) OBS006 and (c, d) OBS013. Observed waveforms were recorded on the vertical component with no application of a deconvolution filter. For observed and synthetic waveforms, digital band-pass filtering of 3.0–12.0 Hz was used. Trace amplitudes are scaled to increase with the offset. Vertical and horizontal axes of observed and synthetic waveforms, respectively, show the traveltime reduced by 8 km/s and the offset in kilometers between the OBS and shots. The red, blue, and green color arrows correspond to a reflected arrival phases as inferred from interpreted to be from the top of the crust-mantle transition layer, the bottom of this layer, and from the interface in the uppermost mantle, respectively.

Download figure to PowerPoint

In the observed waveform on OBS006 deployed in Sagami Bay, two later phases are apparent at offsets from about 90 km at 6 s and from 125 km at 7.8 s (Figures 2a and 3a). One obvious later phase corresponds to a reflected arrival that was interpreted to be from the top of the crust-mantle transition layer (red arrows). Another later phase corresponds to a reflected arrival inferred from the bottom of this layer (blue arrows). Reflection amplitudes from the bottom of the crust-mantle transition layer are larger than those from the top of that layer. It is obscure for later phases around 150 km at offsets, which correspond to the reflected ones from the interface in the uppermost mantle (green arrows).

In the observed waveform on OBS017 deployed around Nii-jima, later phases are visible at offsets of more than about 70 km at 6.3 s, corresponding to a reflected arrivals inferred from the top of the crust-mantle transition layer (red arrows) (Figures 2b and 4a). Later phases corresponding to a reflected arrival inferred from the bottom of the crust-mantle transition layer are obscure at offsets of greater than 80 km at 7 s (blue arrows).

image

Figure 4. Observed and synthetic waveforms for (a, b) OBS017 and (c, d) OBS027. See Figure 3 for detailed explanations.

Download figure to PowerPoint

Regarding the observed waveform on OBS032 deployed between Mikura-jima and Kurose hall, the later phases are apparent at offsets between −60 and −80 km, and around 100 km, corresponding to the reflected ones from the top of the crust-mantle transition layer (red arrows) (Figures 2c and 5a). Furthermore, the later phases are visible at offsets greater than −80 km at 6.5 s and greater than 110 km at 6.8 s, corresponding to the reflected ones from the bottom of that layer (blue arrows) (Figures 2c and 5a). To the north of this OBS, reflection amplitudes from the bottom of this layer are larger than those from the top. To the south of this OBS, reflection amplitudes from the bottom of this layer are almost identical to those from the top. Taken together, those observations imply that the reflection amplitude pattern between the top and the bottom of this layer varies from Sagami Bay to Aoga-shima. It is obscure for the later phases, which correspond to the reflected ones from the interface in the uppermost mantle at offsets of more than −140 km (green arrows).

image

Figure 5. Observed and synthetic waveforms for (a, b) OBS032 and (c, d) OBS034. See Figure 3 for detailed explanations.

Download figure to PowerPoint

In the observed waveform on OBS034 deployed between Mikura-jima and Kurose hall, later phases are apparent at offsets of more than about 100 km at 6.5 s (red arrows) and more than about 120 km at 7 s (blue arrows) (Figures 2d and 5c). Reflection amplitudes from the bottom of the crust-mantle transition layer are slightly larger than those from the top of that layer. It is obscure for later phases that correspond to the reflection ones from the interface in the uppermost mantle at offsets of greater than 150 km (green arrows).

Regarding the observed waveform on OBS056 deployed on the north of Aoga-shima, it is obscure and complex for later phases, which correspond to reflected ones from the top (red arrows) and bottom of the crust-mantle transition layer (blue arrows) (Figures 2e and 7a). However, the reflected phase from the top of the crust-mantle transition layer has strong amplitude at offsets of more than −110 km (red arrows).

In the observed waveform on OBS063 deployed the south of Aoga-shima, the later phases are apparent around 75 km and between −90 and −120 km at offsets corresponding to the reflected ones from the top of the crust-mantle transition layer (red arrows); they are also visible at offsets around −120 km at 6.5 s, and 90 km at 6 s corresponding to the reflected ones from the bottom of this layer (blue arrows) (Figures 2f and 3c). Although reflection phases from the bottom of this layer overlap the long coda from the reflection of the top of the layer, the reflection amplitudes from the top of the crust-mantle transition layer are slightly larger than those from the bottom in the south side. The amplitudes are obscure for later phases, which correspond to the reflection amplitudes from the interface in the uppermost mantle after reflection from the bottom of this transition layer, in the north side (green arrows).

In the observed waveform on OBS078 deployed between Myojin-sho and Sumisu-jima, the later phases are apparent at offsets around −60 and 80 km, corresponding to the reflected ones from the top of the crust-mantle transition layer (red arrows) (Figures 2g and 8c). Furthermore, the later phases are visible at offsets of more than −100 km at 6.8 s and over 120 km at 7.2 s, corresponding to the reflected ones from the bottom of that layer (blue arrows) (Figures 2g and 8c).

In the observed waveform on OBS095 deployed between South Sumisu caldera and Tori-shima, later phases are apparent between −80 and −100 km and around 60 km at offsets corresponding to the reflection amplitudes from the top of the crust-mantle transition layer (red arrows) (Figures 2h and 9a). However, the waveform of later phases that correspond to the reflected ones from the bottom of this layer at offsets around −90 km at 6.8 s is complex (blue arrows) because the phases from the top of this layer have large amplitudes and long coda (Figures 2h and 9a).

3. Analysis

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Observation and Data
  5. 3. Analysis
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusion
  9. Acknowledgments
  10. References

The velocity and velocity contrast values from the crust-mantle transition layer to the uppermost mantle, as obtained not only by travel times but through comparison of the observed and synthetic waveforms simulated by elastic modeling using finite difference wave propagation, are more accurate than travel times calculated solely using the analytical method used by Kodaira et al. [2007a, 2007b]. The reason is that both travel times and seismic waveforms of the refracted and reflection phases from the crust-mantle transition layer to the uppermost mantle are used. In the present study, to obtain velocity contrast values from the crust-mantle transition layer to the uppermost mantle, we computed synthetic waveforms using a finite difference wave propagation program code “e3d” [Larsen and Grieger, 1998] and compared observed waveforms with synthetic ones. The finite difference method is well suited to solving wave propagation problems; the salient advantage of this method is that it can compute full waveforms in heterogeneous media. This program code of the finite difference method solves the wave equation on a staggered grid [Madariaga, 1976; Virieux, 1986; Levander, 1988]. The finite difference model approximation is second order in time and fourth order in space, with respective grid spacing of 30 m and 2 ms. The source time function uses the zero-phase 5 Hz Ricker wavelet to match these reflections from the top and bottom of the crust-mantle transition layer and interfaces in the uppermost mantle. The dominant frequency of these reflections is around 5 Hz from the amplitude spectrum. In this analysis, reciprocity is assumed because the source sets only one point in this program code.

On the basis of the P wave velocity model obtained by seismic refraction tomography and reflection travel time mapping [Kodaira et al., 2007a], the initial P wave velocity structure model was constructed to include the appropriate velocity contrast value at the top and bottom of the crust-mantle transition layer and interfaces in the uppermost mantle. Then, the velocity model was revised using this program code “e3d” with trial and error, considering the comparison of observed and synthetic waveforms, traveltime, and raypath from these interfaces calculated using grid-based traveltime calculation method [Fujie et al., 2003].

The S wave velocity was assumed as 0.15 km/s within the sedimentary layer (Vp < 2.5 km/s) based on the acoustic basement topography using the MCS profile along this survey line [Tsuru et al., 2005] and traveltimes of PPS phases converting from P waves to S waves at this basement beneath the OBS, as calculated using 2-D ray tracing [Zelt and Ellis, 1988]. However, the S wave velocity below the acoustic basement was impossible to estimate from the traveltimes of PSS phases that had been converted from P waves to S waves at this basement below the shotpoint and from the traveltimes of PSS phases that had been transmitted as S waves through the crust and mantle because these were obscure in vertical and horizontal record sections. An earlier study found Poisson ratios in the middle and lower crust as 0.25 in the northern Izu-Bonin island arc [Takahashi, 1997]. The Poisson ratio in the upper part of the crust is known to be larger than 0.25, although this ratio varies across the northern Izu-Bonin island arc [Takahashi, 1997]. For these reasons, we presumed that the S wave velocity below this basement is as presented in Table 1 from the following Poisson ratios: 0.27 in the upper part of the crust (Vp = 2.5–5.6 km/s), and 0.25 in the middle and lower part of the crust and the crust-mantle transition layer (Vp = 5.6–7.6 km/s) and in the mantle (Vp > 7.6 km/s).

Table 1. S Wave Velocity, Q Values for P Waves, and S Waves in the Sedimentary Layer, the Upper Part of the Crust, the Middle and Lower Part of the Crust and the Crust-Mantle Transition Layer, and the Mantle Used for the Computation of “e3d”a
 VsQpQs
  • a

    S wave velocity is Vs, Q values for P waves is Qp, and S waves is Qs. The sedimentary layer is Vp < 2.5 km/s, the upper part of the crust is Vp = 2.5–5.6 km/s, the middle and lower part of the crust and the crust-mantle transition layer is Vp = 5.6–7.6 km/s, and the mantle is Vp &amp;gt; 7.6 km/s.

Sedimentary layer (Vp < 2.5 km/s)0.1 km/s101
Upper part of the crust (Vp = 2.5–5.6 km/s)1.4–3.2 km/s (σ = 0.27)10050
Middle and lower part of the crust, and the crust-mantle transition layer (Vp = 5.6–7.6 km/s)3.2–4.4 km/s (σ = 0.25)300150
Mantle (Vp > 7.6 km/s)>4.4 km/s (σ = 0.25)500250

The seismic attenuation (Q structure) was also introduced for both P waves and S waves within the computation. In the northern Izu-Bonin island arc, a low-Q zone exists in the upper mantle beneath the volcanic front and active rift zone from attenuation of seismic waves [Suyehiro et al., 1996]. However, the Q values for P waves and S waves have not been obtained for this arc. The three-dimensional Q structures show that the low-Q zone exists along the volcanic front in the northeastern Japan arc and Kanto areas of Japan; the Q values for P waves and S waves around 5 Hz in the low-Q zone are about 150–500 and 100–200 at a depth of 0–40 km, corresponding to the crust and uppermost mantle [e.g., Sekiguchi, 1991; Tsumura et al., 2000; Nakamura et al., 2006]. Furthermore, the Q value for P waves of sand in the marine sediment is about 20–30 [Hamilton, 1972]. The Q value for S waves of Quaternary alluvium in California is low (about 10 at 10–20 Hz) [Gibbs et al., 1994]. Allowing for these results, the Q values for P waves and S waves in the sedimentary layer, upper part of the crust, the middle and lower part of the crust and the crust-mantle transition layer, and the mantle were assumed as presented in Table 1. The density is estimated using the following relationship: ρ = −0.6997 + 2.2302Vp − 0.598Vp2 + 0.07036Vp3 − 0.0028311Vp4 in the sedimentary layer, crust, and mantle [Ludwig et al., 1970].

In all, 35 OBSs data are used for this analysis because these OBSs show almost clear reflection phases from the top and bottom of the crust-mantle transition layer and interfaces in the mantle. Moreover, OBSs used for this analysis are distributed uniformly along this survey line (Figure 1).

4. Results

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Observation and Data
  5. 3. Analysis
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusion
  9. Acknowledgments
  10. References

4.1. Comparison of Observed and Synthetic Waveforms

Figures 3456789present comparisons of observed and synthetic waveforms simulated using the final P wave velocity structure model with velocity contrasts presented in Figure 10. For this analysis, we used the observed waveforms without applying the deconvolution filter to retain the original waveforms. For observed and synthetic waveforms in Figures 3–9, digital band-pass filtering of 3.0–12.0 Hz was used. The synthetic waveforms of each OBS from the final velocity model conform well visually to the observed waveforms. Especially, the characteristics of the reflection phases from the top and bottom of the crust-mantle transition layer and from interfaces in the uppermost mantle are almost replicated by this final model. Furthermore, the strong and weak amplitude patterns of the reflection phases to the first arrival phases on observed waveforms are almost completely replicated by this final model. However, in some OBSs, the synthetic waveforms do not even partially explain the observed waveforms. This discrepancy might result from the topography and/or geometry of the interfaces on the local 3-D effect and/or the local difference of Q values in the crust and uppermost mantle because this velocity model is designated as a 2-D velocity model.

image

Figure 6. Observed and synthetic waveforms for (a, b) OBS042 and (c, d) OBS052. See Figure 3 for detailed explanations.

Download figure to PowerPoint

image

Figure 7. Observed and synthetic waveforms for (a, b) OBS056 and (c, d) OBS063. See Figure 3 for detailed explanations.

Download figure to PowerPoint

image

Figure 8. Observed and synthetic waveforms for (a, b) OBS069 and (c, d) OBS078. See Figure 3 for detailed explanations.

Download figure to PowerPoint

image

Figure 9. Observed and synthetic waveforms for (a, b) OBS095 and (c, d) OBS102. See Figure 3 for detailed explanations.

Download figure to PowerPoint

image

Figure 10. The velocity contrast of the top and bottom of the crust-mantle transition layer, and interfaces in the uppermost mantle beneath the volcanic front along the northern Izu-Bonin island arc superimposed on the final P wave velocity structure model (top). White and clear areas respectively show small and zero velocity contrast. The lines are isovelocity contours whose interval is 0.2 km/s. The distribution of velocity contrast values of the top (red line) and bottom (green line) of the crust-mantle transition layer beneath the volcanic front along this arc (bottom). The velocity contrast values and average P wave velocity in the bottom of this figure are smoothed using the moving averages of 10 km and 3 km, respectively.

Download figure to PowerPoint

4.2. Distribution of the Velocity Contrast

The obtained distribution of the velocity contrast and final P wave velocity structure model is depicted in Figure 10. Along the northern Izu-Bonin island arc, the top of the crust-mantle transition layer has a velocity contrast value of about 0.4 km/s. However, between Nii-jima and Kurose hall (100–200 km), the velocity contrast value at the top of this layer is about 0.25 km/s. The variation of this velocity contrast value is small between Nii-jima and Tori-shima along this arc (100–530 km). On the other hand, the velocity contrast at the bottom of the crust-mantle transition layer has a large variation along this island arc. Between Nii-jima and Kurose hall (100–200 km), the velocity contrast value at the bottom of this transition layer is greater than 0.4 km/s, which is higher than that of the top of this layer. However, this value is approximately 0.2 km/s from Kurose hall to Sumisu-jima (200–420 km). Furthermore, little velocity contrast exists between Sumisu-jima and South Sumisu caldera (420–450 km). For interfaces in the uppermost mantle, a velocity contrast (about 0.1–0.2 km/s) exists in some areas (around Miyake-jima (130 km), between Hachijo-jima and South Hachijo caldera (230–290 km), and around Myojin-sho (370 km)).

The average P wave velocity in the crust-mantle transition layer varies from Nii-jima to Tori-shima (Figure 10). Between Nii-jima and Mikura-jima (100–150 km) and between Hachijo-jima and South Hachijo caldera (230–290 km), this velocity is about 7.2 km/s. The velocity around Kurose hall (200 km) and Aoga-shima (310 km) is about 7.3–7.4 km/s. In addition, from the north of Myojin-sho to Tori-shima (350–530 km), this velocity is about 7.5 km/s. Therefore, the velocity in this layer is greater southward. The velocity of the uppermost mantle between Hachijo-jima and Aoga-shima (230–310 km) is about 7.5 km/s: the slowest along this arc.

4.3. Quantitative Estimation of Velocity and Velocity Contrast

The velocity contrasts at the top and bottom of the crust-mantle transition layer and interfaces in the mantle were obtained through a comparison of observed and synthetic waveforms using the program code “e3d.” However, this program code is unable to estimate the uncertainty of the final P wave velocity structure model and these velocity contrasts. First, to demonstrate the uncertainty of the final model, we calculated the mean residual between observed and calculated traveltimes of the first arrival phases on the P wave velocity structure model using the grid-based traveltime calculation method [Fujie et al., 2003]. The mean traveltimes residual of the final velocity model is small (about ±0.15 s) (Figure 11). Almost all the calculated traveltimes of this model fit within the uncertainties of observed traveltimes, which are assigned 0.1 s depending on the signal-to-noise ratio. Furthermore, we compared the calculated traveltimes of first arrival phases on the final velocity model and those on the velocity model obtained using seismic refraction tomography and reflection traveltime mapping method [Kodaira et al., 2007a]. The difference of traveltimes is also small (about ±0.1 s) (Figure 11); the final velocity model is comparable to the velocity model obtained by Kodaira et al. [2007a]. Second, we calculated the mean residual between observed and calculated traveltimes of phases reflected from the top and bottom of the crust-mantle transition layer. The mean residual of traveltimes of reflected phases of this transition layer is obtained in the same way as that of the first arrival phases. Although this mean residual between observed and calculated travel times of reflected phases of this transition layer is larger than that of the first arrival phases, the calculated traveltimes of this model roughly fit within the uncertainties of observed traveltimes, which are assigned as 0.1–0.4 s, depending on the signal-to-noise ratio (Figure 12).

image

Figure 11. Mean residual between observed and calculated travel times of the first arrival phases on the P wave velocity structure model using the grid-based travel time calculation method [Fujie et al., 2003]. The mean residual between observed and calculated traveltimes for each OBS is shown at the top. The residual between observed and calculated traveltimes for each shot at each OBS is shown in the middle. The residual between calculated travel times for each shot at each OBS on the obtained velocity model and those on the velocity model obtained using seismic refraction tomography and reflection travel time mapping [Kodaira et al., 2007a] is shown at the bottom. The horizontal axis shows the distance from the northern end of the survey line; the vertical axis shows the number of OBS.

Download figure to PowerPoint

image

Figure 12. Mean residual between observed and calculated travel times of the reflected phases from the (left) top and (right) bottom of the crust-mantle transition layer on the P wave velocity structure model using the grid-based travel time calculation method [Fujie et al., 2003]. The residual between observed and calculated traveltimes for each shot at each OBS is shown.

Download figure to PowerPoint

Third, to examine the uncertainties of velocity contrasts at the bottom of this transition layer, we compared the observed waveforms to the synthetic ones using various velocity contrasts at the bottom of this layer in the final model on one OBS. We calculated the amplitude ratio of the reflection phases from the bottom of this layer to the first arrival phases at the same offset (Figure 13). In this study, various velocity contrasts at the bottom of this layer are assigned as 0.0, 0.2, 0.4, 0.6, 0.8, and 1.0 km/s uniformly along this survey line.

image

Figure 13. Comparison of observed and synthetic seismograms of various velocity contrasts at the bottom of the crust-mantle transitional layer for OBS032 on the northern shooting line, and the amplitude ratio of reflection phases from the bottom of this transitional layer to the first arrival phases. (a) The observed seismogram is shown. (b) Synthetic seismograms are computed from the velocity model of Figure 10b and from velocity models of the bottom of this transitional layer having each velocity contrast of (c) 0.0 km/s, (d) 0.2 km/s, (e) 0.4 km/s, (f) 0.6 km/s, (g) 0.8 km/s, and (h) 1.0 km/s. Digital band-pass filtering is 3.0–10.0 Hz. No amplitude correction is adopted. On observed and synthetic seismograms, the horizontal and vertical axes show the offset distance from OBS032 and the traveltime reduced by 8 km/s, respectively. (i) The amplitude ratio of reflection phases from the bottom of this transitional layer to the first arrival phases. The amplitude ratios are observed values (black points), synthetic values of this velocity model shown in Figure 10 (thick red line), and assumed velocity contrasts of 0.0 km/s (pink line), 0.2 km/s (dark pale blue line), 0.4 km/s (blue line), 0.6 km/s (green line), 0.8 km/s (gray line), and 1.0 km/s (yellow ochre line).

Download figure to PowerPoint

On the north side of OBS032, the amplitudes of the later phases are stronger than those of the first arrival phases, which were observed between −103 and −125 km at offsets (Figures 5a and 13a). These later phases correspond to the reflected phases from the bottom of this transition layer between 125 and 145 km in the final P wave velocity model. In this final model, the velocity contrast at the bottom of this transition layer between 125 and 150 km is obtained as about 0.5 km/s, which explains this characteristic of having stronger reflected phases than the first phases (Figures 5b and 13b). For comparison of the observed and synthetic waveforms using various velocity contrasts, this characteristic is almost replicated by models with velocity contrasts having greater than 0.4 km/s (Figure 13).

In general, the amplitude of the observed body wave O at frequency f can be written in the following approximate form.

  • equation image

In that equation, A signifies radiation pattern, B is the geometrical spreading factor, S is the source spectrum, P is the path effect, and I is the instrumental response. The value of A is canceled by the amplitude ratio of the reflection phase to the first arrival phase at the same offset because these phases are generated from the same source. Furthermore, the terms of S(f) and I(f) of the amplitude of reflection phase are regarded as identical terms (S(f) and I(f)) to those of first arrival phase because the calculation of the amplitude ratio of these phases was done at the same offset using the same shot and instrument (OBS). The attenuation effect along the raypath is considered as using Q values for P waves and S waves in the sedimentary layer, the upper part of the crust, the middle and lower part of the crust, and the mantle, as shown in Table 1. For this study, the amplitude ratio of these phases includes B, P, and the reflection coefficient, which resembles the method used by Iidaka et al. [2004]. The geometry at the top and bottom of the crust-mantle transition layer and in the uppermost mantle used the final P wave velocity model. The amplitude ratio of these phases is calculated for models having various velocity contrasts at the bottom of this layer. At the calculation of the amplitude ratio, the respective amplitudes of reflection and first arrival phases are used for maximum amplitude values within the time window with a length of 0.2 s for reflection phases, and 0.15 s for first arrival phases.

The observed amplitude ratio of the reflection phases from the bottom of this layer to the first arrival phases at offsets between −90 and −100 km is smaller than that at offsets between −110 and −130 km (Figure 13i). Furthermore, between −110 and −125 km at offsets, this observed amplitude ratio has a variation and is apparently convex downward at −115 km at offsets and convex upward at offsets between −118 and −122 km. These features of these observed amplitude ratios at offsets between −90 and −130 km almost fit theoretical amplitude ratios using velocity contrasts of the final model (Figure 13i). In addition, this feature between −90 and −125 km is explainable using models with velocity contrasts having 0.4 and 0.6 km/s at the bottom of this transition layer (Figure 13i). These reflection phases observed between −90 and −125 km correspond to those from the bottom of the crust-mantle transition layer between 125 and 150 km in the final P wave velocity model. These results show that the velocity contrast at the bottom of the crust-mantle transition layer between 125 and 150 km in the final P wave velocity model are proper and have uncertainties of the velocity contrasts of less than ±0.2 km/s. From quantitative estimation of the uncertainties of the velocity and velocity contrasts, the obtained velocity contrasts at the top and bottom of the crust-mantle transition layer and interfaces in the uppermost mantle are proper.

4.4. Amplitude Effects on the Low-Velocity Zone and Various Q Values

Next, to exclude the possibility of the wide-angle reflections from originating from the top of a low-velocity zone and effects of various Q values in the crust, we compared the observed waveforms to the synthetic ones in each case. First, we computed synthetic waveforms using the velocity model with the low-velocity zone. We designated that the top of the low-velocity zone (Vp = about 5 km/s and zone's thickness = 2 km, and Vp = about 5 km/s and zone's thickness = 0.2 km) corresponds to the top of the crust-mantle transition layer. In both these cases, the strong amplitude pattern and traveltimes of reflected phases from the top and bottom of the crust-mantle transition layer could not be replicated. Moreover, assuming that the top of the low-velocity zone (Vp = about 5 km/s and zone's thickness = 2 km, and Vp = about 5 km/s and zone's thickness = 0.2 km) corresponds to the bottom of this transition layer, the strong amplitude pattern and traveltimes of reflected phases from the bottom of the crust-mantle transition layer could not also be replicated. On the other hand, assuming that top of the low-velocity zone (Vp = 7.4 km/s and zone's thickness = 1 km) corresponds to the interfaces in the uppermost mantle, synthetic waveforms resemble those without the low-velocity zone. For that reason, we were unable to determine whether the low-velocity zone exists or not in this case.

Second, we computed synthetic waveforms using various Q values in the crust and the crust-mantle transition layer (Vp = 2.5–7.6 km/s) in the final model on one OBS. We assigned various Q values (Qp ranging from 100 to 750, Qs ranging from 50 to 500) in the upper, middle, and lower parts of the crust and the crust-mantle transition layer (Vp = 2.5–7.6 km/s) in the final model based on Q values for P waves and S waves in the crust obtained in the various areas [Sato and Fehler, 1998]. Then, to compare the observed waveforms to the synthetic ones, we used the amplitude ratio of the reflection phase to the first arrival phase at the same offset using the same shot and instrument (OBS) in the manner described in section 4.3 for estimation of the velocity and velocity contrast. The variations of the amplitude ratio of these phases varying the Q values in the crust are smaller than those of the varied velocity contrast values (Figure 14). From this result, the amplitude variation to the effect of Q values in the crust is smaller than those to the varied velocity contrast values. On the other hand, for low Q values (Qp = 100 and Qs = 50), the synthetic waveforms were unable to explain the observed waveforms because of the insufficient amplitude around the selected first arrival phases in the observed waveform. Moreover, in the sedimentary layer, the Q values for S waves might be underestimated. Therefore, the difference of synthetic waveforms of Q values for S waves in the sediments assigned 1 and 10 were examined: these synthetic waveforms were equivalent.

image

Figure 14. Comparison of the amplitude ratio of reflection phases from the bottom of this transitional layer to the first arrival phases of various Q values for P waves (Qp) and S waves (Qs) in the crust and crust-mantle transition layer for OBS032 on the northern shooting line. The amplitude ratios are observed values (black points) and assumed Q values (Qp and Qs) in the crust and crust-mantle transition layer (Vp = 2.5–7.6 km/s) of Qp = 200 and Qs = 100 (pink line), Qp = 300 and Qs = 150 (purple line), Qp = 500 and Qs = 250 (dark pale blue line), and Qp = 750 and Qs = 500 (yellow line). In addition to, for Q values in the upper part of the crust (Qp = 100, Qs = 50, and Vp = 2.5–5.6 km/s) and in the middle and lower part of the crust and the crust-mantle transition layer (Qp = 300, Qs = 150, and Vp = 5.6–7.6 km/s), the amplitude ratios represent synthetic values of this velocity model shown in Figure 10 (red line) and the assumed velocity contrasts of 0.4 km/s (blue line), 0.6 km/s (green line), and 0.8 km/s (gray line).

Download figure to PowerPoint

5. Discussion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Observation and Data
  5. 3. Analysis
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusion
  9. Acknowledgments
  10. References

The top and bottom of the crust-mantle transition layer have some velocity contrast values beneath the volcanic front along the northern Izu-Bonin island arc, except for the bottom of this layer between Sumisu-jima and South Sumisu caldera (420–450 km). The top of the crust-mantle transition layer has a large velocity contrast (about 0.4 km/s) along this arc, except for the northern area ranging from Nii-jima to Kurose hall (about 0.25 km/s). On the other hand, the velocity contrast at the bottom of the crust-mantle transition layer (about 0.2 km/s) is smaller than that at the top of this layer along this arc, except for the northern area ranging from Nii-jima to Kurose hall (about 0.4 km/s). These obtained velocity contrast values at the top and bottom of this transition layer are more accurate than the P wave velocity distribution shown by Kodaira et al. [2007a, 2007b] because these values are obtained using methods which incorporate not only traveltimes but also waveforms of the refracted and reflection phases from the crust-mantle transition layer to the uppermost mantle. For this study, we define the crust as the part above this transition layer for reasons we describe later.

On the basis of obtained crustal structures and petrological modeling of the Izu-Bonin and Mariana island arcs, Tatsumi et al. [2008] suggested that the deeper structure having 7.4–7.7 km/s of P wave velocity might comprises mafic residues transformed from crustal differentiation. The crustal-mantle transition layer in this study corresponds to the upper part of this deeper structure. Furthermore, Tatsumi et al. [2008] proposed that substantial contents of olivine cumulates (dunite), formed during the initial island arc creation and basaltic underplating, exist immediately below the top of the crust-mantle transition layer. Such contents explain the difference between observed and calculated P wave velocity values that is attributable to the petrologically modeled values of the lower crust, the crust-mantle transition layer, and the uppermost mantle. For that explanation to be accurate, this transition layer would necessarily comprise a mixture of mafic residues and olivine cumulates. The constituent material of this transition layer differs from the pure crustal materials. The top of this transitional layer has a large velocity contrast attributable to the different materials in the crust-mantle transition layer and the lower crust. Moreover, the velocity contrast at the bottom of the crust-mantle transition layer is small because the lower part of the crust-mantle transition layer includes more olivine cumulates. Each characteristic of the velocity contrast at the top and bottom of this transition layer implies that this layer mainly comprises a mixture of mafic residues and olivine cumulates (Figure 15).

image

Figure 15. Simplified illustration of interpretations in the crust-mantle transition layer beneath the volcanic front along the northern Izu-Bonin island arc.

Download figure to PowerPoint

The characteristics of velocity contrasts at the top and bottom of the crust-mantle transition layer beneath the volcanic front along the northern Izu-Bonin island arc differ in two areas: the southern area between Kurose hall and the Tori-shima (200–450 km) and the northern area between Nii-jima and Kurose hall (100–200 km). In the southern area, the velocity contrast at the top of the crust-mantle transition layer (about 0.4 km/s) is larger than that at the bottom of this layer (about 0.2 km/s). On the other hands, in the northern area, the velocity contrast at the top of this transition layer (0.25 km/s) is smaller than that at the bottom of this layer (about 0.4 km/s). In addition, at the point of the average P wave velocity in this transition layer, the southern area (about 7.5 km/s) is larger than the northern area (about 7.2 km/s). Moreover, in the frequency distribution of the P wave velocity values on the final structure model, the frequency peak of values between 7.0 and 8.0 km/s of the P wave velocity varies along this arc. The parts between Miyake-jima and Mikura-jima in the northern area (125–145 km) show a peak of 7.2 km/s. The south of Hachijo-jima (240–260 km) has two peaks of 7.3 and 7.7 km/s. In addition, Sumisu-jima (410–430 km) has two peaks of 7.5 and 7.7 km/s (Figure 16). As causes of these differences between the southern and northern area in terms of characteristics of the top and bottom of the velocity contrasts and the average P wave velocity of this transition layer, two candidates are proposed: the temperature and the material composition of this transition layer.

image

Figure 16. Histograms showing the P wave velocity values and frequency distribution of these values between Miyake-jima and Mikura-jima in the northern area (125–145 km), and the south of Hachijo-jima (240–260 km) and Sumisu-jima (410–430 km) in the southern area.

Download figure to PowerPoint

Results of both experimental and theoretical studies suggested that the temperature dependence on the P wave velocity for mafic rocks is ∼5 × 10−4 km/s/K [e.g., Christensen and Mooney, 1995; Hacker et al., 2003]. Accepting this value, then a temperature difference of ∼600 K is necessary for explaining the observed difference in P wave velocity. This is, however, quite unlikely, suggesting that the spatial variation in P wave velocity cannot be explained solely by the temperature dependence of the P wave velocity. From the relation between seismic P wave velocity and the composition of anhydrous igneous and meta-igneous rocks, the velocity increases with increasing mean MgO and CaO contents and decreasing SiO2 content [Behn and Kelemen, 2003]. In addition, Tatsumi et al. [2008] suggested, based on petrological modeling and semitheoretical P wave velocity calculations, that the transition layer would be composed of mafic restite consisting of plagioclase, garnet, and clinopyroxene, after extraction of andesitic melts and olivine cumulate. Results of that study also suggested that the seismic reflector at the bottom of transition layer might result from a discontinuous increase in the garnet fraction within the restite through the reaction (plagioclase + clinopyroxene [RIGHTWARDS ARROW] garnet). Olivine has a P wave velocity (∼7.8 km/s) that is higher than that of restite (7.0–7.5 km/s) [Tatsumi et al., 2008]. Therefore, an increase of olivine/restite ratio might engender a higher P wave velocity (Figure 15). Garnet-bearing mafic restite is denser than olivine cumulate and tends to form the lower part of the layer. If so, then the growth or thickening of the arc crust, via basaltic underplating and subsequent partial melting of the basaltic lower crust, might increase the olivine/restite ratio within the transition layer through transportation of restites to the sublayer.

The top and bottom of the crust-mantle transition layer are respectively located at depths of about 20–25 km and 25–30 km. These depths at the top and bottom of this layer have variations with a 70–110 km wavelength. Nevertheless, the thickness of this transition layer is approximately constant along the northern Izu-Bonin island arc, except for the area ranging from South Sumisu caldera to Tori-shima (480–520 km) (Figure 10). Kodaira et al. [2007a] described that the crustal structure along the northern Izu-Bonin island arc has a variation with an 80–100 km wavelength. The pattern of variation of the depth at the top and bottom of this transition layer resembles that of the thickness of the crustal part. This variation pattern similarity implies that the variation of the depth at the top and bottom of this transition layer is mainly attributable to the variation in the thickness of the crustal part.

From the deeper crust and uppermost mantle, reflectors are confirmed at depths of about 20 km and 30–35 km beneath the Mariana island arc [Takahashi et al., 2008]. Takahashi et al. [2008] described that the reflector located at depths of about 20 km corresponds to the bottom of the 6.7–7.3 km/s layer and that the P wave velocity in the space between these reflectors is lower than the typical upper mantle velocities of 8 km/s. The top of the crust-mantle transition layer corresponds to the bottom of the lower crust (Vp = 6.8–7.2 km/s) defined by Kodaira et al. [2007a, 2007b]. Therefore, the reflector located at depths of about 20 km beneath the Mariana island arc can be interpreted as the top of the crust-mantle transition layer. Moreover, although the P wave velocity of the region between these reflectors beneath the Mariana island arc is higher than that of the crust-mantle transition layer, Takahashi et al. [2008] pointed out that the region between these reflectors corresponds to the crust-mantle transition layer, which implies that the reflector located at depths of about 30–35 km beneath the Mariana island arc seems to be the bottom of the crust-mantle transition layer or the reflector in the uppermost mantle. These might show that the crust-mantle transition layer, a mixture of the mafic residues and olivine cumulates formed during crustal growth, exists in oceanic island arcs such as the northern Izu-Bonin and Mariana island arcs. The difference of the P wave velocity in this transition layer between the northern Izu-Bonin and Mariana island arcs might result from the difference of the olivine/restite ratio mentioned above and that of the stage and process during crustal growth.

Furthermore, results of geologic studies suggest that crustal thickening that occurs by the collision between the Izu-Bonin island arc and Honshu arc takes place in the northernmost Izu-Bonin island arc; the influence of this collision extends to the area around Miyake-jima in the southern limit [e.g., Taira et al., 1989]. Around this southern limit of the influence for the arc-arc collision, characteristics of velocity contrasts at the top and bottom of the crust-mantle transition layer and average P wave velocity in this transition layer differ between the northern and southern areas. Moreover, Kodaira et al. [2007a] describes that the average crustal P wave velocity excluding the mafic to ultramafic cumulates (the crust-mantle transition layer) is low and the middle crust's thickness increases toward the northern area along this island arc. It is possible that the difference between the southern and northern areas in terms of characteristics of the crust and the crust-mantle transition layer includes the influence of the arc–arc collision, although the exact reason and process related to this difference by the collision have not been clarified.

The reflectors in the uppermost mantle, which have a velocity contrast (about 0.1–0.2 km/s), exist around Miyake-jima (130 km), between Hachijo-jima and Aoga-shima (230–310 km), and around Myojin-sho (370 km). Although the absolute velocity values below these reflectors are inaccurate, the values show moderate 8.0 km/s. These reflectors are likely to be the base of a transformed crustal component into the mantle during the crustal evolution process, as inferred from results of petrological modeling and the seismic velocity structure [Tatsumi et al., 2008]. Consequently, these reflectors in the uppermost mantle might mark the boundary between the mafic materials including a transformed crustal component and ultramafic materials such as a mantle peridotite. However, these reflectors are distributed to limited locations along the northern Izu-Bonin island arc. It is necessary to clarify and incorporate results of the other geophysical surveys such as those of the tomography using the natural earthquakes to ascertain the reason for the limited reflectors in the uppermost mantle.

The Moho is normally defined as a seismic discontinuity to a value between 7.6 and 8.6 km/s of P wave; it marks the boundary between the crust and the mantle [e.g., Jarchow and Thompson, 1989]. In the southern area ranging from Kurose hall to Tori-shima (200–450 km), a large velocity contrast corresponds to the top of the crust-mantle transition layer, though this transition layer has about 7.3–7.6 km/s of the average P wave velocity. In the northern area from Nii-jima to Kurose hall (100–200 km), the bottom of this layer has a larger velocity contrast than the velocity contrast at the top of this layer, though this velocity in this transition layer is about 7.2 km/s. In addition, some reflection phases from the deeper crust to the mantle have been observed beneath the volcanic front along the northern Izu-Bonin island arc. The Moho beneath the volcanic front along the northern Izu-Bonin island arc has a complex signature and might not be a single plane. In the volcanic arc of the central Andes, a distinct reflection from the Moho cannot be observed because of hydration of mantle rocks, magmatic underplating, partial melting, and intraplating under and into the lowermost crust beneath the volcanic front [ANCORP Working Group, 2003]. In the Aleutian island arc, the characteristics of these reflection phases from the lower crust, the Moho and upper mantle looking eastward, differ from those looking westward [Fliedner and Klemperer, 1999, 2000]. These differences of characteristics underscore that the location at which the continental crust is generated and has progressed presents a complex reflection pattern from the deeper crust to the uppermost mantle beneath the volcanic front. This complex reflection pattern is explainable by the fact that the composition from the deeper crust to the uppermost mantle is a mixture of mafic residues and olivine cumulates; it changes gradually or slightly. Consequently, it is possible that the Moho in such a location has a complex signature because of the influence of the interaction of crust and mantle during crustal growth.

6. Conclusion

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Observation and Data
  5. 3. Analysis
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusion
  9. Acknowledgments
  10. References

This paper presents velocity contrast values determined for the top and bottom of the crust-mantle transition layer and at the interfaces in the uppermost mantle beneath the volcanic front along the northern Izu-Bonin island arc. The values were obtained through comparison of the observed waveforms and synthetic ones using a finite difference wave propagation code. Synthetic waveforms of each OBS from the final velocity model including these velocity contrast values are well explained using the observed waveforms, particularly, the relative amplitude of the later phases to the first arrival ones.

The top of the crust-mantle transition layer has a velocity contrast value of about 0.4 km/s in the southern area ranging from Kurose hall to Tori-shima, although the northern area ranging from Nii-jima to Kurose hall has a small velocity contrast (0.25 km/s). On the other hand, the velocity contrast value at the bottom of this transition layer is greater than 0.4 km/s in the northern area (Nii-jima to Kurose hall) and about 0.2 km/s in the southern area (Kurose hall to Tori-shima). In the uppermost mantle, an interface has a velocity contrast (0.1–0.2 km/s) around Miyake-jima, between Hachijo-jima and Aoga-shima, and around Myojin-sho.

The crust-mantle transitional layer beneath the volcanic front along the northern Izu-Bonin island arc is interpreted as a mixture of mafic restites and olivine cumulates during crustal growth. The top of this transition layer, which has a large velocity contrast, is apparently the boundary between the mixture of mafic residues and olivine cumulates, and the crustal materials. It is inferred from variations of the velocity contrasts at the top and bottom of this transition layer and the average velocity in this layer that the mixing ratio of materials composed of mafic residues and olivine cumulates in this layer differs in the northern (Nii-jima to Kurose hall) and southern areas (Kurose hall to Tori-shima) along the northern Izu-Bonin arc. The crust-mantle transition layer in the southern area seems to contain mainly mantle materials (olivine cumulates) and small mafic residues, in comparison with this layer in the northern area. The depths at the top and bottom of this transition layer show the variations with a 70–110 km wavelength. However, the thickness of this layer is approximately constant except for the area ranging from South Sumisu caldera to Tori-shima. These results suggest that these variations of the depth are mainly attributable to the variation of the thickness above this transition layer. In addition, some obvious reflection phases are apparent from interfaces within the deeper crust and the uppermost mantle. It is likely that the Moho beneath the volcanic front in the oceanic island arc has a complex signature because of the crust-mantle transition layer composed of a mixture of mafic residues and olivine cumulates.

Acknowledgments

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Observation and Data
  5. 3. Analysis
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusion
  9. Acknowledgments
  10. References

The authors acknowledge Shawn Larsen for permission to use the “e3d” program code. We thank Junzo Kasahara and Kei Murase for useful information and helpful advice related to the use of the program code. We also thank Mikiya Yamashita, Yuka Kaiho, Seiichi Miura, Gou Fujie, and Tsutomu Takahashi for fruitful discussions related to this study. We would like to thank the captain, officers, crew, and technical staff of the R/V Kaiyo and Aki Ito for support during data acquisition. We gratefully acknowledge Editor Peter van Keken, Associate Editor, and two anonymous reviewers for critical comments and suggestions to improve the manuscript. The figures and maps presented with this paper were produced using the GMT graphic package [Wessel and Smith, 1998]. This work is partially supported by a Grant-in-Aid for Creative Scientific Research (19GS0211).

References

  1. Top of page
  2. Abstract
  3. 1. Introduction
  4. 2. Observation and Data
  5. 3. Analysis
  6. 4. Results
  7. 5. Discussion
  8. 6. Conclusion
  9. Acknowledgments
  10. References