SEARCH

SEARCH BY CITATION

Keywords:

  • oceanic plate;
  • oceanic plate;
  • wide-angle seismic survey;
  • Vp/Vs ratio;
  • water content;
  • crustal structure

Abstract

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Survey and data
  5. Vp Structure
  6. Vs and Vp/Vs ratio structures
  7. 5 Discussion and Conclusion
  8. Acknowledgement
  9. References
  10. Supporting Information

[1] Recent seismic structural studies in trench-outer rise regions have shown that Vp within the incoming oceanic plate systematically decreases toward the trench, probably owing to bending and fracturing of the plate. To understand the mechanisms acting to reduce Vp, Vs is critical because the Vp/Vs ratio is a sensitive indicator of lithology, porosity, and the presence of fluid. In the outer rise region of the Kuril trench, we conducted an extensive seismic refraction and reflection survey that revealed systematic changes in Vp, Vs, and Vp/Vs. Our results suggest that water content within the incoming oceanic plate increases toward the trench accompanied by the development of bending-related fractures at the top of the oceanic crust, consistent with the seawater percolation. Our results support the idea that plate bending and fracturing during the bending in the outer rise of the trench play an important role in the water cycle of subduction zones.

1 Introduction

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Survey and data
  5. Vp Structure
  6. Vs and Vp/Vs ratio structures
  7. 5 Discussion and Conclusion
  8. Acknowledgement
  9. References
  10. Supporting Information

[2] Dehydration processes and the expulsion of the water from the subducting oceanic plate affect various subduction-zone processes, including arc volcanism and generation of earthquakes and tremor [Hyndman and Peacock, 2003; Yamasaki and Seno, 2003; Tatsumi, 1989]. Although oceanic plates are thought to acquire water through several mechanisms, including hydrothermal circulation near the spreading centers where they form, recent studies of their seismic and thermal structure suggest that most oceanic plate hydration is associated with faulting related to plate bending in the trench-outer rise region [Peacock, 2001; Ranero et al., 2003].

[3] In the last decade, various seismic refraction and reflection surveys have been conducted in the trench-outer rise region of the Central and South American subduction zones, where relatively young oceanic plates (about 20 Ma) are subducting [Ranero et al., 2003; Ranero and Sallarès, 2004; Grevemeyer et al., 2007; Ivandic et al., 2008; Contreras-Reyes et al., 2008a]. These studies showed that bending-related faults extend from the seafloor into the oceanic upper mantle near the trench and that the compressional velocity (Vp) is low in the region where bending-related faults are observed.

[4] The Vp reduction near the trench can be explained by an increase of porosity owing to fracturing, the presence of free fluids, and a rise in the degree of hydration (serpentinization) [e.g., Van Avendonk et al., 2011]. To better understand the mechanism of the Vp reduction, the shear wave velocity (Vs) is critical because the Vp/Vs ratio, which is directly related to the Poisson's ratio, is a sensitive indicator of lithology, porosity, and the presence of fluid [Christensen, 1996; Takei, 2002]. However, systematic changes in Vs before subduction are not well constrained.

[5] To investigate structural changes in an incoming plate, focusing on its water content, we conducted an extensive seismic refraction and reflection survey in the northwestern Pacific margin, where an old part of the Pacific plate, dating from about 130 Ma, is subducting beneath the Okhotsk plate at the Kuril trench [Müller et al. 2008]. We present models of Vp, Vs, and Vp/Vs derived by traveltime inversion and discuss the systematic changes in the structure of the incoming plate as it approaches the subduction zone.

2 Survey and data

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Survey and data
  5. Vp Structure
  6. Vs and Vp/Vs ratio structures
  7. 5 Discussion and Conclusion
  8. Acknowledgement
  9. References
  10. Supporting Information

[6] In 2009 and 2010, we established a 500 km long seismic profile extending from 10 km south of the Kuril trench to well seaward of the outer rise (Figure 1). The crustal magnetic lineations (tectonic spreading fabric) is sub-parallel to the Kuril trench, and the structural variation along the trench axis is expected to be weak here, although the magnetic lineation in the southern area (where distance along the profile x > 350 km) is obscured by pervasive volcanism and normal faulting [Kobayashi et al., 1998].

image

Figure 1. (top) Bathymetric map of the study area. The depth contour interval is 1000 m; the dashed contour represents 5500 m. The yellow circles represent OBSs, and the red line is the air-gun shot line. (bottom) Seafloor topography along the profile. The trench axis is around x = − 10 km.

Download figure to PowerPoint

[7] We deployed 80 ocean bottom seismometers (OBSs) along the profile at intervals of 6 km and fired a large air gun array (total volume 7200 cubic inches) from the R/V Kairei at 200-m intervals for OBSs and at 50-m intervals for a 444-channel, 6-km-long hydrophone streamer. We deviated from the survey line around site 60 to avoid a group of fishing boats. Therefore, shots between x = 352 and 372 km have been excluded from the structural analysis.

[8] In the time-migrated reflection data (Figure 2a), we observed two prominent horizons interpreted as the top of the oceanic crust (“basement” hereafter) and the oceanic Moho discontinuity. Both horizons are smooth in the central part of the profile (80 < x < 350 km) but rough or obscured at both ends. The onset of horst-and-graben structure is at x ∼ 80 km, and the fault throws gradually increase toward the trench, as is commonly observed in this region [e.g., Kobayashi et al., 1998; Tsuru et al., 2000].

image

Figure 2. Obtained seismic data. (a) Time-migrated multichannel seismic reflection section. (b,c,d) Examples of OBS vertical component data reduced by 8 km/s. (e,f,g) OBS radial component data reduced by 8 km/s. (h,i,j) OBS radial component data reduced by 4.62 km/s. A 5–20 Hz band-pass filter was applied to these OBS sections. See the Auxiliary Material for more sections.

Download figure to PowerPoint

[9] Our OBSs were equipped with one vertical and two horizontal-component geophones. On the vertical record sections (Figures 2b–2d), we identified crustal refractions (Pg), Moho reflections (PmP), and uppermost mantle refractions (Pn). The apparent velocity of the Pn refractions was about 8.5 km/s, which is significantly higher than the Vp of typical oceanic mantle [White et al., 1992] but consistent with the Vp in the northwestern Pacific basin derived from a borehole study [Shinohara et al., 2008] and an active-source seismic structure study [Oikawa et al., 2010] in which Vp showed strong anisotropy and was highest along the spreading direction.

[10] To utilize the horizontal data, we calculated the radial and transverse components by the polarization of the direct water arrivals of air gun shots. The S-wave energy should be recorded mainly on the radial component if the across-profile (along-trench) structure variation is small. Therefore, we used only the radial component for the Vs structural analysis.

[11] As air guns in the water generally emit only P-waves, observed shear waves were P to S (PS) converted waves. The PS converted waves are of two types, PPS and PSS. The PPS is a PS conversion of the ascending ray and has almost the same apparent velocity as P. The PSS is a PS conversion of the descending ray and has roughly inline image the apparent velocity of P. We observed both PPS and PSS crustal refractions with good quality (Figures 2e–2i).

Vp Structure

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Survey and data
  5. Vp Structure
  6. Vs and Vp/Vs ratio structures
  7. 5 Discussion and Conclusion
  8. Acknowledgement
  9. References
  10. Supporting Information

[12] To model the Vp structure along our profile, we adopted a tomographic inversion of traveltimes, which simultaneously determines Vp and layer boundaries [Fujie et al., 2002; Fujie et al., 2006]. At first, to clarify the lateral structural variation objectively, we used wide-angle first arrivals and two-way reflection times from the basement and adopted a starting model consisting of three layers that were almost uniform laterally (Figure 3a). The top layer was seawater with a uniform velocity of 1.5 km/s. The seafloor topography was derived from the multi-narrow beam bathymetric data obtained in our survey and held fixed during inversion. The second layer was the sediments, in which the velocity field was defined in terms of the velocities at the top and bottom of the layer, which were distributed along the profile at nodes spaced 3 km apart. The initial value of Vp were 1.6 km/s at the top and 2.5 km/s at the bottom, and the thickness was calculated from the two-way reflection times from the basement. The basement was represented by nodes with 2-km spacing with one degree of freedom in the vertical direction. The third layer (below the basement) was the oceanic crust and mantle, and its velocity field was represented by a regular grid with nodes spaced 3 km (horizontal) and 0.5 km (vertical) apart.

image

Figure 3. Starting models and results of seismic velocity analysis. (a,b) Vp of three-layer model parameterization. (c,d,e) Vp model and CRT of four-layer model parameterization. The black line in Figure 3e represents the Moho. (f,g) Vs of three-layer model. (h,i,j) Velocity perturbations below basement. (k) Vp/Vs ratio derived from Figures 3b and 3f. Shaded areas indicate areas that are not sampled by raypaths. The contour intervals are 0.5 km/s in Vp models, 0.289 km/s in Vs models, 0.02 in the Vp/Vs model, and 3 % in velocity perturbations. Vertical exaggeration is 8 : 1 except in (k), where it is 15:1.

Download figure to PowerPoint

[13] Using 62,260 wide-angle first arrivals and 1443 multichannel seismic two-way times, we simultaneously determined seismic velocities and the depth of the basement (Figure 3b). During the inversion, the root mean square (RMS) of traveltime residuals was reduced from 663 to 29.3 ms. To illustrate the structural changes within the incoming plate, we calculated Vp perturbations below the basement with respect to an average Vp, which shows that the crustal Vp becomes lower toward the trench (Figure 3h)

[14] As first arrivals alone are insufficient to resolve the trade-off between crustal Vp and crustal thickness, we conducted another traveltime inversion to better constrain the lower crust and the upper most mantle by adding Moho reflections. The second starting model was basically the same as the initial model, but the bottom layer was divided into crust and mantle by incorporating the Moho (Figure 3c). Adding both 11,323 wide-angle PmP and 766 normal-incident Moho reflections, we obtained a four-layer Vp model by traveltime inversion (Figure 3d). The RMS of traveltime residuals was reduced from 177 to 40.2 ms.

[15] To evaluate the reliability of the model, we applied a checkerboard resolution test (CRT) to evaluate the spatial resolution of velocity perturbation, although the CRT is not a tool for evaluating uncertainties. The checkerboard patterns were well recovered down to depths a few kilometers below the Moho (Figure 3e), which means that the lateral Vp perturbations within the oceanic crust and uppermost mantle are reliable at the scale of the checkerboard pattern. Comparing the three-layer and four-layer models, Vp perturbations within the upper crust (about 0–2 km below the basement) were very similar to each other, but those within the deeper parts were somewhat different (Figures 3h and 3i). This is probably because Moho reflections gave additional constraints on the four-layer model, and it suggests that Moho reflections are indispensable for discussing the detailed structure in the lower crust.

[16] The Vp perturbations indicate that the seismic structure differs markedly between the abyssal plain (x > 350 km) and the northern area (x < 350 km). The typical Vp at the top of the oceanic crust is about 5.4 km/s in the northern area and 5.7 km/s in the abyssal plains. The abyssal plains formed near the spreading center by pervasive volcanism and normal faulting, and their formation affects the oceanic crust and mantle structure. Therefore, we infer that the difference around x = 350 km is related to the formation of abyssal plains and not related to structural changes related to impending subduction. Hereafter, we confine our discussion to the northern area to investigate the structural changes in the incoming plate prior to subduction.

[17] In the northern area, the upper crustal Vp begins to decrease at around the center of the outer rise bulge (x = 190 km). On the other hand, the Vp decrease for the lower crust and mantle begins at the edge of the horst-and-graben structure (x = 80 km), suggesting systematic structure changes similar to those in the Central and South American subduction zones. To better understand the mechanism of the Vp reduction, Vs is critical.

Vs and Vp/Vs ratio structures

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Survey and data
  5. Vp Structure
  6. Vs and Vp/Vs ratio structures
  7. 5 Discussion and Conclusion
  8. Acknowledgement
  9. References
  10. Supporting Information

[18] Structural modeling based on marine Vs is usually done in three steps: (1) Vp modeling, (2) Vs modeling above the PS conversion interface using PPS, and (3) Vs modeling below the conversion interface using PSS [e.g., Mjelde et al., 2003; Contreras-Reyes et al., 2008b]. To determine Vs by a traveltime inversion, we extended our traveltime inversion method by implementing a calculation method for PPS and PSS phases [Fujie et al. 2003], which is robust even in complicated Vs models. During steps 2 and 3, Vp and layer boundaries were held fixed. In addition, in the final step, Vs above the conversion interface was also fixed.

[19] We observed PPS crustal refractions and PSS crustal refractions (Sg) of good quality at most sites, but we did not observe mantle refractions (Sn) and Moho reflections at many sites. We thus focused on Vs within the upper crust and adopted a three-layer model parameterization. The typical time delay between PPS and P waves was 2.0 s, and the typical two-way time within the sediments was 0.6 s, suggesting that the Vp/Vs ratio within the sediments was roughly 8.0. This estimation is reasonable for Vp/Vs ratios within sediments on oceanic plates [e.g., Shinohara et al., 2008]. The apparent velocity of crustal refractions Pg and Sg suggested that the Vp/Vs ratio within the oceanic crust was roughly 1.75. On the basis of these estimates, we constructed a starting Vs model from the tomographic Vp model of Figure 3b.

[20] We picked 13,888 PPS crustal refractions and 16,755 PSS refractions (Sg) assuming that the conversion interface was the basement, as the basement was by far the most prominent discontinuity in both the multichannel seismic section and the Vp models. Using these picks, we obtained a tomographic Vs model after applying inversion steps 2 and 3 (Figure 3f). The third inversion step reduced the RMS of the traveltime residuals from 244 to 44.9 ms. The CRT for the final Vs inversion step showed that the lateral Vs variation within the upper crust was reliable at the scale of the checkerboard pattern (Figure 3g).

[21] Similar to Vp, the upper crustal Vs, approximately 3.0 km/s at the top of the oceanic crust, begins to decrease at around x = 190 km (Figure 3j). In contrast, the Vp/Vs ratio, a sensitive indicator of water content, sharply changes at x = 140 km, implying a change in the water content within the upper crust (Figure 3k).

5 Discussion and Conclusion

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Survey and data
  5. Vp Structure
  6. Vs and Vp/Vs ratio structures
  7. 5 Discussion and Conclusion
  8. Acknowledgement
  9. References
  10. Supporting Information

[22] Our seismic velocity models derived by traveltime inversion show systematic structural changes toward the trench (Figure 3). First, Vp and Vs within the oceanic upper crust begin to decrease roughly at the center of the outer rise bulge (x = 190 km), probably owing to the increase in fracture porosity caused by bending-related fracturing.

[23] The Vp/Vs ratio at the top of the oceanic crust is about 1.85 and almost constant to the south of x = 140 km but begins to increase from there. The Vp/Vs ratio depends strongly on water content and the aspect ratio of cracks. As the water content increases at the depth of the oceanic crust, the Vp/Vs ratio increases when the aspect ratio is smaller than ∼ 0.03 but decreases when the aspect ratio is larger [Takei, 2002]. Considering the small degree of curvature of the incoming plate in the outer rise, it is reasonable to hypothesize that bending-related extensional stress creates cracks with small aspect ratio (thin cracks) here [Ivandic et al., 2008]. Therefore, we propose that the water content within the upper crust increases toward the trench, starting from x = 140 km.

[24] The seismic reflection section around x = 140 km provides a clue to the mechanism of the change in the water content (Figure 4). The reflection from the basement is smooth and continuous to the south of x = 140 km, suggesting that the basement is little fractured, but rough and discontinuous to the north of x = 140 km, suggesting that the basement is pervasively fractured. We suggest that north of 140 km well-developed fractures enable water to percolate from the bottom of the sediment into the top of the oceanic crust. South of 140 km, the small plate curvature probably creates the bending-related small cracks within the oceanic upper crust but does not cause pervasive fracturing at the basement. Thus, here water cannot penetrate the basement, and the crustal water content does not change south of 140 km despite crustal cracking.

image

Figure 4. Time-migrated section around the peak of the outer rise. The continuity of the basement reflection changes around x = 140 km. Note the extreme vertical exaggeration.

Download figure to PowerPoint

[25] These structural changes are confined to the upper crust until the onset of horst-and-graben structure (x < 80 km), where Vp within the lower crust and uppermost mantle becomes lower. Because the horst-and-graben structure is thought to develop as a result of recurrent bending-related normal faulting that ruptures the entire oceanic crust and reaches deep into the oceanic mantle, the reduction of Vp within the lower crust and uppermost mantle coincides with the extent of the bending-related normal faulting [e.g., Ranero et al., 2003; Ivandic et al., 2008; Contreras-Reyes et al., 2008a]. Although we could not investigate structural changes around the trench axis owing to the limitation of our OBSs applicable water depth, we would expect larger structure changes toward the trench axis.

[26] The low-Vp mantle may represent serpentinization. Mantle serpentinization would be promoted if bending-related normal faults act as conduits for water to percolate from the upper crust into the mantle [e.g., Ranero et al., 2003; Van Avendonk et al., 2011]. If all of the Vp reduction within the mantle can be attributed to serpentinization, the largest amount of serpentinization is estimated to be roughly 20% [Christensen, 2004], and this estimation is roughly equal to those estimated in Central American subduction zone [Ivandic et al., 2010; Van Avendonk et al., 2011]. To verify this possibility, the mantle Vs, which we could not constrain in this study, is a key. In the trench-outer rise region off Chile, Contreras-Reyes et al. [2008b] observed S-wave mantle refractions (Sn) of good quality and confirmed the presence of low-Vs mantle along a trench-parallel wide-angle seismic survey line, but they could not observe Sn along an across-strike seismic survey line like that of our survey. Therefore, a trench-parallel seismic refraction survey in the Kuril trench might yield effective constraints on the mantle Vs and on the degree of serpentinization.

[27] The most significant feature of our structure models is the lateral variation in the Vp/Vs ratio within the upper crust, suggesting an increase in water content toward the trench. Our results support the idea that structural changes as the incoming plate approaches the subduction zone have a profound impact on the water cycle in subduction zones.

[28] The structural changes in our Vp model in the Kuril trench, where 130 Ma old lithosphere is subducting, are similar to those in the Central and South American subduction trenches (15–50 Ma) and in the Tonga trench (80 Ma) [e.g., Ranero and Sallarès, 2004; Ivandic et al., 2008; Contreras-Reyes et al., 2008a; Contreras-Reyes et al., 2011]. This similarity suggests that the structural changes that precede subduction do not depend much on plate age but are likely controlled by bending-related faulting and fracturing. Because bending-related normal faulting is observed at many subduction zones [e.g., Masson, 1991], we expect that systematic structural changes before subduction, including water percolation, may be shown to be common features in those subduction zones.

Acknowledgement

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Survey and data
  5. Vp Structure
  6. Vs and Vp/Vs ratio structures
  7. 5 Discussion and Conclusion
  8. Acknowledgement
  9. References
  10. Supporting Information

[29] We appreciate Harm Van Avendonk and an anonymous reviewer for useful and constructive comments, which helped improve this manuscript.

References

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Survey and data
  5. Vp Structure
  6. Vs and Vp/Vs ratio structures
  7. 5 Discussion and Conclusion
  8. Acknowledgement
  9. References
  10. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. 1 Introduction
  4. 2 Survey and data
  5. Vp Structure
  6. Vs and Vp/Vs ratio structures
  7. 5 Discussion and Conclusion
  8. Acknowledgement
  9. References
  10. Supporting Information

The supporting information may be found in the online version of this article.

FilenameFormatSizeDescription
supplement.pdfPDF document12466KSupporting Information

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.