G

P

Sm

Ya

b

c

a

ARRA

E

D

KSSMRE

1

aitas1tte(a(gz

f

Af

0d

ARTICLE IN PRESSModel

EPI-5268; No. of Pages 8

Physics of the Earth and Planetary Interiors xxx (2010) xxx–xxx

Contents lists available at ScienceDirect

Physics of the Earth and Planetary Interiors

journa l homepage: www.e lsev ier .com/ locate /pepi

eismic discontinuities in the mantle transition zone and at the top of the lowerantle beneath eastern China and Korea: Influence of the stagnant Pacific slab

uan Gaoa,b,∗, Daisuke Suetsugub, Yoshio Fukaob, Masayuki Obayashib, Yutao Shia, Ruifeng Liuc

Institute of Earthquake Science, China Earthquake Administration, Beijing 100036, ChinaInstitute For Research on Earth Evolution, Japan Agency of Marine-Earth Science and Technology, 2-15 Natsushima-cho, Yokosuka, Kanagawa 237-0061, JapanChina Earthquake Network Center, China Earthquake Administration, Beijing 100045, China

r t i c l e i n f o

rticle history:eceived 17 July 2009eceived in revised form 3 February 2010ccepted 18 March 2010

dited by: D. Suetsugu, C. Bina, T. Inoue and

a b s t r a c t

We have analyzed broadband data to identify and determine mantle discontinuities in the mantle tran-sition zone (MTZ) beneath eastern China, where the Pacific slab is stagnant. The depths of the 410 and660 km discontinuities are generally shallower and deeper, respectively, than the global averages inand near the Pacific slab beneath eastern China. The MTZ is thicker in the slab than the global average.This observation is consistent with the thermally controlled olivine to wadsleyite transformation for the410 km discontinuity and the post-spinel transformation for the 660 km discontinuity. Other disconti-

. Wiens.

eywords:eismic discontinuitiestagnant Pacific slabantle transition zone

eceiver functions

nuities appear from 690 to 750 km in the lower mantle part of the Pacific slab. Recent mineralogicalexperiments indicate that the most plausible interpretation of these deep discontinuities is that theyrepresent the ilmenite to perovskite transformation in a cold environment, such as that in the Pacificslab.

© 2010 Elsevier B.V. All rights reserved.

astern China and Korea

. Introduction

The two main global discontinuities in the mantle are locatedt depths of around 410 and 660 km (called the “410” and “660”n this paper), though their depths vary slightly in different tec-onic zones (e.g., Flanagan and Shearer, 1998). The “410” and “660”re considered to be due to the olivine to wadsleyite and post-pinel phase transformations, respectively (e.g., Ito and Takahashi,989). The pressure of both transformations is thermally con-rolled because the olivine to wadsleyite and post-spinel phaseransformations are exothermic (positive Clapeyron slope) andndothermic (negative Clapeyron slope) reactions, respectivelye.g., Bina and Helffrich, 1994). In a cold (hot) environment suchs a subduction zone, the “410” and “660” should be elevateddepressed) and depressed (elevated), respectively, which should

Please cite this article in press as: Gao, Y., et al., Seismic discontinuities in teastern China and Korea: Influence of the stagnant Pacific slab. Phys. Earth

enerate temperature-related topography in the mantle transitionone (MTZ).

However, recent mineralogical studies suggest that phase trans-ormations related to garnet, the other major component of the

∗ Corresponding author at: Institute of Earthquake Science, China Earthquakedministration, 63 Fuxing Road, Beijing 100036, China. Tel.: +86 10 88015660;

ax: +86 10 88015627.E-mail address: gaoyuan@seis.ac.cn (Y. Gao).

031-9201/$ – see front matter © 2010 Elsevier B.V. All rights reserved.oi:10.1016/j.pepi.2010.03.009

mantle, are involved, in addition to the olivine-related post-spineltransformation, at depths from 600 to 750 km (e.g., Vacher et al.,1998), and should therefore appear as multiple seismic discon-tinuities in that depth range. Recent seismological studies havedetected such multiple discontinuities (Simmons and Gurrola,2000; Deuss et al., 2006; Tibi et al., 2007; Schmerr and Garnero,2007; Andrews and Deuss, 2008). For subduction zones in par-ticular, the relationship between subducted slabs and multiplephase transformations has been investigated in order to understandthe fate of subducted slabs. Niu and Kawakatsu (1996) revealed acomplicated 660 km discontinuity at one seismic station (MDJ) innortheastern China, and from a small array of data, Ai et al. (2003)found multiple discontinuities near the “660” in the same region,where the Pacific slab is imaged as high velocity anomalies thatlay horizontally (called a stagnant slab). Chen and Ai (2009) alsofound multiple discontinuities near the “660” beneath the south-eastern part of North China Craton, where high velocity anomaliesare located. The stagnant slab is, however, spreading in the MTZbeneath a wide area of eastern China, where multiple disconti-nuities have never been reliably reported except in northeastern

he mantle transition zone and at the top of the lower mantle beneathPlanet. In. (2010), doi:10.1016/j.pepi.2010.03.009

China. Among other previous studies, Shen et al. (2008) was theonly systematic receiver function study to determine the “410”and “660” depths beneath the entire Chinese region. Their resultshowed a thick MTZ beneath eastern China, which is a signatureof the cold MTZ around the stagnant slab. However, they did not

ARTICLE IN PRESSG Model

PEPI-5268; No. of Pages 8

2 Y. Gao et al. / Physics of the Earth and Planetary Interiors xxx (2010) xxx–xxx

F NC), Is

o““

sCOvieac

2

e

TD

ig. 1. (a) Seismograph stations of the National Seismograph Network of China (NShown by dots. (b) Events analyzed in the present study.

btain clear signals due to the multiple discontinuities below the660,” while they suggested a thin low-velocity layer above the660.”

The aim of the present study is to investigate the complextructure of the MTZ that underlies the entire region of easternhina region using receiver function analysis (e.g., Langston, 1997;wens and Crosson, 1988). We analyzed shorter-period Ps con-erted waves at a larger number of permanent broadband stationsn eastern China and the Korean peninsula than were used by Shent al. (2008), which enabled us to detect multiple discontinuitiesnd to study the interaction of the stagnant slab and mantle dis-ontinuities.

Please cite this article in press as: Gao, Y., et al., Seismic discontinuities in teastern China and Korea: Influence of the stagnant Pacific slab. Phys. Earth

. Data and methods

We used all the permanent broadband stations in the east-rn Chinese region, which includes northeast China, north China,

able 1epths of the “410,” “660,” and “720” discontinuities and MTZ thickness.

Station “410” depth (km) “660” depth (km) M

BJT 421 ± 3 686 ± 3 26BNX 404 ± 4 680 ± 7 27CN2 408 ± 4 668 ± 2 26CNS 411 ± 11 671 ± 6 26DL2 – 656 ± 4HEH 410 ± 5 662 ± 8 25HHC 423 ± 1 666 ± 10 24HIA 415 ± 1 667 ± 2 25HNS 422 ± 1 675 ± 1 25INCN – 658 ± 5LYN 413 ± 10 678 ± 2 26MDJ 427 ± 9 664 ± 12 23NJ2 417 ± 5 677 ± 2 26NNC 413 ± 1 671 ± 6 25QZH – 672 ± 9SNY – 668 ± 11SSE 406 ± 3 659 ± 8 25TIA 417 ± 2 672 ± 10 25TJN 408 ± 3 669 ± 10 26WZH 418 ± 3 669 ± 5 24XLT 413 ± 3 673 ± 2 26

RIS/CDSN stations, and an OHP station used in this study. P660s piercing points are

and Korea. We analyzed data from the National SeismographNetwork of China (NSNC), the Chinese Digital Seismic Network(CDSN), the Incorporated Research Institution for Seismology(IRIS) of the United States of America, and the Ocean Hemi-sphere Project Network (OHP) of Japan. We used a total of 21stations in the research for this paper (Fig. 1a): 15 NSNC sta-tions, 4 IRIS/CDSN stations, 1 IRIS station, and 1 OHP station inKorea. The stations were spread over all of eastern China andthe Korean peninsula, where the stagnant slab is stalling in theMTZ. We selected earthquakes that took place between 2004and 2006, whose magnitude ranged from 6.0 to 7.5 and whoseepicentral distance ranged from 30◦ to 90◦ (Fig. 1b). Before the

he mantle transition zone and at the top of the lower mantle beneathPlanet. In. (2010), doi:10.1016/j.pepi.2010.03.009

receiver function analysis was conducted, we visually inspected allseismograms and discarded data with low signal-to-noise ratios.Finally, more than 30 records were obtained from each station(Table 1). A band-pass filter with corner periods from 1 to 50 swas applied to the selected seismograms to suppress noise, partic-

TZ thickness (km) “720” depth (km) Numbers of data

5 ± 3 – 886 ± 8 – 400 ± 4 – 520 ± 14 – 66– – 691 ± 10 – 754 ± 10 – 1003 ± 3 – 843 ± 2 721 ± 9 48– – 585 ± 11 – 1077 ± 15 700 ± 20 820 ± 6 748 ± 10 558 ± 6 698 ± 9 39– – 49– 698 ± 9 563 ± 8 – 465 ± 10 719 ± 7 402 ± 10 – 512 ± 6 – 370 ± 4 719 ± 8 51

ARTICLE IN PRESSG Model

PEPI-5268; No. of Pages 8

Y. Gao et al. / Physics of the Earth and Planetary Interiors xxx (2010) xxx–xxx 3

Fa

ur

meirs

Fig. 4. All stacked and depth-migrated receiver functions obtained by this study.

Fsai

ig. 2. Receiver functions at HIA are shown as examples. Predicted arrival times arelso shown.

larly the long-period noise observed in the horizontal-componentecords.

In this study, we employed the velocity spectrum stacking (VSS)ethod (Gurrola et al., 1994). To equalize the source effect and

Please cite this article in press as: Gao, Y., et al., Seismic discontinuities in the mantle transition zone and at the top of the lower mantle beneatheastern China and Korea: Influence of the stagnant Pacific slab. Phys. Earth Planet. In. (2010), doi:10.1016/j.pepi.2010.03.009

xtract Ps phases, we computed the receiver functions by perform-ng a frequency-domain deconvolution of a vertical-componentecord from a radial-component record for each event at eachtation. All receiver functions were plotted for each station for

A band-pass filter between 4 and 50 s was used in computing receiver functions.Dashed curves indicate a 95% confidence level. We identified the “720” signals (dots)as later phases to P660s with amplitudes above a 95% confidence level. We did notchoose seismograms showing a continuous oscillation around the MTZ (such asWZH, HHC, and HEH) even when they showed apparent later arrivals.

ig. 3. VSS plot computed from all receiver functions at station HIA. (Top) Amplitudes of stacked receiver functions normalized to 20% of the direct P-wave amplitudes andhown with respect to Ps conversion depths and the average shear wave velocity perturbations above the conversion depths. Two plus signs indicate peaks of the “410”nd “660” obtained using velocities taken from Ritsema and van Heijst (2000). (Bottom) Stacked amplitudes versus depth curve for zero velocity correction. Two plus signsndicate converted signals from the “410” and “660.” The dashed curves indicate a 95% confidence level computed by the bootstrap method.

ARTICLE IN PRESSG Model

PEPI-5268; No. of Pages 8

4 Y. Gao et al. / Physics of the Earth and Planetary Interiors xxx (2010) xxx–xxx

Fig. 5. Depths of mantle discontinuities. Deviations from the global averages (418 and 660 km for the “410” and “660”, respectively, by Flanagan and Shearer, 1998) ares lloweP 001) aa yron st

cvTPedcifgwsssTac1vCg

hown. (a) “410”, (b) “660”, and (c) MTZ thickness. Black and white circles denote shas pierce points for each station. P-velocity tomograms at each depth (Fukao et al., 2nomalies is shown in (d). Straight lines in (d) indicate lines predicted from Claperansformation.

omparison with the theoretical Ps time curves expected for con-ersion depths of “410” and “660”; an example is shown in Fig. 2.he peak alignments of the receiver functions near the expected410s and P660s curves are obvious, though they are not largenough to analyze individually. To constrain the “410” and “660”epths, we employed the VSS method for stacking with moveoutorrection. In this process, the time axis of the receiver functionss converted to a depth axis. The moveout correction is computedor 1 km grids of discontinuity depths from 0 to 800 km and 0.1%rids of shear wave velocity perturbations from −2.5% to +2.5%,ith respect to the Iasp91 model (Kennett and Engdahl, 1991). In

tacking receiver functions, a clear enhanced Ps-converted signalhould be obtained with a moveout computed at the true conver-ion depth and true average shear velocity perturbation (Fig. 3).here is a strong trade-off between the discontinuity depth and theverage velocity perturbation above the discontinuity, which thus

Please cite this article in press as: Gao, Y., et al., Seismic discontinuities in teastern China and Korea: Influence of the stagnant Pacific slab. Phys. Earth

annot be determined independently (e.g., Gurrola and Minster,998). We therefore relied upon existing three-dimensional shearelocity models to determine the discontinuity depths. We usedRUST2.0 (Bassin et al., 2000) for the crustal structure and thelobal shear velocity model of Ritsema and van Heijst (2000) below

r and deeper discontinuities, respectively. Circles are plotted at average positions ofre shown in the background. The correlation between MTZ thickness and P-velocitylopes of −0.5 MPa/K, −1.0 MPa/K, −2.0 MPa/K, and −3.0 MPa/K for the post-spinel

the Moho. The average shear velocity perturbations above the dis-continuities beneath the stations used in the present study rangedfrom −0.1% to +1.0% with respect to the Iasp91 model, and thesevalues were used to determine the discontinuity depths. The aver-age S-velocity anomalies of +1.0% above 410 and 660 km increasedour estimates of the depths of “410” and “660” by +5 and +8 km,respectively. The discontinuity depths and their confidence levelswere computed using the bootstrap method (Efron and Tibshirani,1986), which randomly resamples (500 times) the receiver func-tions prior to stacking. See Gurrola et al. (1994) and Suetsugu et al.(2004, 2005) for more details.

3. Results

3.1. Depths of the “410” and “660”

he mantle transition zone and at the top of the lower mantle beneathPlanet. In. (2010), doi:10.1016/j.pepi.2010.03.009

First, we obtained the discontinuity depths by stacking thelong-period receiver functions at each station. The functions wereobtained using a Gaussian low-pass filter with a corner period of 4 s,which is comparable to the period range used by Shen et al. (2008).The stacked and depth-converted receiver functions are shown for

IN PRESSG

P

d Planetary Interiors xxx (2010) xxx–xxx 5

eep96acae(

vo“atvMfMpP(TagstPlsttdtasrMg

3

buialb(ndss(PoeHlwsses

by a long tail, which may also be the arrival of the “720”, though theseparate signals are not clear. Fig. 7 shows the locations of the “720”conversion, from which we can observe a correlation between theconversion locations and the high-velocity anomalies at the top ofthe lower mantle that may represent the bottom part of the stag-

ARTICLEModel

EPI-5268; No. of Pages 8

Y. Gao et al. / Physics of the Earth an

ach station (Fig. 4). The depths of the “410” and “660” under east-rn China and Korea are presented in Fig. 5a and b, respectively. Welotted the depths obtained from the converted signals above the5% confidence level. Deviations from the global average (418 and60 km for the “410” and “660”, respectively, according to Flanagannd Shearer, 1998) were plotted at the average locations of theonversion points for each station on P-velocity tomographic mapst their respective depths. The degree of uncertainty of the depthstimates was mostly less than 10 km for both the “410” and “660”Table 1).

The “410” is generally elevated by 10–20 km where fast P-elocity anomalies were found and is within ±10 km of the averageutside the high-velocity area (Fig. 5a), which suggests that the410” depths in the studied region are controlled mostly by thermalnomalies associated with the subducted Pacific slab. An exceptiono this is MDJ, where the “410” is depressed by 9 km in the highelocity area. The P410 s signal is broadened or double-peaked atDJ, which makes the depth estimate more uncertain than that

or other stations. Shen et al. (2008) did not obtain “410” depths atDJ, perhaps because the P410s signal is too weak (Fig. 2 in their

aper). The “660” is generally depressed by 10–20 km in the high-velocity anomalies (Fig. 5b), which is consistent with Shen et al.2008). Stations DL2 and INCN show nearly normal “660” depths.hese stations are located in areas where high P-velocity anomaliesre less pronounced [(35◦N, 120◦E) and (35◦N, 125◦E)], which sug-ests that the stagnant slab may have a gap and is not a continuoustructure. The station BJT gives a depth of 686 km (26 km greaterhan the global average) for the “660”, despite the fact that thes conversion points are located near the edge of the fast anoma-ies. Although the deep “660” at BJT has been reported in previoustudies (Niu and Kawakatsu, 1998; Shen et al., 2008), its interpre-ation remains unclear. The MTZ thickness (the “660” depth minushe “410” depth) is presented in Fig. 5c, and was more accuratelyetermined than the “410” and “660” since an estimate of the MTZhickness is less affected by possible errors in velocity correctionbove the “410.” P-velocity anomalies averaged in the MTZ arehown in the background of Fig. 5c, and in most of the studiedegion they are faster than the globally averaged MTZ velocity. TheTZ thickness is also thicker than the global average, which is in

ood agreement with the results of Shen et al. (2008).

.2. Discontinuities below the “660”

As shown in Fig. 4, the “660” peaks and their codas are oftenroadened or composed of multiple peaks, the first of which aresed in the “660” depth plot. We identified significant signals arriv-

ng after the “660” (indicated by dots in Fig. 4) as those withmplitudes above the 95% confidence interval. We did not considerater arrivals that are part of a long oscillation around the “660” toe significant, since they may represent a computational artifactringing) due to deconvolution. To see whether the broadened sig-als were composed of multiple signals, we computed the stackedepth-converted receiver functions with a short period filter (Gaus-ian low-pass filter with a corner period of 2 s). We employed theame shear velocity perturbations as those in the long-period stack4 s) for the velocity corrections. The velocity perturbations in the660s Fresnel zones (about 50 and 100 km in diameter at periodsf 2 and 4 s, respectively) are almost the same, since the short-st wavelength of the shear velocity model of Ritsema and vaneijst (2000) used for velocity correction is 2000 km, which is much

onger than the Fresnel zones. In the short-period stack (Fig. 6)

Please cite this article in press as: Gao, Y., et al., Seismic discontinuities in teastern China and Korea: Influence of the stagnant Pacific slab. Phys. Earth

e can see more isolated peaks at depths of 690–750 km than arehown in the long-period stack (Fig. 4). The long tails after the “660”ignal at SNY and NNC in the long-period stack (Fig. 4) are consid-red to be a combination of the “660” signal and a later wave, ashown in Fig. 6. Depths corresponding to isolated signals, which are

Fig. 6. Stacked and depth-migrated receiver functions of stations where “720” sig-nals were observed. A band-pass filter between 2 and 50 s was used in computingreceiver functions. Dashed curves indicate a 95% confidence level.

listed in Table 1, were determined by the bootstrap method. We callthe discontinuity that generates the later wave the “720,” accordingto its average conversion depth. At MDJ, the “660” is accompanied

he mantle transition zone and at the top of the lower mantle beneathPlanet. In. (2010), doi:10.1016/j.pepi.2010.03.009

Fig. 7. Stations where “720” signals were observed are indicated by red triangles;those with no “720” signals are indicated by blue triangles. A P-velocity tomogramat depths of 700–800 km is shown in the background. P660s piercing points areindicated by crosses. (For interpretation of the references to color in this figurelegend, the reader is referred to the web version of the article.)

ARTICLE IN PRESSG Model

PEPI-5268; No. of Pages 8

6 Y. Gao et al. / Physics of the Earth and Planetary Interiors xxx (2010) xxx–xxx

F d “72o sion o

naa“thtetii“

4

PcvptppfC2tt12s−b

ig. 8. Cross sections of P-velocity tomogram and the “410” (green), “660” (red), anf the references to color in this figure legend, the reader is referred to the web ver

ant Pacific slab. The discontinuities located by the present studyre plotted on vertical cross sections in Fig. 8. The “660” and “410”re detected either inside or outside of the high-velocity slab. The660” depths are deeper inside the high-velocity slab than outsidehe slab. On the other hand, the “720” is located only inside theigh-velocity slab, which implies that the “720” is closely relatedo the subducted slab. The “720” is not visible in the stacks of Shent al. (2008). There are differences between the present study andhat of Shen et al. (2008), e.g., the receiver function methods usedn the present study and the selection of data and frequency, but its not clear why we arrived at different degrees of visibility of the720”.

. Discussion

As shown in Fig. 5, a spatial correlation exists between the-velocity tomograms and the discontinuity depths. To see thisorrelation quantitatively, we plotted MTZ thickness versus P-elocities in the MTZ, as shown in Fig. 5d. We can observe aositive trend (thicker MTZ at a higher P-velocity) with a correla-ion coefficient of 0.72. We compared the observed trend with thatredicted from recent mineralogical results. Assuming the tem-erature dependence of the P-velocity to be −3.0 × 10−4 km/s/Kor minerals in the MTZ (Irifune et al., 2008) and assuming thelapeyron slope of the olivine to wadsleyite transformation to be.5 MPa/K (Katsura and Ito, 1989), we varied the Clapeyron slope ofhe post-spinel transformation from −0.5 to −3.0 MPa/K accordingo the experimental data (Ito and Takahashi, 1989; Irifune et al.,

Please cite this article in press as: Gao, Y., et al., Seismic discontinuities in teastern China and Korea: Influence of the stagnant Pacific slab. Phys. Earth

998; Akaogi and Ito, 1993; Bina and Helffrich, 1994; Katsura et al.,003; Fei et al., 2004; Litasov et al., 2005). The observed trend is con-istent with those expected from the Clapeyron slope gentler than2.0 MPa/K for the post-spinel transformation, which was obtainedy experiments under dry conditions (Katsura et al., 2003; Fei et al.,

0” (yellow) discontinuities along two great circles shown in (c). (For interpretationf the article.)

2004). Fukao et al. (2009) ascertained the Clapeyron slope beneaththe Philippine Sea region to be −2.5 MPa/K from P-velocity pertur-bation and the “660” depth in the lower part of the stagnant Pacificslab (Obayashi et al., 2006; Niu et al., 2005), a value which is slightlysteeper than the slope of −0.5 to −2.0 MPa/K in the present study.This difference in slope may indicate that the MTZ beneath east-ern China is relatively dry compared to that beneath the PhilippineSea, since the Clapeyron slope of the post-spinel transformation isgentler in a dry condition than it is in a wet condition (Litasov etal., 2005). The kinetic (non-equilibrium) effect could also be signif-icant in a cold and dry condition (e.g., Wang et al., 1997; Kubo etal., 2002), a factor which is not addressed in the present study butwhich will be the subject of future studies.

Shen et al. (2008) observed negative Ps signals at a depth ofaround 600 km beneath many Chinese stations, which would rep-resent a downward velocity reduction. Shen and Blum (2003) alsofound negative signals from 570 to 600 km beneath southern Africa.They suggest that these signals originate from a low-velocity zoneat the base of the MTZ, which they attribute to the accumulatedbasaltic layer in the MTZ. We also see negative signals in the stackedreceiver functions at some stations. They are distributed in a widerange of depths from 450 to 650 km and at stations located bothinside and outside the Pacific slab. We cannot conclude that theyare actual Ps signals from a low-velocity zone in the MTZ becausethey do not have spatial coherence. These negative signals may bedue to reverberation phases from shallower discontinuities or maysimply be a computational artifact of the deconvolution process. Toidentify the origin of these signals, further analysis is needed that

he mantle transition zone and at the top of the lower mantle beneathPlanet. In. (2010), doi:10.1016/j.pepi.2010.03.009

is based upon more data from the denser seismic networks thatare now being deployed. Our discussion will next focus upon thecomplex discontinuities around the “660”.

The complex discontinuities around the “660” have been exten-sively studied in the last decade. While most previous studies have

ING

P

d Plan

rzzadfafimKtderMfwaCsZT(nttafti(tmsiopiwrt

notVfentoib2tta1oaitiasM

ARTICLEModel

EPI-5268; No. of Pages 8

Y. Gao et al. / Physics of the Earth an

eported multiple discontinuities around the “660” in subductionones, some have reported similar multiples away from subductionones (e.g., Simmons and Gurrola, 2000; Deuss et al., 2006; Andrewsnd Deuss, 2008). Niu and Kawakatsu (1996) first found complexiscontinuities below the “660” beneath MDJ by using the receiverunction method. They determined discontinuities at depths of 740nd 780 km in addition to the “660” at a depth of 670 km. Our ownndings showed that the “660” signal at MDJ has a long tail, whichay correspond to the 740 km discontinuity reported by Niu and

awakatsu (1996). Ai et al. (2003) also employed the receiver func-ion method to study mantle discontinuities and found multipleiscontinuities at depths from 650 to 750 km beneath northeast-rn China (40–46◦N, 126–132◦E) using broadband array data. Theiresult is consistent with our observation of the “720” at SNY andDJ. Ai and Zheng (2003) analyzed dense broadband array data

rom Shandon Province in eastern China (36–37◦N, 117–120◦E)ith the receiver function method. They found a discontinuity at

round 700 km below the “660” in one area (34–36◦N, 116–122◦E).hen and Ai (2009) reported similar deep discontinuities in theame area, based upon more array data than was used by Ai andheng (2003). Their result is consistent with the “720” found atIA and HNS in the present study. Ai et al. (2003), Ai and Zheng2003), and Chen and Ai (2009) interpreted the multiple disconti-uities as a garnet-related phase transformation. A discontinuityhat is deeper below the “660” has been detected in other subduc-ion zones. Tibi et al. (2007) analyzed broadband data from islandnd ocean-bottom seismic stations in Mariana with the receiverunction method. They found a discontinuity at depths from 740o 770 km near the Mariana slab, which they attributed to thelmenite to perovskite phase transformation. Schmerr and Garnero2007) analyzed the precursors to SS waves in order to study man-le discontinuities in the South America subduction zone and found

ultiple discontinuities at around the “660” in the vicinity of theubducted Nazca plate. One possible interpretation of the high vis-bility of the “720” signals in the subduction zones is the presencef a Ps wave that was converted from a later P wave due to multi-athing caused by the subducted slabs. However, if the later P wave

s present, it should also produce a Ps converted wave at the “410”,hich should arrive after P410s, but this was not observed. For this

eason we conclude that the “720” signal represents the Ps waveshat were converted at the “720”.

The discontinuities below the “660” can be attributed to theon-olivine components of the slab material, since there is nolivine-related phase transformation below the “660” (except forhe post-perovskite transformation at the bottom of the mantle).acher et al. (1998) indicated that a garnet-related phase trans-

ormation should occur around the “660” in a low-temperaturenvironment (1000 K), such as is found in cold slabs. The gar-et to ilmenite phase transformation should take place from 608o 664 km for pyrolite composition, and the ilmenite to per-vskite transformation should take place from 709 to 731 km. Its more likely that this latter transformation would be detectedy the seismic waves used in the present study (wavelengths of0–40 km) than the former, because the ilmenite to perovskiteransformation is relatively sharp. In normal temperature condi-ions (1500–1600 K), a garnet-related transformation should occurt pressures close to the post-spinel transformation (Vacher et al.,998), which could not be isolated seismologically. The depthsf the discontinuities located by the present study (690–750 km)nd their confinement in the lower mantle part of the cold slabndicate that the most likely origin of the deep discontinuities is

Please cite this article in press as: Gao, Y., et al., Seismic discontinuities in teastern China and Korea: Influence of the stagnant Pacific slab. Phys. Earth

he ilmenite to perovskite transformation. The depression of thelmenite to perovskite transformation in the cold slab should causepositive buoyancy of the slab in addition to that due to the depres-ion of the “660,” which should contribute to slab stagnation in theTZ.

PRESSetary Interiors xxx (2010) xxx–xxx 7

5. Conclusions

We have estimated the depths of the mantle discontinuitiesfrom 400 to 750 km using broadband data from the entire regionof eastern China and Korea, which enabled us to study the rela-tionship between the discontinuity depths and the stagnant Pacificslab. Our results are summarized as follows:

(1) The depths of the “410” and “660” are generally shallower anddeeper, respectively, than the global averages in and near thePacific slab beneath eastern China. The MTZ is thicker in the slabthan the global average. These observations are consistent withthose reported by Shen et al. (2008). These observations areconsistent with the thermally controlled olivine to wadsleyitetransformation of the “410” and the post-spinel transformationof the “660.”

(2) There is another discontinuity at a depth from 690 to 750 km inthe lower mantle part of the Pacific slab. Recent mineralogicalexperiments indicate that the most plausible interpretation ofthis deep discontinuity is the ilmenite to perovskite transfor-mation.

(3) We observed negative signals between the “410” and “660”, aswas also reported in previous studies. The origin of this negativephase remains unknown. The deployment of the denser broad-band seismic network which is now being constructed shouldbe useful for making detailed topography maps of the disconti-nuities identified above and for examining the significance andorigin of these negative signals.

Acknowledgements

YG thanks the support from: China NSFC Project 40674021,China IES 2007-13, and a Japanese JSPS invitation fellowship project(2007). We thank two anonymous reviewers much to improve themanuscript.

References

Ai, Y., Zheng, T., 2003. The upper mantle discontinuity structure beneath easternChina. Geophys. Res. Lett. 30 (21), 2089, doi:10.1029/2003GL017678.

Ai, Y., Zheng, T., Xu, W., He, Y., Dong, D., 2003. A complex 660 km discontinuitybeneath northeast China. Earth Planet. Sci. Lett. 212, 63–71.

Akaogi, M., Ito, E., 1993. Refinement of enthalpy measurement of MgSiO3 perovskiteand negative pressure–temperature slopes for perovskite-forming reaction.Geophys. Res. Lett. 12, 1839–1842.

Andrews, J., Deuss, A., 2008. Detailed nature of the 660 km region of the mantle fromglobal receiver function data. J. Geophys. Res. 113, doi:10.1029/2007JB005111.

Bassin, C., Laske, G., Masters, G., 2000. The current limits of resolution for surfacewave tomography in North America. EOS Trans. AGU 81, F897.

Bina, C., Helffrich, G., 1994. Phase transition Clapeyron slopes and transition zoneseismic discontinuity topography. J. Geophys. Res. 99, 15853–15860.

Chen, L., Ai, Y., 2009. Discontinuity structure of the mantle transition zone beneaththe North China Craton from receiver function migration. J. Geophys. Res. 114,B06307, doi:10.1029/2008JB006221.

Deuss, A., Redfern, S.A., Chambers, K., Woodhouse, J.H., 2006. The nature of the 660-kilometer discontinuity in Earth’s mantle from global seismic observations ofPP precursors. Science 311, 198–201.

Efron, B., Tibshirani, R., 1986. Bootstrap methods for standard errors, confidenceintervals, and other measures of statistical accuracy. Stat. Sci. 1, 54–77.

Fei, Y., Van Orman, J., Li, J., van Westrenen, W., Sanloup, C., Minarik, W., Hirose, K.,Komabayashi, T., Walter, M., Funakoshi, K., 2004. Experimentally determinedpostspinel transformation boundary in Mg2SiO4 using MgO as an internal pres-sure standard and its geophysical implications. J. Geophys. Res. 109, B02305,doi:10.1029/2003JB002562.

Flanagan, M., Shearer, P., 1998. Global mapping of topography on transitionzone velocity discontinuities by stacking SS precursors. J. Geophys. Res. 103,2673–2692.

Fukao, Y., Widiyantoro, S., Obayashi, M., 2001. Stagnant slabs in the upper and lower

he mantle transition zone and at the top of the lower mantle beneathPlanet. In. (2010), doi:10.1016/j.pepi.2010.03.009

mantle transition region. Rev. Geophys. 39, 291–323.Fukao, Y., Obayashi, M., Nakakuki, T., the deep slab project group, 2009. Stagnant

slab: a review. Ann. Rev. Earth Planet. Sci. 37, 19–46.Gurrola, H., Minster, J.B., Owens, T., 1994. The use of velocity spectrum for stacking

receiver functions and imaging upper mantle discontinuities. Geophys. J. Int.117, 427–440.

ING

P

8 d Plan

G

I

I

I

K

K

K

K

L

L

N

N

ARTICLEModel

EPI-5268; No. of Pages 8

Y. Gao et al. / Physics of the Earth an

urrola, H., Minster, J.B., 1998. Thickness estimates of the upper mantle transitionzone from bootstrapped velocity spectrum stacks of receiver functions. Geophys.J. Int. 133, 31–43.

rifune, T., Nishiyama, N., Kuroda, K., Inoue, T., Isshiki, M., Utsumi, W., Funakoshi,K., Urakawa, S., Uchida, T., Katsura, T., Ohtaka, O., 1998. The post spinel phaseboundary in Mg2SiO4 determined by in situ X-ray diffraction. Science 279,1698–1700.

rifune, T., Higo, Y., Inoue, T., Kono, Y., Ohfuji, H., Funakoshi, K., 2008. Sound velocitiesof majorite garnet and the composition of the mantle transition region. Nature451, 814–817.

to, E., Takahashi, E., 1989. Postspinel transformations in the systemMg2SiO4–Fe2SiO4 and some geophysical implications. J. Geophys. Res. 94,10637–10646.

atsura, T., Ito, E., 1989. The system Mg2SiO4–Fe2SiO4 at high pressures and tem-peratures: precise determination of stabilities of olivine, modified spinel, andspinel. J. Geophys. Res. 94, 15663–15670.

atsura, T., Yamada, H., Shimei, T., Kubo, A., Ono, S., Kanzaki, M., Yoneda, A., Walter,M.J., Ito, E., Urakawa, S., Funakoshi, K., Utsumi, W., 2003. Post-spinel transition inMg2SiO4 determined by high P-T in situ X-ray diffractometry. Phys. Earth Planet.Inter. 136, 11–24.

ennett, B.L.N., Engdahl, E.R., 1991. Traveltimes for global earthquake location andphase identification. Geophys. J. Int. 105, 429–465.

ubo, T., Ohtani, E., Kato, T., Urakawa, S., Suzuki, A., Kanbe, Y., Funakoshi, K., Utsumi,W., Kikegawa, T., Fujino, K., 2002. Mechanisms and kinetics of the postspineltransformation in Mg2SiO4. Phys. Earth Planet. Inter. 129, 153–171.

angston, C.A., 1997. The effect of planar dipping structure on source andreceiver responses for constant ray parameter. Bull. Seismol. Soc. Am. 67,1029–1050.

itasov, K., Ohtani, E., Sano, A., Suzuki, A., Funakoshi, K., 2005. Wet subduction versuscold subduction. Geophys. Res. Lett. 32, L13312, doi:10.1029/2005GL022921.

Please cite this article in press as: Gao, Y., et al., Seismic discontinuities in teastern China and Korea: Influence of the stagnant Pacific slab. Phys. Earth

iu, F., Kawakatsu, H., 1996. Complex structure of mantle discontinuities at the tipof the subducting slab beneath northeast China—a preliminary investigation ofbroadband receiver functions. J. Phys. Earth 44, 701–711.

iu, F., Kawakatsu, H., 1998. Determination of the absolute depths of the man-tle transition zone discontinuities beneath China: Effect of stagnant slabs ontransition zone discontinuities. Earth Planets Space 50, 965–975.

PRESSetary Interiors xxx (2010) xxx–xxx

Niu, F., Levander, A., Ham, S., Obayashi, M., 2005. Mapping the subducting Pacificslab beneath southwest Japan with Hi-net receiver functions. Earth Planet. Sci.Lett. 239, 9–17.

Obayashi, M., Sugioka, H., Yoshimitsu, J., Fukao, Y., 2006. High temperature anoma-lies oceanward of subducting slabs at the 410-km discontinuity. Earth Planet.Sci. Lett. 243, 149–158.

Owens, T.J., Crosson, R.S., 1988. Shallow structure effects on broadband teleseismicP waveforms. Bull. Seismol. Soc. Am. 78, 96–108.

Ritsema, J., van Heijst, H.J., 2000. Seismic imaging of structural heterogeneity inEarth’s mantle: evidence for large-scale mantle flow. Sci. Prog. 83, 243–259.

Schmerr, N., Garnero, E.J., 2007. Upper mantle discontinuity topography from ther-mal and chemical heterogeneity. Science 318, 623–626.

Shen, X., Zhou, H., Kawakatsu, H., 2008. Mapping the upper mantle discontinu-ities beneath China with teleseismic receiver functions. Earth Planets Space 60,713–719.

Shen, Y., Blum, J., 2003. Seismic evidence for accumulated oceanic crust above the660-km discontinuity beneath southern Africa. Geophys. Res. Lett. 30, L1925,doi:10.1029/2003GL017991.

Simmons, N.A., Gurrola, H., 2000. Multiple seismic discontinuities near the base ofthe mantle transition zone in the Earth’s mantle. Nature 405, 559–562.

Suetsugu, D., Saita, T., Takenaka, H., Niu, F., 2004. Thickness of the mantle transitionzone beneath the South Pacific as inferred from analyses of ScS reverberated andPs converted waves. Phys. Earth Planet. Inter. 146, 35–46.

Suetsugu, D., Shinohara, M., Araki, E., Kanazawa, T., Suyehiro, K., Yamada, T.,Nakahigashi, K., Shiobara, H., Sugioka, H., Kawai, K., Fukao, Y., 2005. Mantle dis-continuity depths beneath the West Philippine Basin from receiver functionanalysis of deep-sea borehole and seafloor broadband waveforms. Bull. Seism.Soc. Am. 95, 1947–1956, doi:10.1785/0120040169.

Tibi, R., Wiens, D.A., Shiobara, H., Sugioka, H., Yuan, X., 2007. Double seismic dis-continuities at the base of the mantle transition zone near the Mariana slab.

he mantle transition zone and at the top of the lower mantle beneathPlanet. In. (2010), doi:10.1016/j.pepi.2010.03.009

Geophys. Res. Lett. 34, L16316, doi:10.1029/2007GL030527.Vacher, P., Mocquet, A., Sotin, C., 1998. Computation of seismic profiles from min-

eral physics: the importance of the non-olivine components for explaining the660 km depth discontinuity. Phys. Earth Planet. Inter. 106, 275–298.

Wang, Y., Martinez, I., Guyot, F., Liebermann, R.C., 1997. The breakdown of olivineto perovskite and magnesiowustite. Science 275, 510–513.