Seismogram Analysis of Indonesian Earthquakes at DAV Observation Station

Earthquake Research in ChinaVolume 23，Number 2, 2009Seismogram Analysis of IndonesianEarthquakes at DA V ObservationStation'Bagus Jaya SantosaProdi Geofisika, FMIPA, ITs, Surabaya 60111, IndonesiaThe S-wave velocity across the earth structure under Indonesia for Indonesia earthquakes hasbeen investigated through seismogram analysis, simultaneously in the time domain and threeCartesian components. The data were recorded at DAV observational station, the Philippines.The main data set is the seismogram comparison between the measured and syntheticseismogram, instead of travel time data， as commonly used in other seismological research.The synthetic seismogram is calculated using the GEMINI method, which is equivalent to ModeSummation.The above seismogram comparison shows that the global earth mantle of PREMAN gives adeviating synthetic seismogram and has earlier arrival times than those of the measurement.The gradient of β in the upper mantle layers is altered into a positive, rather than negativeslope as stated in the PREMAN model, and negative corretions are imposed to the zero orderof the polynomials coefficients in all earth mantle layers. The excellent ftting, as well as traveltime or waveform, is obtained from the surface waves of Love and Rayleigh, surface wave tothe S and ss mantle waves as well as the core reflected waves .This result expresses that part of the earth mantle, due to a collision between India andAsia tectonic released zones, has a negative anomaly in S wave velocity and vertical anisotropyin all of the earth mantle layers.Key words: Seismogram analysis; Vertical anisotropy; Negative velocity anomalyINTRODUCTIONThe main geology of the research area is as follows: Following the closing of the Thetys Ocean inthe Mesozoic geologic time, there was a collision between the Indian and Asian plate which hasformed, created the Himalayas and Qinghai-Xizang (Tibet) Plateau . This process also induces relaxedstrain in South-East Asia and China (Hall, et al., 2004)_ Some fare being upraised by the subduction process due to the coll中国煤化工an and EurasiaTYHCNMHGReceived on July 28, 2008.Volume 23, Number 2211plates. In addition, the East Indonesia part is also being subducted by the Pacific Ocean plate. In thefront of the subduction zone, seismologists ( Replumaz, et al., 2004; Hill, et al., 1998) interpretedthis zone with a negative anomaly in P wave velocity. Fig. 1 ilustrates the main geology of theIndonesia area. It can be seen that the subduction zones are located in the south of Sumatra-Java, andin Kepala Burung of Papua northwards to the Philippines . The subduction zones are marked by curveswith many lttle arrows, mentioned as trenches, where the arrows show the direction of Ocean platemovement. The average speed of Ocean plate movement is ca. 11 cm/a. Large tectonic earthquakesoccurred in this zone, and they will be analyzed using the seismogram data recorded at DA Vobservation station, in the time domain and three Cartesian components .During the last two decades, with improved computer capability and better quality seismic data,seismology opened new fields regarding the Earth's structure . Seismic tomography has been applied tothe mapping of 3-dimensional earth structures ( Replumaz, et al., 2004; Engdahl, et al., 1998;Vasco, et al., 1995; Zhou, 1996; Grand, et al.，1995; Hilst, et al., 1997). For example,forward waveform modeling has been introduced to the study of tomography, and provided detailedinformnation about structural features and small patterns of heterogeneity ( see last reviews ( Zhao,2001，2004)). These studies have developed our understanding of dynamics and the interior of theEarth' s structure. The subducted lithosphere and marine volcanoes are the two features whichcomplement each other (Engdahl, et al., 1998).100PHLPPINEPLATEPACEFIC-107NrsOCEANINDUANJ2rsFig.1The main geology of South East Asia (Hall R., 2002)The span of the Moho depth is measured from approximately 11 km below the ocean plate to 40km~ 70km below the continent plate, and can be performed in the grid model (Vasco, et al., 1995).Discontinuity at depths of 410km and 660km shows a variation of 18km at the global scale ( Garmero,2000). The change of discontinuity of depth influences the travel time and the trajectory of the seismicwave, and therefore this topography has to be calculated in the tomography inversion. The depth of theEarth's crust, as mentioned in the references, can be proven through the study of the Love waveform.The velocity structure was obtained in previous seismological research in the same area(Replumaz，et al., 2004; Engdahl, et al., 1998; Grand中国煤化工al.，1997) byinverting the P travel time data where the amount of P daHCN MH G reled wavephase of pP ca. 0.6x 10*，and PKP deflected in the Eartlrs cortca.I A iu . vala numbers of thiscollection stemmed due to 300, 000 earthquakes from the time interval 1964-01-01 to 2000-12-31212Earthquake Research in China(Engdahl, et al., 1998), and a small number of absolute difference time data through PP-P, PKP-PDIFF, were measured accurately through cros8-correlation of digital broad-band waveform data(Grand, et al., 1995).The objective of this research is to analyze the seismogram data of earthquakes, in Indonesia,which were recorded at DAV observation station, the Philippines. The wave propagation from thehypocenter to the observatory station passes the front area of the subduction zone ( the subduction zonein Sumatra-Java, West Pacific Ocean) . The data set used is all the waveforms in the seismogram, fromthe S wave to Love and Rayleigh surface waves, instead of the travel time data or dispersion analysis[Okabe] or waveform of the S wave only [ Tanimoto, 2007]. Though the data set used by otherseismologists reaches the millions, it contains lttle informnation in the time series of the seismogram.Waveform analysis from S waves, Love and Rayleigh surface waves simultaneously in three Cartesiancomponents, as well as core reflected ScS and up reflected sScS, deals with the entire informationcontent in seismogram.1 METHODOLOGYThe seismogram data was downloaded from the IRIS Databank Center ( Incorporated ResearchInstitutions for Seismology) . The locations of earthquake sources are in Indonesia. Table 1 presents theearthquake hypocenters analyzed in this research, and the data recorded at DAV station， thePhilippines .Table 1 The analyzed earthquake code and hypocenters in Indonesia at the DAV stationNoEarthquake codeLatitude (°N)Longitude (°E)epth (km)C082097B4.3696.4933C1 10895A1.8595.06C031797C- 6.61105.51C010196C0.72119.98C111395A3.56126.67C021395C- 1.31127 .42C021395K- 1.35127.523C012795G-4.46134.46C122297A-5.49147.8779The position of the DAV observation station, the earthquake epicenter and the vertical projectionof the wave paths from earthquake hypocenter to the observation station are presented in Fig .2.Firstly, in the seismogram analysis, the calculation of the travel time of some main phases wascarried out using the TTIMES program ( Bulland and Chapman, 1983) obtained at http://orfeus .knmi. nl. The input for this program is the earthquake source depth and epicentral distance to thestation, while the output is the arrival time of main wave phases . These travel times are used as guidesto identify the wave phases in the seismogram at the observation station. The data set used is theseismogram comparisons between the measured and the synthetic seismogram， where the syntheticseismogram is calculated by the GEMINI method ( Green's function of the Earth by MINor Integration) ，which is equivalent to the Mode Summation method, though the seismoeram can be calculated at higherfrequencies.中国煤化工The GEMINI program calculates the minors of Grel!YHC N M H Gnodel (PREMAN(Dziewonski and Anderson, 1981)) with a certain depth ot the earthquake source. Green's functionsare expanded by filing the boundary conditions at the reflected point of wave propagation, the pointVolume 23, Number 2213100*E120°E140*EC082097B'C111395A.C010196C0°C1 10895AC02139SC、C031797CC021795GC122297A100°E120*EFig.2Vertical projection of wave paths from earthquake epicenter to DAV observational stationof the source depth and the Earth's surface, as well as CMB ( Core Mantle Boundary). Expansion iswritten by using independent variable as complex frequency (w + iσ) to avoid time aliasing .Earthquake tensor moment is applied to calculate the right side of the system equation of motion, whichis detailed in the third line of CMT solution ( Centroid Moment Tensor (Dreger, 2002)), The Green'sfunctions coefficients are calculated by solving the Cramer rule. The coordinates of the earthquakesource and observation stations are transformed to epicentral and azimuth angles. The sphericalharmonic functions are developed using these angles. The DISPEC program ( including in GEMINIpackage) reads the output Green's functions and performs multiplication with expanded sphericalharmonic coefficients from the moment tensor and the spherical harmonic functions . The program sumsthe multiplied Green's functions and yields the synthetic seismogram in the complex frequency domain.The MONPR program ( in the GEMINI package) transformns the synthetic seismogram from thecomplex frequency domain into the time domain. A low pass filter is applied to the synthetic and themeasured seismograms. The inversion of the RESPONSE file of seismometer equipment systems in thereceiver station is applied to the measured seismogram. The RESPONSE file is the descriptionconcerming the amplification and phase change of equipment systems, as altering the input of velocityof ground movement to electric output [ mV ] . Horizontal components of measured seismograms have tobe tumed around the local vertical axis, where the local North-South axis is aimed in the direction fromthe observatory station to the earthquake source, as shown in Fig. 2. This rotation is needed todecompose complex wave motions into P-SV and SH wave motions. Therefore， the synthetic andmeasured seismograms are then compared in the same unit and movement directions .2 SEISMOGRAM ANALYSIS AND DISCUSSION2.1 Seismogram Analysis and FittingThe seismogram analysis of the Indonesian earthquakes recorded at DAV station is presented inFig.3 and is shown as follows: The figure contains 3 curves, the solid curve is the measured data, thedotted curve is the synthetic seismogram from the PREMAN eeth mndel and the, dot-dash curve is thecorrected Earth model. The correctin of the S velocity co中国煤化工g, in the uppermantle layers and the zero order coefficients of the polynom:YHCN M H Gity structure inthe Earth mantle layers. The correction is needed to achieve seismogram ftting throughout the S wavephase and the repetition, as well as the Love and Rayleigh surface wave. The result of the correction of214Earthquake Research in Chinathe S velocity is displayed on a small box on the right side as a dot-dash curve, between the PREMANmodel and the corrected model .The seismogram analysis started with the earthquakes occuring in West Indonesia, and then inEast Indonesia. First is the seismogram of the C082097B earthquake that occurred in Aceh, on thesubduction zone in North Sumatra. The seismogram synthesis from the PREMAN Earth model givesgreat deviation, that is, the synthetic Rayleigh surface wave in the r and z components from PREMANarrive earlier than the appropriate measured Rayleigh wave. While the corrected eath model gives abetter arrival time for the Rayleigh wave, the measured Love wave is well approached by these twosynthetic seismograms from the PREMAN and the corrected Earth model, where the correction coversthe change of the β gradient in the upper mantle layers. The mistake patterm of the amplitudewaveform stems from the mistake in the solution tensor CMT in the time history function, releasedenergy quantity and direction of the fault plane of the earthquake (Dreger, 2002). .970820 NORTHERN SUMAFERA (5.9)SnSnPKiKPPg |cP|28.9NONNDAV" i -Kon:Preman- ; Data- : Syn. Kor.'Syn. PREMAN1216Minute afer the origin time (PDE), corner freq. -20 mHz'Fig.3ismogram analysis and ftting of C082097B earthquake at DAV stationFig. 4 presents the seismogram comparison of the C1 10895A earthquake that occurred in thesubduction zone of Sumatra, recorded at the DAV station. We can see clearly that the synthetic Lovewave from PREMAN arrives a little earlier, whereas the synthetic Rayleigh wave arrives clearly earlierin front of the measured Rayleigh. The Earth model is corrected with the change on the gradient inthe upper mantle layer, and the negative corrections of both styles of β velocity in the mantle layersdown to a depth of 730km provide a synthetic seismogram中国煤化工5 arival times ofcogram well.Fig.5 presents the seismogram analysis of the earthquthe synthetic Love and Rayleigh surface waves from PREM:YHC N M H Gpriate measuredwaves. Despite the application of removal data trend to the preparations for the seismogram analysis,Volume 23, Number 2215951108 OFF W COAST OF NORTHERN (6.1)B。s SPg SnSnPKiKPi:PsPpSSDAV- Kor.一: DataSyn. Kor.: Syn. PREMAN16Minute after the origin time (PDE), comner freq. =20 mHzFig.4Seismogram analysis and fiting of C1 10895A earthquake at DAV970317 SUDAN STRAIT5.8)PnPns sSnpPpPcP |DAvHC|&DAvHMntMf ej-Kom.一: Dta， : Sym. Kor.. : Syn. PREMANMinute ster the origin time (PDE), corner foq,●20 mHz中国煤化工Fig.5MHCNMHGSeismogram Analysis and fiting of C031797 caunquaxl a uni ouaisui216Earthquake Research in Chinathe low amplitude in front of the Love wave is very dificult to simulate with the synthetic seismogram.Analysis is cried out to obtain the fiting of the S wave and the maximum from the Love wave, wherethe correction is performed by changing the negative gradient into the positive gradient of and thecorrection of the velocity of the S wave in the mantle layers .Fig.6 presents the seismogram comparison of the C010196C earthquake that occured in the smallsubduction zone in North Minahasa， North Celebes (Fig.1) that was recorded at the DAV station,where the epicentral distance is quite close to 8.59. However, the synthetic seismogram fromPREMAN gives a great deviation against the measured seismogram. Due to small distance and deviatedwaveform, the imposed negative correction should be of quite large order, as ilustrated in the right boxof the figure. Attention should be given to the geologic conditions in this area, where the subductionzone is parallel to the Celebes Sea, while the wave path crosses the slab of this sea. The analysis ofthis seismogram manifests that the subducted slab of Celebes Sea has a negative velocity anomaly.960101 MINAHASSA PENINSULA (6.2)Bn。&?n sPgin i个个iNA|g3tDAVqKon.一: Data” : Syn. Korr.; Syn. PREMAN3Minute ater the origin time (PDE), corner froq.■20 mHz34567km/Fig.6Seismogram analysis and ftting of C010196C earthquake at DAVFig.7 presents the seismogram comparison between the measured and synthetic seismograms froman earthquake in Talaud, North Celebes , from data recorded at DAV Station . The epicentral distancefrom this earthquake source to the DAV station is small, only 3.4°. With such a small distance as .this, the other seismologists could only measure the arrival time of the P wave. Therefore, until nowthere has been no other seismological research article on the analysis of S waves in the seismogram fordistances as close as this, whilst the PREMAN synthetic seismogram arives earlier. Correcting thisadvance requires a large negative correction to obtain the arival time fitting for the maximum of thewave, both on the Love wave and Rayleigh surface wave.Fig.8 and Fig. 9 present the seismogram comparisq中国煤化工almahera. Thehypocentral locations are close to each other, occurring:MHC NMHG time lag of acouple of hours. The epicentral distances are still small, but the seismogram analysis shows that thesynthetic from the PREMAN for the Love wave arrives earlier with a weakened amplitude pattern. ToVolume 23, Number 221 7951113 TAILAND ISLAND(5.8).B4s只sSn导f,ivcA3.6--DAV一:DiatiI- : Syn. Kor.I Syn. PREMANMinute after the origin time (PDE), corner toq. 20 mHzFig.7Seismogram analysis and ftting of C111395A earthquake at DAV station950213 HALMAHERA(6.1)队4Pn sPgSb 'pPbAA哥,4567DAVE; →Kor( e Preman-: Data.. : Sym. PREMANMinute afer the origin time (PDE), corner froq. =20 mHzFig.8Seismogram analysis and ftting of C0213中国煤化工fit the arrival time and amplitude patterm, the correctionMYHC NMH Gge form in thethickne8s of the Earth's cnust and the correction of the S velocity with the large negative in the uppermantle layers. The wave train at the end of the Love wave with strong amplitude cannot be well218Earthquake Research in China950213 HALMAHERA(6.2)| sPgsSPeP|AA8.6°DAVqAAn一: Data: Syn. Ko.:Syn. PREMANMinue afer the origin time (PDE), corner frq:=20 mHz 345° 7Fig.9Seismogram analysis and ftting of C021395K earthquake at DAV station950127 WEST RAN REGION (6.2)PbPbSn SbSbPcPPgPgSgSg多|门:Preman一: Du- :Syn. Kor.Minute after the origin time (PDE), corner freq. =20 mHz 34中国煤化工Fig.10JMHCNMHGSeismogram analysis and ftting of C0127956 caurquant a int Siauunt .Volume 23, Number 2219imitated by these two synthetic seismograms, because of the reverberation characteristics of the fineEarth layering system between the bypocenters and the DAV observation station.(@) 9122 EAST PAPUANEW GUINEARE (6.3)sSjSPg25.8DAV一:Data- ; Syn. Kor.: Syn. PREMAN1012Minute sner the origin time (DE,orefoq.-20mHz 3m 6 7(0) 971222 EAST PAPUANEW GUNEARE (6.3)pPKiKPSKiKPDAVE一: Du-.:Syn.Kor.●: Syn. PREMAN2Minute after the origin time (PDE),corner foq. =20 mHzFig.11Seismogram analysis and ftting of C122297A中国煤化工，time window of (a) S, Love and RayleighYHCNMHGFig.10 presents the seismogram comparison of the eartnquake tnat occurrea in the KepalaBurung, Irian Jaya. The synthetic seismogram from PREMAN gives real discrepancies against the220Earthquake Research in Chinameasured seismogram. By applying a positive gradient to wave velocity in the upper mantle layersand negative corrections to the zero order cefficients of the two kinds of S wave in the mantle layersdown to a depth of 730km, seismogram ftting can be achieved. The S is immersed in the Love wave'samplitude when the epicentral distance is small.Fig.11 presents the seismogram comparison of the earthquake in P. N. G recorded at DAVstation. The wave path from the earthquake source to the station passes through the structure of theOcean slab. However, the maximum amplitude of the Love wave from both synthetics is higher thanthat of the measured , which is contributed to the error in the CMT solution of this earthquake ( Dreger,2002). Small negative corrections on β, and β in the upper mantle layers also icontribute to thewaveform ftting of the ScS wave. Since the earthquake has a focal depth of 176km， it also gives thepossibility of checking the ScS wave reflected from Earth surface, that is, sScS . The path's differencebetween the ScS and the sScS is 2 times, 352km, so the ScS and sScS wave will be self interference ，as seen in Fig.11(b) . These two core reflected waves could be well simulated by the corrected Earthmodel .2.2 DiscussionThe PREMAN earth model is presented with the vertical anisotropic characteristics only in theupper mantle layers. It can be seen in Fig.3~ Fig.11 that all the analyzed earthquakes show anarrival time earlier than the synthetic Rayleigh surface wave' s constructed from the PREMAN model .This demands a larger amount of negative correction for the S velocity structure. This shows thatvertical anisotropy is bigger and deeper, different from that written in the PREMAN Earth model. Tofit both the S waves (SV and SH) on seismograms of several earthquakes, the weaker negativecorrections in the mantle layers down to the depth 730km are needed. This shows that thecharacteristics of vertical anisotropy also occurred in the mantle layers beneath the upper mantle layers .This anisotropic characteristic is not used in the seismology research based on travel time data, becauseit is dificult to observe the arrival time dfference of the S wave in the three Caresian components.The results of this research match the research results of Replumaz, et al. (2004) and Hilst, et al .(1997) on the structure of the P wave velocity under South- East Asia and South China Sea.Seismogram analysis of earthquakes that occurred around Celebes indicates that at small epicentraldistance stations, the research can analyze and make a fitting on the strong-motion seismogram usingthe CEMINI method. With the arrival time data in the short distance seismograms, all wave phaseswill crumple, making the arrival time data of P wave the only available data, as the first-breaks. Withthe GEMINI method, we can precisely analyze waves with big amplitude, which are not the firstbreak .Deep earthquakes can be exploited to comprehend wave velocity structure in the shallow layersand in the base mantle layers, through wave analysis of ScS and sScS. Research of ScS waves usingsmall epicentral distance station data has never been performed in other seismological research, wherethe arrival time difference data of ScS-S is required, and quantitative measurement can only be carriedout at large epicentral distance stations (Wysession, et al., 1998).Determination of earthquake CMT solutions is performed by CMT routines ( Dreger, 2002).Supposing we have an isotropic earth model, known as the 1066B Earth model, the calculation ofGreen' S functions for various depths of the earthquake source is carried out， and the syntheticseismogram is reconstructed using a CMT tensor that varies with the Monte Carlo Method, by payingattention to the absolute spectrum of the seismogram.中国煤化工Iwersions is theseismogram comparison in the time domain, but the execuler frequency of8 mHz. However, the GEMINI program used in this reseanMYHcNMHGsmogramsupto40 mHz，which is 2.5 higher. Therefore， the Green routine used in the CMT inversion should beVolume 23, Number 2221replaced with the GEMINI routine, 8o that seismogram comparison can be carried out in higherfrequencies .3 SUMMARYThe structure of the S wave velocity in front of the subduction zones and the non-tectonics areaunder Indonesia were investigated through the analysis of the seismograms of earthquakes that bappenedin Indonesia, with the seismograms recorded at the DAV station, the Philippines, in time domain andthree Cartesian components simultaneously. The data set used in this research was seismogramcomparisons, which used all information available in the seismogram, unlike the data set of arrivaltime, where the analyzed information in the seismogram is small.The synthetic seismogram is calculated using the GEMINI method, which is equivalent to theNormal Mode method. To simplify the waveform analysis, the low pass filter was applied lo theseismograms, where the comer frequency was set to 20 mHz.The seismogram comparison between the measured and synthetic seismogram shows that the arrivaltimes of S waves are earlier than the appropriate measured wave phase and the surface waves of Loveand Rayleigh, where the synthetic seismogram is reconstructed from the PREMAN Earth model. Toobtain the seismogram ftting, the Earth model must be corrected with negative anomalies on S wavevelocity. This correction covers the change of the gradient, which is positive in the upper mantlelayers, and the zero order cofficients of β, in the polynomial function of the S velocity wave in eachlayer of the earth mantle layers. The main negative correction is aplied to β, to obtain the ftting forthe Rayleigh wave. The correction is also carried out for S and ss and deep ScS and sScS waves. Thisshows that the vertical anisotropic characteristics are stronger than supposed in the PREMAN Earthmodel that occurs only in the upper mantle layers.The negative corrections in the upper mantle layers have also provided a good ftting for body anddepth waves. This result indicates that the Earth mantle part, due to the collision between the Indiaand Asia，Pacific and Asia tectonic released zones, has negative anomaly in S wave velocity andvertical anisotropy in all Earth mantle layers.ACKNOWLEDGEMENTSGratitude is addressed to Dr. Dalkomo and Prof. Friedrich who developed the GEMINI programwith the author, as well as to Prof. Wielandt. Gratitude is also addressed to IRIS ( IntemationalSeismogram Network) for providing the seismogram data for this research and its complementarysoftware. The main program was written with the non-commercial software Intel FORTRAN, and thefigures in this paper were written using the PGPLOT and CMT software.REFERENCESBoschi L., Dziewonski A. High- and low- resolution images of the Eart's mantle: Implications of diferent approaches totomographie modeling[J]. J. Geophys. Res.. 1999, 104:25567 ~ 25594.Bulland R. and Chapman, C. Travel time calculation [J]. BSSA, 1983, 73:1271 ~ 1302.Dalkolmo J. Synthetische Seismogramme fuer eine sphaerisch aymmetrische, nichtrotierend Erde durch direkteBerechnung der Greenschen Funktion. Diplomarbeit, Inst. fuer Geophys., Uni. Stuttgart, 1993.Dreger D. s. Time-Domain Moment Tensor INVerse Code (TDMT中国煤化工logical Labortoy(BSL), report mumber 8511, 2002.Dziewonski A. M. and Anderson D. L. Preliminary reference Earth"TYHC N MH Gand Plan. Int..1981, 25:297 ~ 356.Engdahl E.R., Van Der Hilset R.D., Buland R. P. Clobal tleisic earthquake relocation with improved travel times222Earthquake Research in Chinaand procedures for depth determination [J]. Bull. Seism. Soc. Am., 1998, 88:722~ 743.Friederich W. and Dalkolmo J. Complete synthetic seismograms for a spherically symmetric earth by a mumericalGarmero E. Heterogeneity of the lowermost mantle [J]. Annu. Rev. Earth Planet. Sei., 20 b00, 28:509 ~ 537.Crand s., van der Hilst R., Widiyantoro s. Clobal seismic tomography: a snapshot of convection in the Earth [J].GSA Today, 1997, 7: 1~7.Hall R. Cenozoic geological and plate tectonic evolution of SE Asia and the sw Pacific: computer based reconstructions,model and animations [J]. Joumnal of Asian Earth Seiences, 2002, 20:353~ 431.Replumaz A, Krason, H, van der Hilst R.D., Besse J. and Tapponnier P. 4-D evolution of SE Asia's mantle fromgeological reconstructions and seismic tomography [J]. Earth and Planetary Science Ltters, 2004, 221 :103 ~ 115.Van der Hilst, R., Widiyantoro, s., Engdahl, E. Evidence for deep mantle circulation from global tomography [J].Nature, 1997, 386: 578 ~ 584.Vasco D.. Johnson L., Plliam R. Lateral variations in manle velocity structure and discontinuities determined from P,PP, S, ss, and SS-ScS travel time reiduals [J]. J. Geophys. Res.， 1995, 100:24037 ~ 24059.Wysession M., Lay T.，Revenaugh J. The D" discontinuity and its implications. In: Gumis， M., Bufel， B.,Knittle, K., Wysession, M. (Eds.), The Core Mantle Boundary. AGU, 1998. 273 ~ 297.Zhao D. Global tomographic images of mantle plumes and subducting slabs: insight into deep Earth dynamics [J].Plhysics of the Earth and Planetary Interiors, 2004, 146:3~ 34.Zhao D. Seismic structure and origin of hotspots and mantle plumnes[J]. Earth Planet. Sei. lt., 2001, 192: 251 ~Zhou H. A high-esolution P wave model for the top 1200km of the manle [J]. J. Geophys. Res., 1996, 101 :27791~ 27810.About the AuthorBagus Jaya Santosa, bom on August 2, 1962, is a Lecturer of the Physics Department, FMIPA,ITs. B.Sc 1986 in UGM University, M.Se. 1988 in UGM, Dr. Rer. Nat. 1999 in the Institute furGeophysik，Stuttgart University. Major interest is waveform seismology . Research interests include theimaging of the S-wave velocity structure of Earth structure using waveform analysis. E-mail: bjs @physics . its . ac.id中国煤化工MYHCNMHG