Shear wave anisotropy in the crust, mantle ... - Wiley Online Library

Article Volume 12, Number 1 13 January 2011 Q01002, doi:10.1029/2010GC003343 ISSN: 1525‐2027

Shear wave anisotropy in the crust, mantle wedge, and subducting Pacific slab under northeast Japan Zhouchuan Huang Department of Geophysics, Tohoku University, Sendai 980‐8578, Japan ([email protected]) Also at School of Earth Sciences and Engineering, Nanjing University, Nanjing 210093, China

Dapeng Zhao Department of Geophysics, Tohoku University, Sendai 980‐8578, Japan ([email protected])

Liangshu Wang School of Earth Sciences and Engineering, Nanjing University, Nanjing 210093, China [1] To study the anisotropic structure beneath northeast (NE) Japan, we made 4366 shear wave splitting

measurements using high‐quality seismograms of many earthquakes occurring in the crust and the subducting Pacific slab. Our results provide important new information on the S wave anisotropy in the upper crust, lower crust, mantle wedge, and subducting Pacific slab. In the upper crust, the anisotropy is mainly caused by the stress‐aligned fluid‐saturated microcracks. The measured delay times (DTs) increase to 0.10 s at 10– 11 km depth; the fast velocity directions (FVDs) are parallel to either the tectonic stress or the strike of active faults. The maximum DTs for the low‐frequency earthquakes near the Moho are 0.15–0.17 s, suggesting strong anisotropy at the base of the crust or in the uppermost mantle. The measurements for the intermediate‐depth earthquakes in the Pacific slab show dominant E‐W (trench‐normal) FVDs in the back‐arc area and N‐S (trench‐parallel) FVDs in the fore‐arc area. The trench‐normal FVDs in the back‐arc area are caused by the corner flow in the mantle wedge as a result of the subduction of the Pacific plate. The maximum DTs for the slab earthquakes reach 0.30–0.32 s at 100 km depth, but only half of the total DTs are produced in the mantle wedge. The small DTs in the mantle wedge may result from an isotropic or weak anisotropic zone in the middle of the mantle wedge. In the fore arc, the dominant trench‐parallel FVDs for the slab earthquakes are consistent with those for the upper crust earthquakes, and ∼80% of the total DTs can be accounted for by the anisotropy in the crust. In the subducting Pacific slab, the trench‐parallel FVDs may reflect either the original fossil anisotropy in the Pacific plate when the plate was produced in the mid‐ocean ridge or the preferred orientations of the crystals and cracks in the upper part of the subducting slab. Components: 8500 words, 15 figures. Keywords: shear wave splitting; anisotropy; subduction zone; low‐frequency earthquake; mantle wedge; subducting slab. Index Terms: 7240 Seismology: Subduction zones (1207, 1219, 1240); 7270 Seismology: Tomography (6982, 8180). Received 24 August 2010; Revised 22 October 2010; Accepted 29 November 2010; Published 13 January 2011. Huang, Z., D. Zhao, and L. Wang (2011), Shear wave anisotropy in the crust, mantle wedge, and subducting Pacific slab under northeast Japan, Geochem. Geophys. Geosyst., 12, Q01002, doi:10.1029/2010GC003343.

Copyright 2011 by the American Geophysical Union

1 of 17

Geochemistry Geophysics Geosystems

3

G

HUANG ET AL.: SHEAR WAVE ANISOTROPY UNDER NE JAPAN

10.1029/2010GC003343

Figure 1. (a) Tectonic background of the study area. The bold curves denote the plate boundaries. The dashed line shows the estimated boundary between the Amurian and Okhotsk plates. The solid triangles denote the active volcanoes. The bold arrow indicates the motion direction of the Pacific plate relative to the Eurasian plate. (b and c) Two east‐west vertical cross sections showing the seismicity within a 100 km width along the profiles of 39°N and 40°N under NE Japan. Open circles denote the low‐frequency earthquakes (M ≥ 0.5) between 1997 and 2007 determined by the JMA. The horizontal bold line and solid triangles at the top denote the land area and the active volcanoes, respectively. The three dashed lines denote the Conrad and Moho discontinuities and the upper boundary of the subducting Pacific slab. VF, the volcanic front.

1. Introduction

of the inclined low‐V zone in the mantle wedge [Hasegawa et al., 2009; Zhao et al., 2009].

[2] The northeast (NE) Japan arc is a typical subduction zone where the Pacific plate is subducting beneath the Eurasian and Okhotsk plates at a rate of ∼9 cm/yr (Figure 1a) [Bird, 2003]. Thanks to the dense seismic networks installed on the Japan islands, NE Japan is by far the best studied subduction zone in the world (see recent reviews by Hasegawa et al. [2009] and Zhao [2009]). The double‐planed deep seismic zone found so far in many subduction zones was first discovered unambiguously in this region over three decades ago (Figures 1b and 1c) [Umino and Hasegawa, 1975; Hasegawa et al., 1978]. The detection and analysis of converted and reflected waves revealed that the upper boundary of the subducting Pacific slab is a sharp seismic discontinuity and the P wave velocity contrast between the slab and the mantle wedge amounts to 6% [Hasegawa et al., 1978; Matsuzawa et al., 1986, 1990; Zhao et al., 1997]. Traveltime and attenuation tomography imaged clearly the high‐velocity (high‐V), low‐attenuation (high‐Q) subducting Pacific slab and arc‐magma related low‐velocity (low‐V) and high‐attenuation (low‐Q) anomalies in the crust and upper mantle wedge under the active arc volcanoes [Zhao et al., 1992, 2009; Tsumura et al., 2000; Hasegawa et al., 2009]. Recent studies revealed along‐arc variations

[3] The influences of fluid and its transportation on the subduction dynamics have been discussed extensively based on the above mentioned seismological results and the numerical modeling [e.g., Iwamori and Zhao, 2000; Hasegawa et al., 2009; Zhao, 2009]. Water is released from the subducting slab down to 70–90 km depth and is hosted by a hydrated layer of serpentine and chlorite just above the slab [e.g., Nakajima et al., 2009]. The hydrated layer is brought down to 150–200 km depth by the subducting slab where further dehydration occurs. The dehydrated fluids rise up to the mantle wedge and subsequently is transported upward by the subduction‐induced, secondary mantle convection [Iwamori and Zhao, 2000; Hasegawa et al., 2009; Zhao et al., 2009; Huang et al., 2011]. [4] Studying seismic anisotropy can provide important constraints on the mantle flow pattern. Both observations and laboratory experiments show that flow‐associated shear strain preferentially aligns olivine and pyroxene crystals in mantle peridotite [e.g., Nicolas and Christensen, 1987; Wiens et al., 2008]. Many researchers suggested that the fast velocity direction (FVD) of mantle minerals aligns with the mantle flow direction [Mainprice, 2007; Karato et al., 2008, and references therein]. A shear 2 of 17


3

G


wave splits into two components with orthogonal polarizations when propagating in the anisotropic media, with one component (fast wave) travels faster than the other (slow wave). Since the early studies [e.g., Ando et al., 1980; Fukao, 1984; Silver and Chan, 1988], shear wave splitting has emerged as a popular tool for characterizing anisotropy in the Earth [Long and Silver, 2009]. The splitting parameters, i.e., the FVD and the delay time between the fast and slow waves (DT) dt, have been determined at stations worldwide. [5] Shear wave splitting under NE Japan has been investigated by using the intermediate‐depth earthquakes in the subducting Pacific slab [Okada et al., 1995; Nakajima and Hasegawa, 2004]. The trench‐ normal FVDs are observed in the back‐arc area, and they are considered to be the results of the subduction‐induced convection in the mantle wedge. Nevertheless, by investigating the dominant deformation mechanism in the olivine‐rich mantle, Katayama [2009] argued that most of the mantle wedge is isotropic due to the diffusion creep and two thin anisotropic layers of dislocation creep is located above the subduction slab and right beneath the island arc crust (i.e., in the uppermost mantle). [6] It is very interesting and important to clarify which portion of a subduction zone is actually isotropic and anisotropic. By using local S waves and the SKS‐type phases, the anisotropy in the layers above the subducting slab and the subslab mantle can be isolated [e.g., Long and Silver, 2008, 2009]. In this study, we studied systematically the anisotropic structures under NE Japan by investigating shear wave splitting for local earthquakes of different depths under NE Japan, i.e., the earthquakes in the upper crust, lower crust, and the subducting Pacific slab (Figures 1b and 1c). The anisotropies in different layers are isolated by using our shear wave splitting measurements. The present results shed new light on the seismic anisotropy and dynamic processes in the Japan subduction zone.

2. Data and Analysis [7] The rotation cross‐correlation method (RCC) [Fukao, 1984; Okada et al., 1995; Wüstefeld et al., 2008] was used to determine the splitting parameters (, dt). A grid search approach was used to identify the best fitting splitting parameters by rotating and time shifting the horizontal components. The RCC method seeks to maximize the cross correlation between the corrected fast and

10.1029/2010GC003343

slow components when the effect of splitting is accounted for [Long and Silver, 2009]. The horizontal components were rotated between −90° and 90° with a step of 1° and the time was shifted from 0 s to 0.5 s with the step of 0.01 s. The F test was used to estimate individual errors of the FVD and DT with 95% confidence [Silver and Chan, 1991]. The shear wave splitting analysis was carried out by using the SplitLab software developed by Wüstefeld et al. [2008], which allows us to evaluate the result better by estimating the quality of the original seismograms, the corrected fast and slow components, the initial and corrected particle motions and the distribution of correlation coefficient simultaneously. [8] The seismograms used in the present work were recorded during 2002 to 2009 by the seismic stations deployed in NE Japan by Japan Meteorological Agency (JMA) and National Research Institute for Earth Sciences and Disaster Prevention (NIED) for the High‐Sensitivity Seismic Network (Hi‐net) (Figures 2a, 3a, and 4a). The earthquakes can be divided into four groups: (1) the earthquakes occurring in the upper crust (Figure 2a), (2) the low‐frequency (LF) earthquakes occurring near the Moho discontinuity under the active volcanoes (Figure 3a), and (3) the upper plane and (4) the lower plane intermediate‐depth earthquakes occurring in the double‐planed deep seismic zone in the subducting Pacific slab (for short, slab earthquakes) (Figure 4). The upper crust and LF earthquakes are larger than M 0.5 and the slab earthquakes are greater than M 2.5. Note that many small earthquakes occur also in the lower crust and mantle wedge under the Pacific coast area due to the very low temperature there (Figure 1) [Furukawa, 1993]; we selected some of these earthquakes (≤25 km) in our data set. [9] We follow the criteria described by Okada et al.

[1995] to select the appropriate seismograms for shear wave splitting analysis. The incident angles of the raypaths are smaller than 40° to avoid the phase shift of particle motions by converted phases at the surface [Nuttli, 1961]. The data with the maximum cross‐correlation coefficient 0.5, respectively (see the text for details). The dashed square shows the location of Figure 10.

filtered the seismograms of the LF earthquakes at 2.0–6.0 Hz and obtained identical measurements (Figures S2 and S3), which indicates that the splitting parameters, specially DT, of the LF earthquakes differ little in the two frequency bands. [10] Thanks to the dense seismic networks and the

high seismicity in NE Japan, more than 20,000 high‐ quality records are available for the shear wave splitting analysis. We only chose the very reliable seismograms showing visible splitting in our analysis. For the upper crust and slab earthquakes (Figures S1 and S4), the uncertainties of FVD and DT are generally smaller than 20° and 0.02 s, respectively. The uncertainties for the LF earthquakes are slightly larger due to the lower signal‐noise ratio (SNR), and the errors can reach to 0.05 s for DT and 50° for FVD (Figures S2, S3, and 12). Figure S5 shows the 4366 measured FVDs as a function of the initial polarizations of the seismograms [Vidale, 1986]. Most of the measurements concentrate near

the areas where the FVDs are 45° away from, not parallel or perpendicular to, the initial polarizations, which indicates the measurements are reliable [Wüstefeld and Bokelmann, 2007]. [11] The mean splitting parameters are calculated

at each station by using the method of Audoine et al. [2004]. The averages of DTs are calculated by using the standard Gaussian statistics. The mean FVD mean can be expressed as: mean ¼

hX i X 1 sinð2 i Þ= cosð2 i Þ ; atan i¼1;nobs i¼1;nobs 2

where i is the FVD for the ith measurement and nobs is the number of observations. The deviation S, also called circular variance, is given by: S ¼ 1

rhffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi i2 hP i2 P i¼1;nobs sinð2 i Þ þ i¼1;nobs cosð2 i Þ nobs

:

4 of 17


3

G


10.1029/2010GC003343

Figure 3. (a) The seismic stations (solid squares) and LF earthquakes (open circles) used for the shear wave splitting analysis. Solid triangles denote the active volcanoes. AB denotes a profile shown in Figure 11. (b) Map showing the splitting parameters plotted at the middle of each raypath. The orientation and length of the bars denote the FVD and DT with the scale shown in the bottom right corner. The bold arrow indicates the motion direction of the Pacific plate (PAC) relative to the Eurasian plate.

It is equal to 0 if all the measurements are aligned and equal to 1 if they are completely random [Audoine et al. 2004]. [12] Figure 2b shows the 1299 measured splitting

parameters for the upper crust earthquakes (Figure 2a) plotted at the middle of each raypath. Figures 2c and 2d show the rose diagrams and mean splitting parameters at each station (Table S1). The FVDs are complex in the back arc and the volcanic front (VF) but are generally N‐S in the fore‐arc area. Most of the DTs are smaller than 0.1 s, but large DTs appear at many stations in the fore‐arc area (Figure 5). The maximum DT reaches 0.26 s at station TU.WYG (Figures 2c and 5a). We have carefully checked these measurements and confirmed that they are reliable (Figure 6). [13] Figure 3b shows the 60 measurements for the LF

earthquakes occurring near the Moho (Figure 3a). As a whole, the spatial pattern of the FVD is very complex, particularly near the VF. The FVDs are NE‐SW in the back arc, normal to the subduction direction of the Pacific plate. The DTs are larger than that measured from the upper crust earthquakes (Figure 2) and the maximum values reach 0.15–0.17 s. [14] Figure 7 shows the 3007 measurements for the

slab earthquakes (Figure 4 and Table S2). The dom-

inant FVDs are E‐W in the back arc and N‐S in the fore‐arc area, similar to previous results [Okada et al., 1995; Nakajima and Hasegawa, 2004]. The DTs are larger in the back arc (0.10–0.14 s) than those in the fore arc (0.07–0.10 s) and the maximum values are 0.32 s and 0.27 s, respectively (Figure 8). The measured DTs, however, are much smaller than those in other subduction zones [e.g., Long and Silver, 2008, and references therein].

3. Anisotropy in the Upper Crust [15] Wüstefeld and Bokelmann [2007] demonstrated

that the RCC method is unstable when the initial polarization of the seismograms becomes parallel with or perpendicular to the FVD of the media. They also showed, through synthetic experiment, that the splitting parameters (, dt) follow a specific pattern as a function of the initial polarizations [see Wüstefeld and Bokelmann, 2007, Figure 2]. Figure 9 shows the measured splitting parameters plotted against the initial polarizations and the corresponding density map at station N.ICWH. The station is located close to the main shock epicenter of the 2008 Iwate‐Miyagi earthquake (Figures 2c and 10) and so has abundant high‐quality seismograms for reliable shear wave splitting analysis. The pattern of FVDs, with the mean of −10°∼−40°, agrees well 5 of 17


3

G


10.1029/2010GC003343

iations in DT are only statistically valid in averages of substantial data sets [Crampin and Peacock, 2008]. [16] The DTs for the upper crust earthquakes are

generally smaller than 0.1 s and they increase as a function of focal depth to ∼11 km (Figure 5b). The trend becomes more visible at station N.ICWH when only the measurements with high SNR are plotted [Verdon, 2010] (Figure S6). In the fore‐arc area where large DTs are obtained, the values do not increase at depths greater than ∼13 km. These results suggest strong anisotropy in the upper crust and the anisotropic layer is mainly constrained in the top 11–13 km. [17] The FVDs show significant lateral variations

(Figure 2c); they are complex in the back arc and VF and are nearly N‐S in the fore‐arc area. Figure 10 shows the mean splitting parameters at six stations near the source area of the 2008 Iwate‐Miyagi earthquake (M 7.2); the abundant aftershocks have provided many high‐quality data for the splitting analysis. By considering the error bars, the FVDs are consistent with the tectonic stress at the four stations far away from the main shock [Seno, 1999]. The anisotropy in the upper crust is mainly caused by the cracks, specifically stress‐aligned fluid‐ saturated microcracks [Crampin and Peacock, 2008]. The FVDs are usually consistent with the direction of the maximum horizontal compression (tectonic stress), as shown here (Figure 10) and in the previous studies [e.g., Kaneshima, 1990]. [18] At two stations near the main shock, however, Figure 4. (a) Seismic stations (solid squares) and earthquakes occurring in the Pacific slab used for the shear wave splitting analysis. Earthquakes occurring in the upper and lower planes of the double seismic zone are shown in crosses and circles, respectively. The solid triangles denote the active volcanoes. (b) East‐west vertical cross section of the earthquakes shown in Figure 4a.

with the synthetic experiment (Figure 9b). We divide the measurements into four groups (Figure 9a): two groups with polarizations parallel (k) or perpendicular to (?) the mean FVD and the other two “confident” (C) groups with polarizations ∼45° away from the mean FVD. The trend of the DTs, however, is not clear (Figure 9), which indicates that the errors of the DTs are much larger. Crampin et al. [2004] showed that the DT measurements for small earthquakes typically display a scatter of as much as ±80% about the mean values. Consequently, var-

the FVDs agree well with the strikes of the active faults, being normal to the tectonic stress. The FVDs at station N.ICWH do not show significant temporal variations (Figure S7), suggesting that they formed before the 2008 Iwate‐Miyagi earthquake. The stress‐normal FVDs were also found in many other regions, such as southwest Japan [Kaneshima, 1990], Parkfield, central California [Liu et al., 1997, 2008; Zinke and Zoback, 2000] and Iceland [Crampin and Peacock, 2008]. Crampin et al. [2004] suggested that the 90° flips in shear wave polarizations are caused by high pore fluid pressures near the seismically active fault zone. The microcrack distribution is rearranged by opening and closing fluid‐saturated microcracks due to varying pore pressures [Crampin and Peacock, 2008], which has been observed in the rock samples taken at the Parkfield, central California [Almeida et al., 2005; Liu et al., 2008]. Thus the cracks or fractures near active faults are dominantly parallel to the fault planes rather than directly controlled by the tectonic stress [Kaneshima, 1990; Leary 6 of 17


3

G


10.1029/2010GC003343

Figure 5. (a) The measured DTs (crosses) for the upper crust earthquakes (Figure 2a) plotted against longitude at the middle of each raypath. Open circles with bold bars denote the mean values and the standard deviations at each station. Solid triangle at the top indicates the location of the volcanic front (VF). (b) The measured DTs plotted against the focal depths. Open circles with bold bars denote the mean values and the standard deviations in a 4 km bin around the mean depth.

et al., 1990; Zinke and Zoback, 2000]. Figure S8 shows the azimuth and angle stereogram of the splitting parameters at station N.ICWH. The FVDs rotate clockwise from NW‐SE in the northwest to nearly N‐S in the southeast, which suggests that the anisotropy in the upper crust can change significantly even in a small area.

4. Anisotropy in the Lower Crust [19] The FVDs for the LF earthquakes are complex

under the VF and they are NE‐SW in the back arc (Figure 3b). As mentioned above, the FVDs in the upper crust are parallel to either the tectonic stress

or the strikes of active faults. The complex measurements for the LF earthquakes mainly reflect the seismic anisotropy in the lower crust since the rays pass through both the upper and lower crust (Figure 11). Unlike in the upper crust where the anisotropy is caused by microcracks, the anisotropy in the lower crust reflects the intrinsic rock anisotropy resulting from the LPO of minerals such as mica and amphibole [Kaneshima, 1990; Weiss et al., 1999; Mahan, 2006; Moschetti et al., 2010]. As a result of ductile deformation and related mineral alignment, most of the samples of typical lower crustal rocks exhibit a significant azimuthal variation of seismic velocity. Under the active volca7 of 17


3

G


10.1029/2010GC003343

Figure 6. Six examples showing the seismograms of the fast (top curve) and slow (bottom curve) waves at stations in the fore arc. The time scale is shown at the bottom.

noes, magmas usually intrude through the existed complex channels and cracks which were generated in the geological history. The LPO of the minerals in the lower crust induced by the magma intrusions is hard to predict, which results in the complex anisotropy and nearly random FVDs (Figure 3). In the back‐arc area where the present volcanic activity is weaker, the LPO of minerals will develop

under the nearly E‐W compression due to the motion of the Pacific plate relative to the Eurasian (or Amurian) plate. [20] Figure 12 shows the 60 measured DTs plotted

against the focal depths (Figures 12a and 12c) and against the distance from the LF hypocenters to the Moho discontinuity (Figures 12b and 12d). For the

Figure 7. The same as Figure 2 but for the slab earthquakes (Figure 4). Figures 7a, 7b, and 7c correspond to Figures 2b, 2c, and 2d, respectively. 8 of 17


3

G


10.1029/2010GC003343

Figure 8. The measured DTs for the slab earthquakes (Figure 4) plotted against the longitude at the middle of each raypath. Open circles with bold bars denote the mean values and the standard deviations in a 0.4° bin. Solid triangle at the top denotes the location of the volcanic front (VF).

LF earthquakes with focal depths 0.3 s) are generally measured from the upper plane earthquakes in the double seismic zone (Figure 14b) which occur right beneath the upper boundary of the subducting Pacific slab [Matsuzawa et al., 1990; Zhao et al., 1997]. Therefore we consider that the seismic anisot-

NE‐SW (nearly trench‐parallel) FVDs (TU.IW3 and YUWA in Figure 7). The trench‐parallel FVDs may be mainly caused by the anisotropy in the crust, because the FVDs for both the upper crust and LF earthquakes in this area are consistent with the FVDs for the slab earthquakes, considering the range of uncertainties of the measurements (Figures 2 and 3). [27] In the fore‐arc region, our measurements for

both the crustal and slab earthquakes show dominant trench‐parallel FVDs (Figure 13) and the maximum DTs for the crustal earthquakes make up as large as 80% of the total DTs for the slab earthquakes (Figure 14c). The results suggest that the observed shear wave splitting in the fore arc is mainly caused by the anisotropy in the crust. Thus we can easily explain the lateral variations in the splitting parameters in a short spatial scale in the fore arc, such as at stations OFUNAI and TU.SN3 (Figure 7b). With a separation of only ∼15 km, the FVDs at the two stations are nearly perpendicular to each other (Figure 7b and Table S2). While the rays to the two stations are very close in the subducting slab and the mantle wedge, they are separated in the crust (Figure S9). Therefore the differences in the splitting parameters between 12 of 17


3

G


10.1029/2010GC003343

Figure 14. (a) Raypaths for the upper crust (blue lines), LF (red lines), upper plane (gray lines from black crosses), and lower plane earthquakes (gray lines from the open circles) in the subducting Pacific slab used in this study. Solid squares at the top denote the seismic stations, and the triangle indicates the location of the volcanic front (VF). The dashed lines denote the Conrad and the Moho discontinuities and the upper boundary of the subducting Pacific slab at the latitude of 40°N. The numbers of the measurements from the four types of earthquakes are shown in the inset. (b) The DTs plotted against the focal depth for the earthquakes whose raypaths only pass though the back‐arc area shown in Figure 14a. The dashed line indicates the envelope of the maximum DT. The values of the maximum DT for earthquakes in different layers are shown with arrows. (c) The same as Figure 14b but for earthquakes in the fore‐arc area.

the two stations are most possibly caused by the anisotropy in the crust. [28] Yet another question arises: how to explain the

different measurements at the same station from different earthquakes if the shear wave splitting is

caused by the crustal anisotropy? For example, at station N.RZTH, the measurements for two different groups of slab earthquakes yielded different FVDs (Figure S10). For the two groups of earthquakes which were recorded by the same station, the raypaths separate in most of the depth ranges, 13 of 17


3

G


particularly in the shallow crust (Figure S10a). Hence the result does not conflict with our conclusion that the shear wave splitting in the fore arc is mainly caused by the anisotropy in the crust. This inference can also explain the shear wave splitting at station TU.WYG where the measurements for the crustal and slab earthquakes (with different azimuths) show different FVDs though the DTs are comparable (Figure S11). [29] Many lower plane earthquakes in the double

seismic zone are used in the present work (Figures 4 and 14). The rays of these earthquakes travel 30– 60 km in the subducting Pacific slab and they provide much information on the anisotropy in the slab. The measurements for the lower plane earthquakes, however, do not show significant difference from those of the upper plane earthquakes. In particular, the DTs for the lower plane earthquakes are in the same order as those for the upper plane earthquake. Since the crust accounts for 80% of the total DTs for the slab earthquakes in the fore‐arc area, the slab only produces the remaining DTs in the order of 0.1 s. The relatively weak anisotropy in the subducting slab is inconsistent with the model of Faccenda et al. [2008] who proposed that the strong (1–2 s) trench‐parallel SKS splitting observed in many subduction zones worldwide can be explained with strong anisotropy in the shallow part of the slab. The inconsistency may result from the frequency‐dependent measurements as mentioned above. The dominant frequency of the SKS phase (e.g., 0.2–0.02 Hz or 5–50 s) is much lower than that of S waves from the local earthquakes (2–6 Hz). The seismic anisotropy induced by the large faults in the slab (2.5 km apart) along with serpentinization could be detected by the low‐ frequency SKS but not by the high‐frequency local S waves [Faccenda et al., 2008]. [30] Recently P wave anisotropic tomography carried out in NE Japan shows dominant N‐S (trench‐ parallel) FVDs in the subducting slab [Ishise and Oda, 2005; Wang and Zhao, 2008, 2010; Huang et al., 2011], which is consistent with this study. The trench‐parallel FVDs in the slab may reflect the original fossil anisotropy when the Pacific plate was produced in the mid‐ocean ridge [Hess, 1964; Francis, 1969; Ishise and Oda, 2005; Wang and Zhao, 2008]. Studies of submarine geology and thermal‐mechanical models provide another explanation [Masson, 1991; Kobayashi et al., 1998; Faccenda et al., 2008; Eberhart‐Phillips and Reyners, 2009; Healy et al., 2009]. Many trench‐parallel normal faults are produced in the

10.1029/2010GC003343

outer slope of the Pacific plate due to the strong extension (bending) when the plate enters the mantle (Figure 15) [Masson, 1991; Kobayashi et al., 1998]. The faults are closed under compression due to the slab unbending after the plate subducts beneath Honshu [Hasegawa et al., 1978]. The trench‐parallel (stress‐normal) lattice‐preferred orientation of the crystals and cracks may develop in the upper part of the subducting slab [Faccenda et al., 2008; Healy et al., 2009]. Although the local S waves are not sensitive to the anisotropy induced by the large faults, they may detect the anisotropy induce by the LPO of crystals and cracks which produce the trench‐parallel FVDs.

6. Conclusions [31] We made detailed shear wave splitting analy-

ses to study the anisotropic structure beneath NE Japan using a large amount of high‐quality waveform data from many upper crust earthquakes, low‐ frequency earthquakes in the lower crust, and intermediate‐depth earthquakes in the subducting Pacific slab. A total of 4366 shear wave splitting measurements provide important new information on the seismic anisotropy in the upper crust, the lower crust, the mantle wedge and the subducting Pacific slab under NE Japan. Main findings of this study are summarized as follows and a cartoon (Figure 15) is made to illustrate the anisotropic structures under NE Japan by combining the shear wave splitting measurements with the results of P wave anisotropy tomography [Wang and Zhao, 2008, 2010; Huang et al., 2011]. [32] 1. The splitting parameters for the upper crust

earthquakes suggest strong anisotropy in the upper crust. The DTs increase with the focal depth to ∼11 km where the maximum DTs reach ∼0.10 s. The anisotropy in the upper crust is mainly caused by the stress‐aligned fluid‐saturated microcracks, and the FVDs are parallel to either the tectonic stress or the strikes of active faults. Large DTs (0.10–0.26 s) are found at many stations in the fore arc where the N‐S (trench‐parallel) FVDs dominate. [33] 2. For the LF earthquakes occurring in the

lower crust only under active volcanoes along the volcanic front and in the back‐arc area, the DTs increase rapidly from 0.10 s at 29 km depth to 0.15–0.17 s near the Moho discontinuity. The result suggests an ∼5 km thick anisotropic layer at the

14 of 17


3

G


10.1029/2010GC003343

Figure 15. A cartoon showing the anisotropic structures under NE Japan. Anisotropy is considered to exist in the gray shaded area. The dark and light colors indicate strong and weak anisotropy, respectively. The horizontal bars and circled crosses denote the trench‐normal and trench‐parallel FVDs, respectively. The curve with a downgoing arrow denotes the mantle flow in the mantle wedge driven by the subduction of the Pacific slab, while the upgoing arrow indicates the secondary mantle flow containing partial melts. The thick horizontal bar at the top denotes the land area. The triangles indicate the active volcanoes. The inverted triangle denotes the Japan Trench. Horizontal solid arrows denote the compressional regime in NE Japan, while double blank arrows denote the bending and unbending in the subducting Pacific slab. Dipping lines in the outer rise part of the Pacific plate denote the normal faults. The dashed lines denote the Conrad and the Moho discontinuities and the upper and lower boundaries of the subducting Pacific slab, respectively.

base of the crust which results from the LPO of minerals such as mica and amphibole. The FVDs in the lower crust are complex under the volcanic front and are NE‐SW (nearly trench‐parallel) in the back‐arc area. [34] 3. The dominant FVDs for the slab earthquakes

under the back‐arc area are nearly E‐W (trench‐ normal) and the maximum DTs reach 0.30–0.32 s for some earthquakes deeper than 100 km. The trench‐normal FVDs are related to the corner flow in the mantle wedge driven by the subduction of the Pacific plate. For the upper plane slab earthquakes in the back‐arc area, only a small fraction (20%–30%) of the rays pass through the crust, while a large portion (>60%) of the rays pass through the mantle wedge. But the mantle wedge produces only about half of the total DTs. The apparent small DTs in the mantle wedge may reflect an isotropic or weak anisotropic zone in the middle of the mantle wedge. [35] 4. The splitting parameters for the slab earth-

quakes in the fore arc show dominant N‐S (trench‐ parallel) FVDs which are consistent with those for the crustal earthquakes. The shear wave splitting in the fore arc is mainly caused by the anisotropy in the crust, because the crust can account for 80% of the total DTs for the slab earthquakes.

[36] 5. The trench‐parallel FVDs in the subducting

Pacific slab may reflect the original fossil anisotropy when the Pacific plate was produced in the mid‐ocean ridge. The LPO of the crystals and cracks in the upper part of the subducting slab can also explain the trench‐parallel FVDs.

Acknowledgments [37] We thank the data centers of the Japan University Seismic Network, High‐Sensitivity Seismic Network (Hi‐net), and the Japan Meteorological Agency seismic network for providing the waveform data used in this study. We thank A. Wüstefeld and J. Verdon for their help on the SplitLab software. We are grateful to Thorsten Becker (the Editor) and two anonymous reviewers who provided constructive comments and suggestions. This work was supported partially by a grant (Kiban‐A 17204037) to D. Zhao from Japan Society for the Promotion of Science, by a grant (40634021) from the National Natural Science Foundation of China, and by the Scientific Research Foundation of Graduate School of Nanjing University. Figures 1–5, 7–14, and S5–S11 were made by using GMT [Wessel and Smith, 1998].

References Almeida, R., J. Chester, F. Chester, T. Waller, and D. Kirschner (2005), Lithology and structure of SAFOD Phase I core samples, Eos Trans. AGU, 86(52), Fall Meet. Suppl., T21A‐ 0454. 15 of 17


3

G


Ando, M., Y. Ishikawa, and H. Wada (1980), S‐wave anisotropy in the upper mantle under a volcanic area in Japan, Nature, 286, 43–46, doi:10.1038/286043a0. Audoine, E., K. Savage, and K. Gledhill (2004), Anisotropic structure under a back arc spreading region, the Taupo Volcanic Zone, New Zealand, J. Geophys. Res., 109, B11305, doi:10.1029/2003JB002932. Bird, P. (2003), An updated digital model of plate boundaries, Geochem. Geophys. Geosyst., 4(3), 1027, doi:10.1029/ 2001GC000252. Crampin, S., and S. Peacock (2008), A review of the current understanding of seismic shear‐wave splitting in the Earth’s crust and common fallacies in interpretation, Wave Motion, 45, 675–722, doi:10.1016/j.wavemoti.2008.01.003. Crampin, S., S. Peacock, Y. Gao, and S. Chastin (2004), The scatter of time‐delays in shear‐wave splitting above small earthquakes, Geophys. J. Int., 156, 39–44, doi:10.1111/ j.1365-246X.2004.02040.x. Eberhart‐Phillips, D., and M. Reyners (2009), Three‐dimensional distribution of seismic anisotropy in the Hikurangi subduction zone beneath the central North Island, New Zealand, J. Geophys. Res., 114, B06301, doi:10.1029/ 2008JB005947. Faccenda, M., L. Burlini, T. Gerya, and D. Mainprice (2008), Fault‐induced seismic anisotropy by hydration in subducting oceanic plates, Nature, 455, 1097–1100, doi:10.1038/ nature07376. Francis, T. (1969), Generation of seismic anisotropy in the upper mantle along the mid‐ocean ridges, Nature, 221, 162–165, doi:10.1038/221162b0. Fukao, Y. (1984), Evidence from core‐reflected shear wave for anisotropy in the Earth’s mantle, Nature, 309, 695–698, doi:10.1038/309695a0. Furukawa, Y. (1993), Depth of decoupling plate interface and thermal structure under arcs, J. Geophys. Res., 98, 20,005– 20,013, doi:10.1029/93JB02020. Greve, S., and M. Savage (2009), Modelling seismic anisotropy variations across the Hikurangi subduction margin, New Zealand, Earth Planet. Sci. Lett., 285, 16–26, doi:10.1016/ j.epsl.2009.05.035. Hasegawa, A., and A. Yamamoto (1994), Deep, low‐frequency microearthquakes in or around seismic low‐velocity zones beneath active volcanoes in northeastern Japan, Tectonophysics, 233, 233–252, doi:10.1016/0040-1951(94)90243-7. Hasegawa, A., N. Umino, and A. Takagi (1978), Double‐ planed structure of the deep seismic zone in the northeastern Japan arc, Tectonophysics, 47, 43–58, doi:10.1016/00401951(78)90150-6. Hasegawa, A., J. Nakajima, N. Uchida, T. Okada, D. Zhao, T. Matsuzawa, and N. Umino (2009), Plate subduction, and generation of earthquakes and magmas in Japan as inferred from seismic observations: An overview, Gondwana Res., 16, 370–400, doi:10.1016/j.gr.2009.03.007. Healy, D., S. M. Reddy, N. E. Timms, E. M. Gray, and A. V. Brovarone (2009), Trench‐parallel fast axes of seismic anisotropy due to fluid‐filled cracks in subducting slabs, Earth Planet. Sci. Lett., 283, 75–86. Hess, H. (1964), Seismic anisotropy of the uppermost mantle under oceans, Nature, 203, 629–631, doi:10.1038/203629a0. Huang, Z., D. Zhao, and L. Wang (2011), Seismic heterogeneity and anisotropy of the Honshu arc from the Japan trench to the Japan Sea, Geophys. J. Int., in press. Ishise, M., and H. Oda (2005), Three‐dimensional structure of P wave anisotropy beneath the Tohoku district, northeast

10.1029/2010GC003343

Japan, J. Geophys. Res., 110, B07304, doi:10.1029/ 2004JB003599. Iwamori, H., and D. Zhao (2000), Melting and seismic structure beneath the northeast Japan arc, Geophys. Res. Lett., 27, 425–428, doi:10.1029/1999GL010917. Kaneshima, S. (1990), Origin of crustal anisotropy: Shear wave splitting studies in Japan, J. Geophys. Res., 95, 11,121–11,133, doi:10.1029/JB095iB07p11121. Karato, S., H. Jung, I. Katayama, and P. Skemer (2008), Geodynamic significance of seismic anisotropy of the upper mantle: New insights from laboratory studies, Annu. Rev. Earth Planet. Sci., 36, 59–95, doi:10.1146/annurev.earth.36. 031207.124120. Katayama, I. (2009), Thin anisotropic layer in the mantle wedge beneath northeast Japan, Geology, 37, 211–214, doi:10.1130/G25413A.1. Kobayashi, K., M. Nakanishi, K. Tamaki, and Y. Ogawa (1998), Outer slope faulting associated with the western Kuril and Japan trenches, Geophys. J. Int., 134, 356–372, doi:10.1046/j.1365-246x.1998.00569.x. Leary, P. C., S. Crampin, and T. McEvilly (1990), Seismic fracture anisotropy in the Earth’s crust: An overview, J. Geophys. Res., 95, 11,105–11,114, doi:10.1029/JB095iB07p11105. Liu, Y., S. Crampin, and I. Main (1997), Shear‐wave anisotropy: Spatial and temporal variations in time delays at Parkfield, Central California, Geophys. J. Int., 130, 771–785, doi:10.1111/j.1365-246X.1997.tb01872.x. Liu, Y., H. Zhang, C. Thurber, and S. Roecker (2008), Shear wave anisotropy in the crust around the San Andreas fault near Parkfield: Spatial and temporal analysis, Geophys. J. Int., 172, 957–970, doi:10.1111/j.1365-246X.2007.03618.x. Long, M. D., and P. G. Silver (2008), The subduction zone flow field from seismic anisotropy: A global view, Science, 319, 315–318, doi:10.1126/science.1150809. Long, M. D., and P. G. Silver (2009), Shear wave splitting and mantle anisotropy: Measurements, interpretations, and new directions, Surv. Geophys., 30, 407–461, doi:10.1007/ s10712-009-9075-1. Mahan, K. (2006), Retrograde mica in deep crustal granulites: Implications for crustal seismic anisotropy, Geophys. Res. Lett., 33, L24301, doi:10.1029/2006GL028130. Mainprice, D. (2007), Seismic anisotropy of the deep Earth from a mineral and rock physics perspective, in Treatise on Geophysics, vol. 2, Mineral Physics, edited by G. Schubert, pp. 437–491, doi:10.1016/B978-044452748-6.00045-6, Elsevier, Oxford, U. K. Marson‐Pidgeon, K., and M. Savage (1997), Frequency‐ dependent anisotropy in Wellington, New Zealand, Geophys. Res. Lett., 24, 3297–3300, doi:10.1029/97GL03274. Masson, D. (1991), Fault patterns at outer trench walls, Mar. Geophys. Res., 13, 209–225, doi:10.1007/BF00369150. Matsuzawa, T., N. Umino, A. Hasegawa, and A. Takagi (1986), Upper mantle velocity structure estimated from PS‐converted wave beneath the northeastern Japan arc, Geophys. J. R. Astron. Soc., 86, 767–787. Matsuzawa, T., T. Kono, A. Hasegawa, and A. Takagi (1990), Subducting plate boundary beneath northeastern Japan arc estimated from SP converted waves, Tectonophysics, 181, 123–133, doi:10.1016/0040-1951(90)90012-W. Michibayashi, K., N. Abe, A. Okamoto, T. Satsukawa, and K. Michikura (2006), Seismic anisotropy in the upper most mantle, back‐arc region of the northeast Japan arc: Petrophysical analyses of Ichinomegata peridotite xenoliths, Geophys. Res. Lett., 33, L10312, doi:10.1029/2006GL025812.

16 of 17


3

G


Michibayashi, K., T. Oohara, T. Satsukawa, S. Ishimaru, S. Arai, and V. Okrugin (2009), Rock seismic anisotropy of the low‐velocity zone beneath the volcanic front in the mantle wedge, Geophys. Res. Lett., 36, L12305, doi:10.1029/ 2009GL038527. Moschetti, M. P., M. H. Ritzwoller, F. Lin, and Y. Yang (2010), Seismic evidence for widespread western‐US deep‐ crustal deformation caused by extension, Nature, 464, 885–889, doi:10.1038/nature08951. Nakajima, J., and A. Hasegawa (2004), Shear‐wave polarization anisotropy and subduction‐induced flow in the mantle wedge of northeastern Japan, Earth Planet. Sci. Lett., 225, 365–377, doi:10.1016/j.epsl.2004.06.011. Nakajima, J., Y. Tsuji, and A. Hasegawa (2009), Seismic evidence for thermally controlled dehydration reaction in subducting oceanic crust, Geophys. Res. Lett., 36, L03303, doi:10.1029/2008GL036865. Nicolas, A., and N. Christensen (1987), Formation of anisotropy in upper mantle peridotites—A review, in Composition, Structure and Dynamics of the Lithosphere‐Asthenosphere System, Geodyn. Ser., vol. 16, edited by K. Fuchs and C. Froideveaux, pp. 407–433, AGU, Washington, D. C. Nuttli, O. (1961), The effect of the Earth’s surface on the S wave particle motion, Bull. Seismol. Soc. Am., 51, 237–246. Okada, T., T. Matsuzawa, and A. Hasegawa (1995), Shear‐ wave polarization anisotropy beneath the north‐eastern part of Honshu, Japan, Geophys. J. Int., 123, 781–797, doi:10.1111/ j.1365-246X.1995.tb06890.x. Saltzer, R., J. Gaherty, and T. Jordan (2000), How are vertical shear wave splitting measurements affected by variations in the orientation of azimuthal anisotropy with depth?, Geophys. J. Int., 141, 374–390, doi:10.1046/j.1365246x.2000.00088.x. Seno, T. (1999), Syntheses of the regional stress fields of the Japanese islands, Isl. Arc, 8, 66–79, doi:10.1046/j.14401738.1999.00225.x. Silver, P. G., and W. W. Chan (1988), Implications for continental structure and evolution from seismic anisotropy, Nature, 335, 34–39, doi:10.1038/335034a0. Silver, P. G., and W. W. Chan (1991), Shear wave splitting and subcontinental mantle deformation, J. Geophys. Res., 96, 16,429–16,454, doi:10.1029/91JB00899. Tsumura, N., S. Matsumoto, S. Horiuchi, and A. Hasegawa (2000), Three‐dimensional attenuation structure beneath the northeastern Japan arc estimated from spectra of small earthquakes, Tectonophysics, 319, 241–260, doi:10.1016/ S0040-1951(99)00297-8. Umino, N., and A. Hasegawa (1975), On the two‐layered structure of deep seismic plane in northeastern Japan arc (in Japanese), J. Seismol. Soc. Jpn., 27, 125–139. Verdon, J. P. (2010), Passive seismic monitoring and geomechanical modelling of CO2 storage in subsurface reservoirs, Ph.D. thesis, Dep. of Earth Sci., Univ. of Bristol, Bristol, U. K.

10.1029/2010GC003343

Vidale, J. E. (1986), Complex polarization analysis of particle motion, Bull. Seismol. Soc. Am., 71, 1511–1530. Wang, J., and D. Zhao (2008), P wave anisotropic tomography beneath northeast Japan, Phys. Earth Planet. Inter., 170, 115–133, doi:10.1016/j.pepi.2008.07.042. Wang, J., and D. Zhao (2010), Mapping P wave anisotropy of the Honshu arc from Japan Trench to the back‐arc, J. Asian Earth Sci., 39, 396–407, doi:10.1016/j.jseaes.2010.04.009. Weiss, T., S. Siegesmund, W. Rabbel, T. Bohlen, and M. Pohl (1999), Seismic velocities and anisotropy of the lower continental crust: A review, Pure Appl. Geophys., 156, 97–122, doi:10.1007/s000240050291. Wessel, P., and W. Smith (1998), New improved version of the Generic Mapping Tools released, Eos Trans. AGU, 79, 579, doi:10.1029/98EO00426. Wiens, D., J. Conder, and U. Faul (2008), The seismic structure and dynamics of the mantle wedge, Annu. Rev. Earth Planet. Sci., 36, 421–455, doi:10.1146/annurev.earth.33. 092203.122633. Wirth, E., and M. Long (2010), Frequency‐dependent shear wave splitting beneath the Japan and Izu‐Bonin subduction zones, Phys. Earth Planet. Inter., 181, 141–154, doi:10.1016/ j.pepi.2010.05.006. Wüstefeld, A., and G. Bokelmann (2007), Null detection in shear‐wave splitting measurements, Bull. Seismol. Soc. Am., 97, 1204–1211, doi:10.1785/0120060190. Wüstefeld, A., G. Bokelmann, G. Barruol, and C. Zaroli (2008), Splitlab: A shear‐wave splitting environment in Matlab, Comput. Geosci., 34, 515–528, doi:10.1016/j.cageo. 2007.08.002. Zhao, D. (2009), Multiscale seismic tomography and mantle dynamics, Gondwana Res., 15, 297–323, doi:10.1016/j. gr.2008.07.003. Zhao, D., A. Hasegawa, and S. Horiuchi (1992), Tomographic imaging of P and S wave velocity structure beneath northeastern Japan, J. Geophys. Res., 97, 19,909–19,928, doi:10.1029/92JB00603. Zhao, D., T. Matsuzawa, and A. Hasegawa (1997), Morphology of the subducting slab boundary in the northeastern Japan arc, Phys. Earth Planet. Inter., 102, 89–104, doi:10.1016/S0031-9201(96)03258-X. Zhao, D., Z. Wang, N. Umino, and A. Hasegawa (2009), Mapping the mantle wedge and interplate thrust zone of the northeast Japan arc, Tectonophysics, 467, 89–106, doi:10.1016/j.tecto.2008.12.017. Zinke, J., and M. Zoback (2000), Structure‐related and stress‐ induced shear‐wave velocity anisotropy: Observations from microearthquakes near the Calaveras fault in central California, Bull. Seismol. Soc. Am., 90, 1305–1312, doi:10.1785/ 0119990099.

17 of 17