SKS splitting beneath continental rift zones

View metadata, citation and similar papers at core.ac.uk brought to you by CORE provided by Missouri University of Science and Technology (Missouri S&T): Scholars' Mine Missouri University of Science and Technology Scholars' Mine Geosciences and Geological and Petroleum Engineering Faculty Research & Creative Works Geosciences and Geological and Petroleum Engineering 01 Oct 1997 S K S Splitting beneath Continental Rift Zones Stephen S. Gao Missouri University of Science and Technology, sgao@mst.edu Paul M. Davis Kelly H. Liu Missouri University of Science and Technology, liukh@mst.edu Philip D. Slack et. al. For a complete list of authors, see https://scholarsmine.mst.edu/geosci_geo_peteng_facwork/97 Follow this and additional works at: https://scholarsmine.mst.edu/geosci_geo_peteng_facwork Part of the Geology Commons Recommended Citation S. S. Gao et al., "S K S Splitting beneath Continental Rift Zones," Journal of Geophysical Research, vol. 102, no. B10, pp. 22781-22797, American Geophysical Union (AGU), Oct 1997. The definitive version is available at https://doi.org/10.1029/97JB01858 This Article - Journal is brought to you for free and open access by Scholars' Mine. It has been accepted for inclusion in Geosciences and Geological and Petroleum Engineering Faculty Research & Creative Works by an authorized administrator of Scholars' Mine. This work is protected by U. S. Copyright Law. Unauthorized use including reproduction for redistribution requires the permission of the copyright holder. For more information, please contact scholarsmine@mst.edu. JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 102, NO. B10, PAGES 22,781-22,797,OCTOBER 10, 1997 $K$ splitting beneath continental rift zones S. Gao,1 P.M. Davis,H. Liu,2 P. D. Slack,andA. W. Rigor Department of Earth and SpaceSciences,University of California, Los Angeles Y. A. Zorin, V. V. Mordvinova,V. M. Kozhevnikov,and N. A. Logatchev Institute of the Earth's Crust, Siberian Branch, Russian Academy of Sciences,Irkutsk Abstract. We presentmeasurements of $KS splitting at 28 digital seismicstations and 35 analog stations in the BaikM rift zone, Siberia, and adjacent areas, and at 17 stations in the East African Rift in Kenya and compare them with previous measurements from the Rio Grande Rift of North America. Fast directions in the inner region of the Baikal rift zone are distributedin two orthogonaldirections, NE •nd NW, •pproxim•tely p•r•llel •nd perpendicularto the NE strike of the rift. In the adjacent Siberian platform and northern Mongolianfold belt, only the riftorthogonalfast directionis observed.In southcentralMongolia,the dominantfast direction changesto rift-parMlel again, althougha small numberof measurements are still rift-orthogonal. For the axial zonesof the East African and Rio Grande Rifts, fast directionsare orientedon averageNNE, that is, rotated clockwisefrom the N-S trending rift. All three rifts are underlain by low-velocityupper mantle as determinedfrom teleseismictomography. Rift-related mantle flow providesa plausibleinterpretation for the rift-orthogonMfast directions. The rift-parallel fast directionsnear the rift axes can be interpreted by oriented magmatic cracksin the mantle or small-scalemantle convectionwith rift-parallel flow. The agreement betweenstressestimatesand correspondingcrackorientationslendssomeweight to the suggestion that the rift-parMlelfast directionsare causedby orientedmagmatic cracks. flow,and of volcanism, magma-filled orientedcracksin the mantlecangiveriseto the observed polarization. A recent review of polarization anisotropymeasured Polarizationanisotropydeterminedby SKS splitting on the continents[Silver,1996]concludes that in trans- isdeveloping asa toolto examine finitestraincaused by pressireregionsanisotropyis causedby vertically co- flowin the uppermost mantle[Fuchs,1977;Andoet al., 1. Introduction herent deformation of the mantle with fast directions parallel to the compressional features. While this behavior is observedfor compressireregionssuch as the Alps, Tibet, and the Tien Shan, the sameprocessoperating in regionsof extensionis expectedto generatefast directionsparallel to the extension.Contrary to this expectation,we presentresultsfrom three continentalrift zones,the Rio GrandeRift (RGR), the Baikalrift zone 1983; Silver and Chan, 1988, 1991; Vinnik et al., 1989, 1992; Savageet al., 1990; Silver, 1996]. The earliest indicationsthat seismicanisotropyis related to mantle flow were observationsof fast mantle velocitiesperpen- dicular to mid-oceanridges[Hess,1964; Raitt et al., 1969; Forsyth,1975]. Azimuthal dependence of velocities of long-periodRayleigh waveshas been observed by Tanimotoand Anderson[1985]and Montagnerand (BRZ) andthe EastAfricanRift (EAR) wherefast di- Tanimoto[1991],whonotedthat the globalmap of fast rectionsare observedorthogonalto the extension. We phasevelocitiescorrelateswith absoluteplate motions. proposethat a plausibleexplanationfor theseobserva- Subsequently,numerousobservationshave been made tionsis that in regionsof mantleextension,of highheat of SKS splitting in diversetectonicsettings(see re- viewsby Vinniket al. [1992]and Silver[1996]). Vinnik •Now at Departmentof TerrestrialMagnetism,Carnegie et al. [1992]report singlestationmeasurements from Institution of Washington, D.C. rift zones such as the Red Sea rift, the Baikal rift, the 2Also at Department of Terrestrial Magnetism, Carnegie continentalextensionof the Arctic ridge, and the RhineInstitution of Washington, D.C. graben. They concludethat in regionsof rifting, the inferred Copyright 1997 by the American GeophysicalUnion. Paper number 97JB01858. 0148-0227/ 97/ 97JB-01858509.00 directions fast directions of flow in the mantle from the are close to the extension directions in the crust. However, for the Rio Grande and the French Massif Central, they find that the fast directionsare ori22,781 22,782 GAO ET AL.: $KS SPLITTING BENEATH CONTINENTAL RIFT ZONES ented nearly 50o to the direction of extension. Recent with the a, b, and c axesof olivine becomingaligned measurements of near ridgesplittingin Iceland[Bjar- with the longest, shortest, and intermediate axes of nasone! al., 1996]givefast directionsthat are closerto the finite strain ellipsoid.McKenzie's[1979]approach waspursuedby Ribe[1992]who developed orientation parallel to the ridge than perpendicularto it. In this paper we examine about 80 continentalrift zone measurements in order to addressfurther the questionwhether in regionsof continental extensionfast directionsare parallel or perpendicularto the extensiondirection. $K$ splitting can be used to characterizeseismic anisotropybeneatha seismicstation[e.g.,Kind e! al., 1985; Silver and Chan, 1988, 1991; Vinnik e! al., 1989; Savage e! al., 1990; Makeyeva e! al., 1992; Silver and Kaneshima, 1993; Gao e! al., 1994a; Liu e! al., 1995; Gao, 1995;Silver, 1996]. The azimuthalanisotropyof distribution formulas for deformation-induced LPO of Ribe and Yu [1991]. Accordingto the theory,under uniaxial compression,the a axis of olivine turns to be perpendicularto the maximum compressional strain direction; under pure shear, it becomesperpendicularto the shorteningdirection; and under progressive simple shear, it aligns in the flow direction. This interpretation has been supported by numerousobservationsof anisotropictextures in deformedmantle rockscomplemented by measurementsand theoretical estimatesof the mantle is readily detectedby birefringenteffectson $K$ phasesfrom earthquakeswith epicentral distance greater than 85ø. The $K$ phasetravelsas an $ wave in the crust and mantle and as a P wave through the liquid outer core. At the core-mantleboundary on the receiver side, the $ wave is radially polarized. If this shear wave encountersanisotropicmaterial on its path to the receiver, it may split into two different polarizations having differentvelocitiesdenotedfast and slow. The splitting parametersare the polarizationof the fast their seismicanisotropy[Hess,1964; Crossonand Lin, P wave has the intermediate velocity with Ve - 8.65 cracks. 1971; Christensen,1971, 1984; Baker and Carter, 1972; Nicolas et al., 1973; Peselnick et al., 1974; Christensen and Salisbury,1979; Christensen,1984; Mercier, 1985; Nicolas and Christensen,1987; Silver and Chan, 1991; Chastelet al., 1993]. A continentalrift is defined as an elongatedtopographic depressionoverlying a place where the entire lithospherehas been significantlymodifiedin extension [OlsenandMorgan,1995].In continentalrift zoneswe split shearwave• (measuredclockwise fromthe north) wouldexpect that if the entire lithosphereis extended and the travel time difference between the fast and slow the a axes and thereforethe splitting fast directions shear wavesSt. The method used to calculate the opti- would clusterin the directionof rift opening. Some mal splitting parametersand their standarddeviations commonfeatures of a typical continentalrift zone inand the wasdescribed extensivelyby Silverand Chan[1991]. clude(1) the centralpart of rift valleysubsides It is generallybelievedthat the main causefor $K$ edgesof the adjacentblocksare uplifted;(2) flanking splitting in the mantle is the lattice preferredorienta- normalfaults;(3) negativeBouguergravityanomalies; tion (LPO) of crystallographicaxes of elastically (4) higherthannormalheatflow;(5) shallow, tensional anisotropicmineralsin the upper mantle [e.g., Hess, andhigherthan normalseismicity; and (6) thinningof 1964; Francis, 1969; Karato, 1989; Savageet al., 1990; the crustbeneaththe rift valley[TurcotteandSchubert, Babuskaand Cara, 1991; Silver and Chan, 1991; Chas- 1982; Fowler,1990]. All thesefeaturesmark the initel et al., 1993; Silver, 1996]. Both olivine and or- tial stagesof continentalbreakupwherethe lithosphere thopyroxeneare highly anisotropicupper mantle con- pulls apart and upwellinghot mantle risesbeneaththe stituent minerals. Mantle peridotirestypically contain spreadingrift to eventually form oceaniclithosphere morethan 65%olivineand 20%orthopyroxene [Ander- and an ocean basin. son,1989].Sinceorthopyroxene is not asabundantand The mechanisms of continentalrifting can be sepanisotropicas olivine, it is generallyassumedthat the arated into two end-memberscalledpassiveand active upper mantle beneath a station is composedof a certain rifting. For passiverifting the entirelithosphere extends percentageof oriented olivine. The rest is assumedto beneaththe rift zone. For this case,LPO is expected be randomlyorientedolivine,orthopyroxene,and other to alignin the directionof extension.Activeriftinginmantle minerals. It must be mentioned that in realvolveserosionof the lithosphereby mantle convection ity the alignmentof orthopyroxenemay reduceoverall causingthe asthenosphereto upwarp beneath the rift anisotropycausedby olivine alignment,sinceits fast axis. If a fabric developsin the asthenosphere, its direcaxis doesn'talign parallelto olivinefast axisduringde- tion will depend on the details of the convection.Two formation[Christensen andLundquist, 1982]. mechanisms for the developmentof anisotropyin upFor an individual olivine crystal, the fastest P wave wellingasthenosphere havebeenconsidered by Kendall propagationdirectionis alongthe a axiswith Va - 9.87 [1994]for the mid-oceanridgecase.The first involves km/s; alongthe b axis the P wavehas the slowestve- the preferentialalignmentof fast axesof olivinecrystals; locity with ¬ - 7.73 km/s; and alongthe c axis the the secondinvolvespreferentiallyalignedmagma-filled km/s [Verma,1960].The fastest$ wavesare thosepoWe compare$K$ and tomographicmeasurements larizedalongthe a axis. McKenzie[1979]assumed that from the three major Cenozoic continental rift areas under sufficientlyhigh stressand finite strain, seismic (Figure1): theRioGrandeRift (RGR)of NorthAmeranisotropyin peridotiresis controlledby finite strain ica [Sandvol et al., 1992],the EastAfricanRift (EAR) GAO ET AL.: SKS SPLITTING 60' • :' '"• ::•'::;'" 180' 5.... BENEATH CONTINENTAL -90' O' RIFT ZONES 22,783 55' 45' •-•'••••d•t,' 180' 95' 100' 105' 110' 115' 120' 90' ....... : 35' ..... _..--•i•i...;.;-4!:: .. .......... .,--•!-:•:-• ;:::t?.-" t•*' ,.. . ................ ".'•*•;,. "..••.::.•.... ':..... ";-":",'"" *-:*;•'•-"';":'::,b. -5' * 30' .............. ---. ................ -35' 40' 25' 45' ' 15' -110' -105' -100' -95' Figure 1. Mercatorprojection mapsshowing locations of the threeriftsandtheirtopography andseismicity. (a) Locations of the 3 riftson a globalmap;(b)-(d) Stationsusedin the study (solidsquares), locationof My _>4.5 earthquakes that occurred 1965-1992 (opencircles),and topographyfor Baikal rift zone, East African Rift, and Rio Grande Rift, respectively. The geographicareasin Figureslb, lc, and ld haveapproximatelythe samedimensions. [Gao et al., 1994c],and alonga 1280km profileacross southernMongolianfold belt. The stationswithin an the Baikalrift zone(BRZ) usingthe datafroma portable area are numberedfrom north to south, e.g., the northexperiment[Gao et al., 1994a;Gao, 1995]alongwith ernmost station in area A is A01 and the southernmost new estimates from data provided by the Baikal Seis- one is All. mic Network. The 63'stationsbelongto two seismicnetworks. Stations named 92xx in Table la were from a portable Program for Array Seismic Studies of the Continen- 2. Data and Results 2.1. Baikal Rift Zone (BRZ) tal Lithosphere(PASSCAL) experimentconductedin the summer of 1992 by a field team from the InstiIn this sectionwe extend our previousSKS splitting tute of Earth'sCrust (IEC), Irkutsk,Universityof Wis- results[Gao et al., 1994a]to includea data set that consin(UW), and Universityof California, Los Anwe recently collectedfrom the permanent seismicarray geles(UCLA). SKS splittingparameterson most of thesesiteswerepreviouslyreported[Gao et al., 1994a]. operated in the Baikal rift zone. 2.1.1. Data. Station locationsare shownin Figure 2 and listed in Table 1. We separatedthe region into six subareas which approximately correspondto the major tectonic provincesof the area. Lake Baikal sits on the southeastmargin of the Siberian craton in a regionthat underwentcollisionaltectonicsin the Mesozoic followedby regionalextensionin the Cenozoic.To the southlies the Mongolianfold be]t, a regionof transpressionthought to be related to the collisionof India with Asia. Areas A, B, and C comprisethe BRZ with area A includingthe NE rift valleys,area B including the central rift valleys,and area C includingthe southern rift valleys;area D is the Siberianplatform, area E is the northern Mongolianfold be]t, and area F is the The remaining stations with alphabetical names such as "boda" belong to the Baikal SeismicNetwork op- eratedby IEC. For this studythe analog, seismograms werescannedand digitized usingthe NXSCAN software package[Humphreyand Helmberger,1993]. Locations of the 31 eventsusedin the analysisare shownin Figure 3, and event information is listed in Table 2. Figures4 and 5 showoriginal SKS signalsbefore and after correctionfor splittingalongwith their particle motion patterns at station E01 for two events. One of the evnets(event8) exhibitsexcellentsplitting,the other(event17) doesnot. Whenthe difference between the radial direction of an event-stationpair and fast SKS direction is near 0ø or 90ø, the anisotropyeffect 22,784 GAO ET AL' SKS SPLITTING BENEATH CONTINENTAL RIFT ZONES 99' 102' 105' 108' 111 ø 114' 117' 120' I I I I I I I I r',,,.o .:.,..:2• o I_ s o • . • .•,•:...• .... o • .• ' I ß '----' , 60 ø 80' lO0 ø 120' •'/ • o/-:.-•/•.•..•,,, I F ø •o.... •:"L -?---'."•.•/•?:•i:;•i•! •.......... n 0' /,',_ o ""_ o '•o / , i ........... 48' / , 45' 140' leigure 2. A Mercator projection map fortheBaikal-Mongolia area showing locations ofstations used inthestudy, regional stress fields associated withBRZ, and local events thatoccurred during thefieldexperiment (open cirlcles). Triangles areBaikal Seismic Network analog stations, and squares aredigitalstations operated in thesummer of 1992.Solidsquares represent stations withprevious SKSsplitting measurements [Gaoetal.,1994a]. Arrows show regional stress fields oftheBaikal riftzone obtained from surface geological structure analysis andearthquake focal mechanism studies [Sherman, 1992]. Thestudy region isdivided intosixareas: areas A,B, andCaretheBaikal riftzone which isoutlined bythethick gray lines. Area AistheNEpartof theBRZ,areaB isthemainriftzone,andareaC istheSWpartoftheBRZ.AreaD islocated ontheSiberian Platform, areaE isthenorthern Mongolian foldbelt,andareaF islocated on theMongolian foldbelt.Thelowerleftinsetshows coastlines andnational borders ofAsiaand the locationof the studyarea. cannot beobserved andthemeasurement isnull[Silver normalizedrosediagramsof the fast directionsfor the andChan,1991].Threenondistinguishable possibilities six areas. The rosediagramsindicatethat thereare may applyin sucha case:the uppermantlebeneath two dominantdirections.One of the directionsis about thestationis (1) isotropic; (2) anisotropic withanun- 60o:t:20ø, whichis approximately parallelto thestrike known splitting timeanda fastdirection parallel tothe of the surfaceexpression of the BRZ aswell asthe strike radialdirection; and(3) anisotropic withan unknownof theuppermantlelow-velocity structureinferredfrom splitting timeanda fastdirection perpendicular to the traveltimeinversion [Gao,1995].The second direction transverse direction. Asdemonstrated in Figures 4 and isabout140o:t:20ø,which isapproximately orthogonal 5, if onlyevent17wereavailable, theresultfromstation E01wouldbea nullmeasurement. Theavailability of to the rift axis. 2.1.2.1. Area A: The peakvalueson the rosedi- event8, whichis froma differentazimuth,resolves the agramfor thisareaare60o for the rift-parallel group measurement. and 130o for the rift-orthogonal group.Of the 11sta2.1.2.Results.Thefinalresults ofSKSsplittingtionsin thisarea,two(A03,A10)havenullmeasure- measurements (Figure6 andTablela) wereobtainedments. Threeofthefourstations (A05,A06,A07,A08) byweighted averaging according to the95%confidence in therift-parallel grouparelocated in themiddlepart interval ofeach individual measurement. Thesplitting of thisarea.Thetwostations whichbelong to therift- ranges from0.3to2.1swhich could have been caused by orthogonalgroupare locatednear the northernboundlayers of30to210kmthick,respectively, characterized ary(A01)andonthesouthern boundary (All). by4%anisotropy. Although directions inthevicinity of 2.1.2.2. Area B: Most of the stations with null theBRZappear toberandomly distributed, examina-measurements are located within this area. Five stationofthescatter reveals some trends. Figure 7 showstions(B01,B02,B07,B12,andB15)neartherift axis GAO ET AL.: $KS SPLITTING BENEATH CONTINENTAL RIFT ZONES 22,785 Table la. $K$ Splitting MeasurementResultsin the Baikal Rift Zone Station A01 A02 A03 A03 A04 A05 A06 A07 A08 A09 A10 A10 All B01 B02 B03 B04 B04 B05 B05 B06 B06 B07 B08 B09 B09 B10 Bll B12 B13 B14 B15 B15 C01 C01 C02 C03 C04 C05 C05 C06 C07 C07 C08 C09 C09 D01 D02 D03 D03 D04 D05 D06 D07 D07 D08 D08 D09 D10 Dll Dll D12 D12 Station Name Coordinates Latitude, Longitude, deg deg Fast Direction, deg boda char 57.800 56.900 114.000 118.300 136.0J:13.0 107.9J: 6.0 nely nely 56.500 56.500 115.700 115.700 29.0J: -119.0J: -- turi 56.400 113.100 86.0 J: 24.0 ozer 56.300 114.000 49.0J: 7.0 tonn 56.300 113.400 64.0J: 8.0 kovo kala kumo uaki uaki 56.100 55.900 55.900 55.500 55.500 113.100 117.400 111.200 113.600 113.600 57.0J: 9.0 61.0 J: 35.0 161.0 J:24.0 38.5J: -128.5J: -- Splitting Time, s 0.70J:0.50 0.82J:0.24 __ __ 0.80-1-0.50 0.70-t-0.30 1.40-I-0.30 0.80-1-0.30 0.50-1-0.30 0.70•0.50 1.40J:0.90 0.84J:0.19 0.30J:0.20 0.60J:0.30 tsip 54.900 113.300 120.0J: 6.0 nizh sbaJ ulun solo solo 55.800 55.600 54.900 54.200 54.200 109.600 109.400 111.200 108.400 108.400 41.3J: 3.2 46.0J: 50.0 76.0 J: 20.0 26.0J: -116.0J: -- ongu ongu 53.600 53.600 107.600 107.600 21.0:J:-111.0J: -- suvo suvo 53.600 53.600 110.000 110.000 82.0 J:35.0 15.2J: 5.5 0.40J:0.30 1.80J:0.64 tyrg 52.800 106.300 44.6J: 3.3 9212 51.847 104.893 144.0J:19.0 0.85J:0.12 0.60J:0.40 9250 9250 9233 9224 51.799 51.799 51.541 51.526 106.015 106.015 104.942 105.121 22.0J: -112.0J: -127.0 J: 10.0 148.0 J: 17.0 9235 9221 51.320 51.292 105.761 105.339 45.7J: 8.7 131.5J: 6.3 9222 9223 51.021 50.791 105.682 105.970 137.9J: 6.2 37.6J: 3.8 9223 orli orli arsh mond t•.la 9270 9270 9272 9271 9271 9237 zaka zaka 9200 9215 50.791 52.600 52.600 51.900 51.700 51.700 51.336 51.336 51.167 51.153 51.153 50.780 50.400 50.400 55.965 55.560 105.970 99.800 99.800 102.400 101.000 103.600 103.458 103.458 104.407 103.877 103.877 104.089 103.300 103.300 101.410 101.803 129.0J: 7.0 70.0J: 10.0 139.6J: 4.8 92. lJ: 11.9 125.5J: 2.9 128.7J: 4.3 24.0J: -114.0J: -141.0J: 7.1 21.0 :J: 111.0J: -58.8J: 5.5 66.5J: 8.3 127.0J: 12.0 133.0J: 7.0 170.0J: 13.0 9213 55.022 102.055 144.8J: 9213 9203 9204 9205 9206 9206 9207 9207 9209 55.022 54.516 54.193 53.929 53.649 53.649 53.243 53.243 52.778 102.055 102.070 102.649 102.934 103.255 103.255 103.767 103.767 104.105 3.5 144.8J: 3.5 127.5:k 3.7 143.8+ 2.6 149.2J: 8.9 20.0-4- -110.0:J: -20.0J: 110.0J: -137.7J: 4.8 9210 irku 52.622 52.200 104.234 104.300 138.2J: 7.4 25.0J: -- irku 52.200 104.300 115.0J: -- 9211 9211 52.169 52.169 104.469 104.469 20.0J: -110.0J: -- __ __ __ __ 0.70J:0.30 1.30J:0.30 0.39J:0.17 0.91J:0.25 1.09J:0.14 1.00-4-0.21 1.60J:0.41 0.80-1-0.20 1.00-1-0.17 1.18-1-0.50 1.48-I-0.20 0.82-1-0.20 1.10-1-0.21 __ 0.83-1-0.17 0.86-I-0.24 1.70-1-0.50 1.00-1-0.50 0.90-1-0.40 0.82-1-0.07 0.82-t-0.07 1.18-1-0.12 0.72-1-0.09 0.58J:0.15 __ __ __ 0.90J:0.16 0.60J:0.13 __ Number Events of 22,786 GAO ET AL.' $KS SPLITTING BENEATHCONTINENTAL RIFT ZONES Table la. (continued) Station Station Coordinates Name Fast Sphtting Number of Latitude, Longitude, Direction, Time, Events deg deg deg s E01 tupi 54.400 119.900 101.0-t- 8.0 0.90-t-0.30 1 E02 E02 E03 E04 E05 E06 chit chit 9280 9291 9282 9283 52.000 52.000 50.193 49.747 49.738 49.288 113.600 113.600 106.254 106.188 106.202 106.412 30.0-t120.0-t- -137.0-t-16.0 145.3-t-16.0 147.24-23.5 134.14- 9.0 1.50-t-0.40 0.78-t-0.14 0.344-0.30 0.374-0.16 9 9 1 4 2 4 F01 hapc 49.700 112.400 71.24- 7.2 0.624-0.15 3 F02 F02 9284 9284 48.931 48.931 106.682 106.682 39.14-10.9 132.04- 9.9 1.044-0.39 0.844-0.57 2 2 F03 F04 F05 F06 F07 F07 F08 F09 9285 9286 9292 9287 9288 9288 9289 9290 48.383 47.921 47.866 47.209 46.635 46.635 46.115 45.262 106.783 106.954 107.051 107.422 107.758 107.758 107.619 108.260 105.24- 5.9 64.44- 4.5 69.24- 6.2 44.24- 6.7 56.24- 9.6 132.24- 3.1 54.74-21.8 84.7 4- 34.4 1.134-0.26 0.754-0.14 0.704-0.13 0.324-0.08 0.404-0.14 1.504-0.19 1.284-0.32 1.424-0.62 4 3 3 4 2 3 3 5 showrift-parallel fast directions. The peak value on the rose diagram for this area is 50ø, which correspondsto the rift-parallel group. Most of the other stationshave fast directionsorthogonalto the rift axis, as evidenced by the 130o fast direction peak on the rose diagram trast liessignificantlydeeperthan this,the Fresnelzone is broader and the transition in the fast directions would spreadout over a longerdistancethan observed. 2.1.2.3. Area C: The rift-parallel fast directions are not as commonas the rift orthogonalones,asshown (Figure 7). The rapid changeof the fast directionsat in the rose diagram for this area. The most common stationsB12 and B13, which are only 30 km apart, may valuefor the rift-parallel fast directionsis about 600, indicate that the main part of the sourceof anisotropy and for the rift-orthogonalgroupit is 1300. 2.1.2.4. Area D: Of the 12 stations in this area, in the area is shallow, probably at a maximum depth four have null measurements. The rest of the measureof less than 50 km, as indicated by the Fresnel zone forwardmodelingof Gao [1995].If the anisotropycon- ments belong to the rift-orthogonalgroup. Table lb. SKS Splitting MeasurementResultsin the East Africa Rift Zone Station Coordinates Latitude, deg Longitude, deg Fast Direction, deg Splitting Time, s K04 K07 -0.19 -0.24 35.44 35.75 20.4428.54- 7.3 8.5 0.864-0.21 1.434-0.31 K09 K10 -0.37 0.64 35.92 36.01 20.04- 3.0 32.24-10.3 2.004-0.20 K12 K15 K16 K24 K25 K31 K45 K47 -0.10 -0.62 0.40 -0.17 -0.47 -0.07 0.29 0.24 35.94 36.30 36.29 36.63 36.82 37.25 35.80 36.07 KS0 K51 K52 K53 -0.33 -0.11 0.06 0.12 36.32 36.46 36.67 37.02 26.04-12.0 19.04-30.0 176.04-13.0 35.04-19.0 26.44- 8.1 10.04- 7.0 26.04-12.0 55.04-28.0 22.04-27.0 33.04-20.0 43.04-13.0 10.04- 9.0 1.504-0.40 2.404-0.50 0.904-0.40 1.104-0.70 1.244-0.38 1.904-0.70 2.304-0.40 1.504-0.80 1.904-0.90 1.504-0.40 0.804-0.30 1.204-0.50 K54 -0.40 37.17 172.04- 0.804-0.30 7.0 1.554-0.38 Number Events of GAO ET AL.: SKS SPLITTING BENEATH CONTINENTAL RIFT ZONES 22,787 Table lc. SKS Splitting MeasurementResultsin the Rio Grande Rift Zone Station Coordinates Latitude, Longitude, deg deg Fast Direction, deg Splitting Time, s CZL ANMO 36.28 34.95 -105.91 -106.46 26.0 41.0 0.9 1.5 SEVI WTX WSMR ELPA 34.27 34.07 32.34 31.77 -106.73 -106.95 -106.49 -106.51 41.0 46.0 11.0 5.0 1.1 1.1 1.2 1.5 From Sandvolet al. [1992]. 2.1.2.5. Area E: The dominant fast direction is rift-orthogonal. The peak value in the rosediagram for this area is 1400 . 2.1.2.6. Area F: The dominant fast directions in this area range from 400 to 800. The fast direction for the southernmost station is nearly E-W. At the northern part of this area, the transition from the dominant rift-orthogonal fast directions in area E to the dominantly ENE fast directions in this area takes place over a distanceof about 50-90 km. Once again, usingthe re- across the Tibetan Plateau[McNamaraet al., 1994]and in the Tien Shaharea[Makeyevaet al., 1992].Both the Tibetan and the MongolianPlateausand the Tien Shan area have been deformed by Cenozoic deformation related to the collision between the Indian and Eurasian plates. The observedfast directionsin these areasmay have the same origin, i.e., the recent continental colli- sion[Gao et al., 1994a]. sult of Fresnelzoneforwardmodeling[Gao, !995], the 2.2. East African Rift (EAR) maximum depth of the source of the anisotropy near Usingdata werecorded duringthe 1989/1990Kenya Rift InternationalSeismicProject(KRISP) [Slacket this transition is about 200 km. In Figure 8 we comparethe BRZ resultswith other clustered measurements in Asia. The fast direction for the southern part of the BRZ profile is roughly ENE, which is consistent with the dominant direction found al., 199d],we found SKS arrivalsat 17 stationssuit- ableforsplittinganalysis (Figure9 andTablelb). The backazimuth for all of the eighteventsusedin measuring SKS splittinglieswithin a narrowrangeof 600to 0 (A) 31 ß 27 2O 21 . , . 24 0 20 50 180 180 Figure 3. Azimuthalequidistant projection mapsshowing epicenters ofeventsusedin thestudy of $KS splittingfor (a) BRZ and (b) EAR. The centerof Figure3a is stationBll, and that of Figure 3b is station K12. The three circlesrepresent45ø,90ø, and 1350epicentraldistances. Numberson the peripheryindicateazimuthsof eventsrelativeto the centerof map. 22,788 GAO ET AL.: SKS SPLITTING BENEATH CONTINENTAL RIFT ZONES Table 2a. Events Used in $K$ Splitting Studiesfor the Baikal Rift Zone Event Origin Year Day Coordinates Time, UT Latitude, Longitude, øN øE 1804 -23.466 -177.297 1453 0308 1237 0411 0003 1021 1811 1906 0128 18.211 -22.293 -18.083 11.866 -25.849 -15.860 -10.562 -20.190 -21.702 -98.240 171.742 -172.492 -13.529 -175.911 -173.643 41.424 178.860 -176.616 Depth, Back km azimuth Distance, deg deg 79 114.12 100.71 72 33 33 11 21 128 10 538 188 23.29 121.58 107.16 299.91 114.63 106.63 242.75 114.96 112.51 107.61 93.72 99.33 97.68 103.35 96.92 82.53 95.95 99.72 99.38 1 2 3 4 5 6 7 8 9 10 1980 1980 1981 1982 1983 1984 1984 1985 1986 1986 104 298 187 153 356 272 289 134 146 303 11 1987 279 0419 -17.940 -172.225 16 106.87 12 1988 282 0446 -18.771 -172.415 35 107.54 99.91 13 1988 284 1820 -28.644 -177.553 27 117.61 104.59 14 15 1989 1990 070 004 0505 0532 -17.766 -15.397 -174.761 -172.850 230 53 108.67 105.74 97.74 97.03 16 1990 050 0648 -15.465 166.385 12 121.96 85.26 17 1990 174 2138 -21.568 -176.483 180 112.33 99.70 18 1991 305 1623 -30.255 -177.981 21 118.96 105.60 19 1992 177 0630 -28.063 -176.735 18 116.65 104.60 20 1992 178 0318 -33.682 -179.076 33 121.97 107.65 21 1992 193 1044 -22.284 -178.507 381 114.27 99.09 22 23 24 25 1992 1992 1992 1992 217 217 238 241 0658 2108 0824 1715 -21.584 -12.023 -20.620 -22.154 -177.322 166.496 -175.151 -179.620 278 109 67 597 112.96 119.97 110.75 115.02 99.22 82.46 99.73 98.36 26 1992 241 1818 -0.996 -13.557 10 292.53 108.25 27 1992 243 2009 -17.738 -178.775 573 111.69 95.36 28 29 30 1992 1992 1992 254 255 259 1043 0357 2104 -22.518 -6.091 -14.122 -175.052 26.680 167.263 39 10 196 111.87 257.19 120.51 101.26 87.58 84.61 31 1992 285 1924 -19.276 168.913 157 122.06 89.75 900 (Table2b). The stationsusedweredistributedin a has the effectof orientingmantlecrystals[Gao et al., 400by200km E-W elongated areacentered at (36.5øE, 1994c]. 0øN). Most of the stationswerewithinor very closeto the rift valley. Fourteenof the 17 stationsanalyzed 2.3. Rio Grande Rift (RGR) havea fast directionof 30 •: 20ø, whichis subparallel In the RGR study by Sandvolet al., [1992],the six to the N-S striking rift system.Thus fast directionsare stations used were within or very closeto the rift valcloserto orthogonalto the extensiondirection rather ley. The resultingfast $K$ polarizationdirectionsare than parallel to it, whichwasexpectedif the extension NNE, i.e., parallelor subparallel to the rift axis(Figure Table 2b. Events Used in $K$ Splitting Studiesfor the East Africa Rift Event Year Origin Day Time, UT Coordinates Latitude, Longitude, oN 1 2 3 4 5 6 7 8 1989 1989 1989 1989 1989 1989 1990 1990 328 342 343 350 350 354 048 073 0035 1023 2038 0033 0053 0835 0228 0344 0.989 10.094 0.141 8.431 8.396 8.192 29.533 4.575 Depth, km oE 126.007 126.495 123.340 126.942 126.848 126.852 130.732 122.620 Back deg 25 43 151 36 29 39 65 639 Distance, azimuth deg 89.0 88.8 80.0 91.1 89.8 87.6 81.6 91.5 81.7 91.0 81.9 91.0 60.6 93.3 85.4 86.7 GAOET AL.' $KS SPLITTINGBENEATH CONTINENTAL RIFT ZONES I 22,789 i 4 (A) (B) 2 0 -2 -4 -5 I o 5 lO 15 (c) 2o -5 I 10 5 _ øt _ i I i (D) -5- -10 0 o 5 10 15 20 -10 I I -5 0 5 10 Time (S) 180 •_•..•1 ,,..._f (,E) Station' tupi (E01) 120 EQ851341811 (No.08) BAZ= 254.6 o, Dist= 91.9 o 60 $= 101.0 o, St= 0.9 s (•_•)=8.0 o, (•_5t= 0.3 s 0 0 1 (•t(5;)2 3 Figure4. Diagrams showing example dataandtheprocedure forfinding theoptimal parameters forevent 8. (a)theoriginal seismograms ofradial (black) andtransverse (gray) components, (b) theirparticle motion pattern, (c) radialandtransverse components ofcorrected seismograms, (d)theparticle motion pattern ofthecorrected components, and(e)thecontour mapofanerror function, theminimum ofwhich corresponds to theoptimal parameter pair.Theseismograms in Figure4cwerecorrected fromthose shown in Figure 4ausing theoptimal parameter pair indicated bya dotonFigure 4e. Station andeventnames, backazimuth (BAZ)anddistance (DIST)oftheevent relative tothestation, aswellastheoptimal parameters andtheirerrors, areindicated at thebottom right.The95%confidence region isconfined bya thickgraycontour line in Figure 4e. 10andTablelc). The splittingtimewasfoundto range 3. from 0.9 to 1.5 s; no measurementswere made on the flanksof the rift zone. Westof the rift, in Californiaand Discussion We had expectedto find that rift zoneswould have Nevada,the dominantfast directionis east-west [e.g., fast directions in the extension direction if the manVinniket al., 1992;SavageandSilver,1993;Liu et al., tle beneaththe riffsextends, aligning olivinecrystals. 1995]. The rift-parallelfast directionwasinterpretedHowever,we havefoundthat for the Kenyaand Rio by Sandvolet al. [1992]to resultfromrift-parallelflow GrandeRiffsandpartsof the Baikalriff zone,fastdiin a small-scale mantle convectioncell beneath the rift. rectionsare orthogonalto the extensiondirection. For 22,790 GAO ET AL.' $KS SPLITTING i BENEATH CONTINENTAL RIFT ZONES i (B) 4 2 _ 0 -2 -4 i o 5 lO 15 -5 2O f -5 0 5 1 lO (D) 5 5 o 0 -5 -5 -lO I 0 5 10 - 15 -10 2O -5 0 5 10 Time (S) !--) 180 Station' tupi (E01) EQ901742138 (No. 17) 120 BAZ= 123.4 o, Dist= 93.1 o (1) = 28.0 o, St= 0.4 s (•_(1) = 104.0 o,(•_St=0.6 s 0 1(•t($)2 3 Figure 5. Sameas Figure4, exceptfor event17. Baikal, fast directions are found in the rift zone that are both parallel and perpendicular to extension. Interpretation of the S'K$ splitting measurementsis dif- ficult in that the anisotropymay be a remnantof past tectonic strain, or caused by mantle flow, or in regionswhere the mantle is melting, couldbe causedby magma-filledcracks. Below we discussthese possibilities given what has been learned from recentrift zone tomographicstudiesand work on paleostressand recent stress directions. 3.1. Tomographic Results: Hot Mantle beneath Rifts 1994b;Gao, 1995]to probethe state of the mantlebeneath these rifts. Travel times in the crust determined from refraction surveyswere removedfrom the data. The remainingsignalsare causedby broad-scalelowvelocity anomaliesin the tippermostmantle beneath the rifts whichextendto a muchgreaterdistanceei- ther side than the surfaceexpression of rifting. The P wave velocity perturbations so determinedare-12% for the EAR,-8% for the RGR, and-5% for the BRZ. The greatestamountof volcanismis associatedwith the largestvelocityanomalyin East Africa, aboutan order of magnitudelessfor the RGR, and lesseramountsin the vicinity of BRZ. Attenuation measurementsindicate Tomographicexperimentshavebeenconductedacross that high-frequency P wavesare preferentiallyabsorbed the RGR [Davis et al., 1984, 1993;Slacket al., 1996] in the low-velocitymantle anomalies[Haldermanand the EAR [Slacket al., 1994],and the BRZ [Gao et al., Davis,1991;Slacket al., 1994,1996;Gaoet aL,1994b]. GAO ET AL.' $KS SPLITTING BENEATH CONTINENTAL RIFT ZONES 53ø..•o• 22,791 '•O•o D .... ø- C 99 ø 100 ø 101 ø 102 ø 103 ø 104 ø 105 ø 106 ø 107 ø 108 ø 109 ø 110 ø A 57 ø A09A Ol A06 "1]•,, A03 •02 A00•A05: B01 Bi "..•. A10 D03 D04 D05 54 ø •D06 DO7 :"I•..D08 ß , .o ß •.o4 •o,•._.:" ........;"-- '.... F02 ,' ,' '•,F03 F04•F05 48 ø- 0.5 1.0 1.5 2.0 sec ! F {'.,........" ....'-. , F06 ••_F07 •F08 ooo0 99 ø ; i 102 ø -•F09 i 105 ø i 108 ø ..... '"" ,.• .--. i 111 ø i ,; ..... 114 o {" i 117 ø i 120 ø Figure 6. Maps showingSKS splitting resultsin the Baikal rift and adjacentareas. The diagramon the top is a two times enlargementof the rectanglein the bottom diagram. The line drawn through each.circle givesthe fast polarization direction. Those with two inconsistent results are plotted'as double circles. Stations representedby squaresare null measurementson which anisotropyeffect cannot be clearly observed. Usinglaboratoryestimatesof the variationof velocity fer to bulk averagesand do not precludethe presence of and attenuationof mantlerockswith temperatureby pocketsof melt in isolatedareascorresponding to the Satoet al. [1989],ttaldermanandDavis[1991],Slack observedpatchesof Cenozoicvolcanismin the vicinity et al. [1994,1996],andGaoet al. [1994b] findthat the of the rift [Zonenshain andSavostin, 1981]. tomographyresultscan be explainedif the mantle imRelation between Paleo and Recent mediatelybeneaththe rifts containspartial melt with 3.2. Stress Directions and Fast Directions the fraction being 2-3% beneaththe EAR, possiblya fractionof a percentbeneaththe RGR and essentially In continental rift zonestheregionalpaleostress operno melt beneath the BRZ. For the BRZ the results reating duringthe rifting processis inferredto be exten- 22,792 GAOET AL.:SKSSPLITTING BENEATH CONTINENTAL RIFTZONES sionin the directionof opening. For the EAR, Bosworth et al. [1992]use boreholebreakoutsand alignedQuaternary vents to showthat the least principal horizontal stress(LPHS) directionis alignedapproximatelyNWSE in Kenya(Figure9). In contrast,the maingeological structures 54 ø .• I) associated with the EAR which have devel- oped over the last 30 Myr have formed when the LPHS wasalignedE-W resultingin the N-S striking rift direction. Thus the rotation of the stressfield is relatively recentevent,thoughtto occurin the Quaternary(0.40.6 Ma), possiblyrelatedto the late uplift and doming of the Kenyan and Ethiopian Rifts [Bosworthet al., 1•2]. 51 ø 48' l• In a similar fashion, the stressesin the RGR and neighboring areas are thought to have undergone a clockwiserotation. Paleostressdata throughoutmuch -......... ..'.... ,-.,. of the western ß ß .: 45 ø , -• underwent United States indicate a transition from WSW-ENE that the LPHS to WNW-ESE ,,. at about 10 Myr ago, a time when Farallon subduction alongthe westernboundaryof North Americahad •'igure 7. Rosediagramsof SKS fast polarization been replacedby significantdevelopmentof the transdirections for each of the six areas in the BRZ and ad- form boundary of the San Andreas fault. The modern jacent areas. Null measurements were excludedin the LPHS direction on the Rio Grande maintains an E-W rosediagrams. The sectorwidth of the rosediagrams direction, but within the transition zone between the is 200. The two most common fast direction ranges are 400 -80 ø, which is approximatelyparallel to the rift and the Colorado Plateau, the LPHS is thought to rift axis; and 1200- 1600, which is approximatelyrift- be WNW-rSr (Figure 9) [Zobackand Zoback,1980; Aldrich et al., 1986]. The seismictomographyresults orthogonal. 60' 30' '• 20' • ,-•:.• ß •" 10' '• ,,•,,., O' , 60' ' ½'••••'•" •• ,•. • 70' ::• " .... • • •'•' ' ::' ""' ..-.... . ..--:::.., '.:. •-.,:.. t •.'•,.... '"' ' ."• -'•.. ' ' ' '"'"' ß •..:•:....•. %, ..'....,.,.-., .....,• ...: . . ....... •"'....... 80' ' •,, ,•:'..• •.'..:.-..• .• 90' • •. '...:.•.. .... 100' . . '., 110' '..,".. J.. ,• •.. ... .,,•=• ' .....• •, .•.'" . .-.," 120' 130' 140' •igure 8, Rosediagrams of SKS splittingmeasurements in the zoneof Tibet-TienShanMongolia-Baikal fromportableseismic arrays.Stationsarerepresented by opensquares.•he measurements werefromt•e followingstudies:Ma•yeva • a•. [1992](Ticn S•an); Mcnamara • a•. [1994](•i•t); 6ao • a•. [1994•]•nd t•i• •tudy(Si•ri•-•ongoli•). GAO ET AL.' SKS SPLITTING BENEATH CONTINENTAL RIFT ZONES 22,793 o o o Uganda Kenya O o .1 ø _2 ø Tanzania ooO0 _3 ø _4 o 33 ø 34 ø 35 ø 36 •, 37 ø 38 ø 39 ø 40 ø Figure 9. Mapsshowing SKS splittingresultsin the EastAfricanRift [Gaoet al., 1994c]. Arrowsrepresent regional stressdirections [Bosworth et al., 1992]. [Slacket al., 1996]reveala low-velocity bodyin theup- localscale,Sherman[1992]findsthat the SW and NE per 200 km of the mantlethat strikesNNE to NE, and regionsof the rift zone are dominated by shear at the extends from beneath the rift into the transition zone. centerwith a transition to tensionalstressesawayfrom The resultswereinterpretedasarisingfrom lithospheric the center(Figure2). The centralpart of the BRZ has extension in the modern(<;5Ma) LPHS direction,i.e., the LPHS oriented NW-SE. WNW-ESE causedby the modernstressfield [Slack et al., 1996]. The magnitudeof the velocityanomaly 3.3. Splitting from Fossil Anisotropy, and the associated attenuation in the mantle beneath Rift-Related Mantle Flow or Magmatic Cracks? the RGR were taken to imply that temperaturesin the anomalouszone are closeto, or even just above, the It is possiblefor all theserifts that SKS splitting is causedby fossil anisotropy in which LPO of olivine solidus.Thus Slacket al. [1996]concludethat it is crystalswascausedby pasttectoniceventsandhasbeen probablythe mantle sourcezoneof the most recentvol- maintained in the lithosphereever since. Each rift zone canicsthat lie alongthis trend (e.g., the Jemezlinea- wasthe siteof continentalconvergence priorto rifting. ment). However,mobility of olivine crystalsat temperatures The Baikal rift zoneis a regionof transtension.Fig- above900øCis high [Vinnik ½t al., 1992],and thereure 2 showsthe regionalstressfieldsof the BRZ [Sher- fore the survivalof fossilanisotropyis very unlikelyin man, 1992]. The stressfieldswereobtainedfrom ge- the mantle beneaththe rifts, especiallyif the tomograologicalstructuralanalysisand from earthquakefocal phy interpretationis correctsuggesting that the velocity mechanisms. Sherman [1992]findsthattheresults from anomaliesare causedby manfie at temperaturesdose the two approaches agreewith eachother. The regional to the solidus. The tomographyresultsindicate that stresspattern is alsoin agreementwith the globalstress the 900øC isotherm may upwarp to a depth of about map of Zoback[1992]and the subcrustalstresses that 50 km beneath these rifts, as suggestedby Zorin and of the layercoolerthan havebeeninferredfromsatellitegravitydata [Liu, 1978, Osokina[1984].The thickness 1983].The globalstressmap indicatesthat the LPHS 900øC is probably too small to generatethe observed lies in the direction of rift opening. However,on a more splitting if all the anisotropyis fossilanisotropy. 22,794 GAO ET AL.: $KS SPLITTING BENEATH CONTINENTAL RIFT ZONES 38 ø 36 ø 34 ø 32 ø o o00 0.5 -110 ø -108 ø -106 ø -104 ø 1.0 1.5 2.0sec -102 ø Figure 10. MapsshowingSKS splittingresultsin the Rio GrandeRift [Sandvolet al., 1992]. Arrowsrepresent regionalstressdirections [Zoback andZoback,1980;Aldrichet al., 1986]. An alternative model holds that the fast directions of partial melt to generatethe observedanisotropy,proare caused by mantle flow. Three-dimensionalsmall- videdcrackaspectratiosare high [O'ConnellandBudiscaleconvectionin the mantle as suggestedby Sandvol ansky,1977;Anderson,1989]. This modelcan explain et al. [1992]for the RGR, or by Makeyevaet al. [1992] both the P wave delays and why the fast directions to explain anomalousdirectionsin the Tien Shan, could appear to be better correlated with the recent stressdi- give rise to rift-parallel directions.At this stageof our rections on the RGR and EAR rather than the stress understandingof mantle processesbeneath rifts, it is directionsrelated to the original rift events. We expect difficult to rule out this possibility. that the crack anisotropy would respondrapidly to a A third model, which may be applicablein regionsof regionalstressrotation. hot mantle beneath rift zones,holds that aligned magIn that the tomographyresultsimply that the BRZ matic cracksin the mantle give rise to the fast direc- mantle is cooler than either the EAR or RGR, the mixtions. The effecton $KS splittingis similar to that of ture of rift-parallel and rift-orthogonaldirectionsin the fluid-filledcracksin the rigiduppercruston splittingof immediate vicinity of that rift may arisefrom a combishearwavesfrom local earthquakes[e.g., Leafy et al., nation of LPO and orientedcracks,dependingon man1990;Liet al., 1994].Parallelverticalmagmaticcracks tle temperature. However,we add that resolvingbeform a transverseisotropywith rift-orthogonalaxis of tween fossilanisotropy,recent mantle flow, or aligned symmetry. The fast direction of the anisotropyis par- cracksis sufficientlydifficult that favoringone over the allel to the strike of the crac. ks, i.e., parallel to the rift other cannot be made with a high degreeof certainty. axis. If the melt is distributedin crackssuchasalonggrain- 4. grain contactsor in planar dike-likeregions,in a region of extension Conclusions there will be a bias in the distribution of Observationsof $KS splitting on the East African crackplaneswith greater numbersorientedperpendicu- and Rio Grande Rifts give fast directionsthat are orlar to the extension.It takesjust a fractionof a percent thogonal to the least principal horizontal stress. For GAO ET AL.: $KS SPLITTING BENEATH CONTINENTAL RIFT ZONES 22,795 calculated for ultramafic minerals and aggregates,in Flow and Fracture o• Rocks,Geophys. Monogr. Set., Vol. 16, with fast directionsboth orthogonaland parallelto edited by H. C. Heard et al., pp. 157-166, AGU, Washthe leastprincipalhorizontal stress.Tomography reington, D.C., 1972. sults and rift-related volcanismindicatethat the man- Bjarnason, I. T., C. J. Wolfe, S.C. Solomon, and G. Gudmundson, Initial results from the ICEMELT experimenttle underthe EAR is hottestand probablyabovethe Body-wave delay times and shear-wave splitting across solidus, thatundertheRioGrande it isnearthesolidus, Iceland, Geophys. Res. Lett., œ3,459-462, 1996. whereasthat beneath Baikal it is below the solidus. Bosworth, W., M. R. Strecker, and P.M. Blisniuk, InteSuchhot mantlemaynot develop LPO fabricdueto angration of east African paleostressand present-day stress nealing effects such asrotational recrystallization [Nico- data: Implications for continental stress field dynamics, las andPoirier,1976]. Magmaticcracksin partially J. Geophys. Res., 97, 11,851-11,865, 1992. moltenmantlecouldexplainthe splittingresultssince Chastel, Y. B., P. R. Dawson, H. R. Wenk, and K. Bennett, Anisotropic convection with implications for the upper a biasin crackpopulation isexpected withmorecracks mantle, J. Geophys.Res., 98, 17,757-17,771, 1993. oriented perpendicular totheLPHS.ForRGRandEAR Christensen,N. I., Shear wave propagationin rocks, Nature, Baikal,in thevicinityof therift, theresults aremixed thefastdirections appearto havefollowed therecentroœœ9,549-550, 1971. tationof thestressfield. Thisisreadilyexplained if they Christensen, N. I., The magnitude, symmetry and origin arecaused by a corresponding redistribution of oriented of upper mantle anisotropy based on fabric analyses of cracks. For Baikal the inferred coolermantle may be ca- pableof generating a fabricawayfromtherift zone,as observed, but in patchesof hot mantlenearthe rift, magmatic crackscouldaccount for the rift-parallel directions.If mantlemagmaticcrackscauseanisotropy in regions of extension, thiscanexplainwhyobserved fast directions areparallelto LPHSin compressive zonesbut perpendicular to it in extension zones.In the former, temperatures areloweranda fabriccandevelop. In the latter, temperatures arehigher,possibly annealing out the fabricbut causing meltpocketsto formwhichgive rise to volcanismand alignedcracks. Acknowledgments. The fieldworkfor the Baikalrift projectwasdonein conjunction with the late BobMeyer, Universityof Wisconsin, whosepresence is sorelymissed. We dedicatethis paperto his memory.The analogseismo- gramswererecorded by the BaikalSeismic Network of the RussianAcademyof Sciences whichhasbeenmanagedby a teamheadedby O. Masalskiof IEC. We thankR. Meyer's groupmembers, especially P. Burkholder andL. Delitsin for discussions and field cooperation.The authorsare grateful to R. Girdlerfor helpfuldiscussions. D. Helmberger andX. Dingat Caltechkindlyprovided assistance for the digiti- ultramafic tectonites, Geophys. J. R. Astron. $oc., 76, 89-111, 1984. Christensen, N. I., and S. Lundquist, Pyroxene orientation within the upper mantle, Geol. Soc. Am. Bull., 93, 279288, 1982. Christensen, N. I., and M. H. Salisbury, Seismic anisotropy in the oceanic upper mantle: Evidence from the Bay of Islands ophiolite complex, J. Geophys.Res., &t, 4601-4610, 1979. Crosson,R. S., and J.-W. Lin, Voigt and Reussprediction of anisotropic elasticity of dunite, J. Geophys. Res., 76, 570-578, 1971. Davis, P.M., E. C. Parker, J. R. Evans, H. M. Iyer, and K. H. Olsen, Teleseismicdeep sounding of the velocity structure beneath the Rio Grande Rift, Field Con]. Guide B., N.M. Geol. Soc., 35th, 29-38, 1984. Davis, P.M., P. Slack, H. A. Dahlheim, W. V. Green, R. P. Meyer, U. Achauer, A. Glahn, and M. Granet, Teleseismic tomography of continental rift zones,in Seismic Tomography: Theory and Practice, edited by H. M. Iyer and H. Hirata, pp. 397-439, Chapman and Hall, New York, 1993. Forsyth, D. W., The early structural evolution and anisotropy of the oceanic upper mantle, Geophys. J. R. Astron. Soc., 4{3, 103-162, 1975. Fowler, C. M. R., The Solid Earth, An Introduction to Global Geophysics,472 pp., Cambridge Univ. Press, New York, 1990. zation. Most of the diagramsin this paperwereproduced Francis, T. J. G., Upper mantle structure along the axis by GMT version 3.0, a freegeneric mapping tool [Wessel of the Mid-Atlantic Ridge near Iceland, Geophys. J. R. andSmith,1991].Reviews fromD. Blackman, A. Nyblade, Astron. Soc., 17, 507-520, 1969. and R. Russoare gratefullyacknowledged. The studywas Fuchs,K., Seismicanisotropyof the subcrustallithosphere supported by DARPAundercontracts F2901-91-K-DB17 as evidencefor dynamical processesin the upper mantle, and F49620-94-1-0161 to UCLA and by ISF under grants Geophys.J. R. Astron. Soc., •9, 167-179, 1977. RLN000 and RLN300 to IEC. R. Russo is thanked for as- Gao, S., Seismic evidence for small-scale mantle convection sumingtheroleof Editor,JGR-Solid Earth,forthereview under the Baikal rift zone, Siberia, Ph.D. thesis,221 pp., process. Univ. of Calif., Los Angeles,1995. Gao, S., P.M. Davis, H. Liu, P. D. Slack, Y. A. Zorin, V. V. Mordvinova, V. M. Kozhevnikov,and R. P. Meyer, References Seismic anisotropy and mantle flow beneath the Baikal rift zone, Nature, 371, 149-151, 1994a. Aldrich,M. J., Jr., C. E. Chapin,andA. W. Laughlin,Stress history and tectonicdevelopmentof the Rio GrandeRift, Gao, S., P.M. Davis, H. Liu, P. Slack, Y. A. Zorin, N. A. Logatchev,M. Kogan, P. Burkholder, and R. P. Meyer, New Mexico, J. Geophys.Res., 91, 6199-6211,1986. Asymmetric upwarp of the asthenospherebeneath the Anderson,D. L., Theory of the Earth, 366 pp., Blackwell Baikal rift zone, Siberia, J. Geophys.Res., 99, 15,319Sci., Cambridge, Mass., 1989. 15,330, 1994b. Ando, M., Y. Ishikawa,and F. Yamazaki,Shearwavepolarization anisotropyin the upper mantle beneath Honshu, Gao, S., P.M. Davis, H. Liu, P. D. Slack,Y. A. Zorin, V. V. Mordvinova, V. M. Kozhevnikov,and R. P. Meyer, SeisJapan, J. Geophys.Res., 88, 5850-5864,1983. mic Anisotropy beneath the Baikal and Kenya rift zones Babuska,V., and M. Cara, SeismicAnisotropyin the Earth, (abstract),Eos Trans.A GU, 75(3),West. Pac. Geophys. 217 pp., Kluwer Acad., Norwell, Mass., 1991. Baker,D. W., and N. L. Carter, Seismicvelocityanisotropy Meet. Suppl., 67, 1994c. 22,796 GAO ET AL.: $KS SPLITTING BENEATH CONTINENTAL RIFT ZONES Halderman, T. P., and P.M. Davis, Qr beneath the Rio Grande and East African Rift zones, J. Geophys.Res., 96, 10,113-10,128, 1991. Hess, H. H., Seismic anisotropy of the upper most mantle under oceans, Nature, œ03,629-631, 1964. Humphrey, N., and D. Helmberger, NXSCAN, IRIS Data Manage. Cent., Seattle, Wash., 1993. Karato, S., Seismic anisotropy: Mechanisms and tectonic imphcations, in RheologyoI solidsand oj' the Earth, edited by S. Karato and M. Toriumi, pp. 393-422, Oxford Univ. Press, Oxford, 1989. Kendall, J. M., Teleseismicarrivals at a mid-oceanridge: Effects of mantle melt and anisotropy, Geophys. Res. Lett., 21,301-304, 1994. Ribe,N.M., On the relationbetweenseismicanisotropy and finite strain, J. Geophys.Res.,97, 8737-8747,1992. Ribe, N.M., and Y. Yu, A theoryfor plasticdeformation andtexturalevolutionof ohvinepolycrystals, J. Geophys. Res., 96, 8325-8355, 1991. Sandvol,E., J. Ni, S. Ozalaybey,and J. Schlue,Shear-wave sphttingin the Rio GrandeRift, Geophys. Res.Left., 19, 2337-2340, 1992. Sato,H., I. S. Sacks,andT. Murase,The useof laboratory velocitydata for estimatingtemperatureand partial melt fractionin the low-velocityzone: Comparisonwith heat flowand electricalconductivity studies,J. Geophys. Res., 9•, •½8•-•704, •8•. Savage,M. K., and P. G. Silver,Mantle deformationand tectonics:Constraintsfrom seismicanisotropyin the western Kind, R., G. L. Kosarev, L. I. Makeyeva, and L. P. Vinnik, Observationsof laterally inhomogeneousanisotropyin the United States,Phys. Earth Planet. Inter., 78, 207-227, 1993. continental hthosphere, Nature, 318, 358-361, 1985. Leafy, P. C., S. Crampin, and T. V. McEvilly, Seismicfrac- Savage,M. K., P. G. Silver,and R. P. Meyer,Observations ture anisotropy in the Earth's crust: An overview, J. Geoof teleseismic shear-wave sphttingin the basinrangefrom phys. Res., 95, 11,105-11,114, 1990. portable and permanentstations,Geophys.Res. Left., 17, 21-24, 1990. Li, Y. G., T. L. Teng, and T. L. Henyey,Shear-wavesphtting observationsin the northern Los AngelesBasin, southern Sherman,S. I., Faults and tectonicstressesof the Baikal rift California, Bull. Seismol. Soc. Am., 8•, 307-323, 1994. zone, Tectonophysics, 208, 297-307, 1992. Liu, H. S., Mantle convectionpattern and subcrustal stress Silver, P. G., Seismicanisotropybeneath the continentsfield under Asia, Phys. Earth Planet. Inter., 16, 247-256, Probingthe depthsof geology, Annu. Rev. Earth Planet. 1978. Sci., 2•, 385-432, 1996. Liu, H. S., Geodynamicsof the Baikal-Stanovoyseismicbelt, Silver, P. G., and W. W. Chan, Imphcationsfor continental Phys. Earth Planet. Inter., 31, 77-82, 1983. structureand evolutionfrom seismicanisotropy,Nature, 335, 34-39, 1988. Liu, H., P.M. Davis, and S. Gao, SKS sphtting beneath southern California, Geophys. Res. Lett., œœ,767-770, Silver,P. G., and W. W. Chan, Shearwavesplittingand 1995. subcontinental mantledeformation,J. Geophys.Res.,96, 16,429-16,454, 1991. Makeyeva, L. I., L. P. Vinnik, and S. W. Roecker., Shearwave splitting and small-scaleconvectionin the continen- Silver, P. G., and S. Kaneshima, Constraints of mantle anisotropy beneath Precambrian North America from a tal upper mantle, Nature, 358, 144-147, 1992. McKenzie, D., Finite deformation during fluid flow, Geotransportableteleseismic experiment,Geophys.Res. Left., phys. J. R. Astron. $oc., 58, 689-715, 1979. 20, 1131-1134, 1993. McNamara, D. E., T. J. Owens, P. G. Silver, and F. T. Wu, Slack,P. D., P.M. Davis, H. A. Dahlheim,A. Glahn, J. Shear wave anisotropy beneath the Tibetan Plateau, J. R. R. Ritter, W. V. Green,P. K. H. Maguire,and R. P. Geophys. Res., 99, 13,655-13,665, 1994. Meyer, Attenuation and velocityof P-wavesin the mantle Mercier, J. C. C. Olivine and pyroxenes,in Preferred Oribeneaththe East African Rift, Kenya, Tectonophysics, entation in Deformed Metals and Rocks:An Introduction 236, 331-358, 1994. to Modern Texture Analysis, edited by H. Wenk, pp. 407- Slack,P. D., P.M. Davis,W. S. Baldridge,K. H. Olsen,A. 430, Acad. Press,Orlando, Flor., 1985. Montagner,J., and T. Tanimoto,Globaluppermantletomographyof seismicvelocitiesand anisotropies,J. Geophys. Res., 96, 20,337-20,351,1991. Glahn, U. Achauer,and W. Spence,The upper mantle structureof the centralRio GrandeRift regionfrom teleseismicP and S wavetraveltime delaysand attenuation, J. Geophys.Res., 101, 16,003-16,023,1996. Nicolas,A., andN. I. Christensen, Formation of anisotropy Tanimoto,T., and D. L. Anderson,Lateralheterogeneity and azimuthal anisotropyof the upper mantle: Loveand in uppermantleperidotires:A review,Geodynamics Set., 16, 111-123, 1987. Rayleighwaves100-250s, J. Geophys.Res., 90, 18421858, 1985. Nicolas,A., F. Boudier,and A.M. Boullier,Mechanisms of Applicaflowin naturallyandexperimentally deformed peridotires, Turcotte, D. L., and G. Schubert,Geodynamics, Am. J. $ci., 273, 853-876, 1973. tions of continuumphysicsto geological problems,John Wiley, New York, 1982. Nicolas,A., andJ.P. Poirier,CrystallinePlasticityandSolid StateFlowin MetamorphicRocks,444 pp., JohnWiley & Verma, R. K., Elasticityof severalhigh-densitycrystals,J. Sons, London, 1976. Geophys.Res., 65, 757-766,1960. O'Connell,R. J., and B. Budiansky, Viscoelastic properties Vinnik, L. P., V. Farra, and B. Romanowicz,Azimuthal anisotropy in the Earth from observationsof SKS at Geoof fluid-saturatedcrackedsohds,J. Geophys.Res., 82, 5719-5735, 1977. scopeand NARS broadbandstations,Bull. Seismol. Soc. Am., 79, 1542-1558, 1989. Olsen,K. H., and P. Morgan,Introduction:Progress in understanding continentalrifts,in ContinentalRifts: Evolu- Vinnik, L. P., L. I. Makeyeva,A. Milev, and A. Y. Usenko, Global patternsof azimuthalanisotropyand deformations tion, Structureand Tectonics, editedby K. H. Olsen,pp. 3-26, Elsevier, New York, 1995. in the continentalmantle,Geophys.J. Int., 111,433-447, Peselnick,L., A. Nicolas,and P. R. Stevenson, Velocity 1992. anisotropyin a mantle peridotire from the Ivrea Zone: Wessel,P., and W. H. F. Smith,Freesoftwarehelpsmapand displaydata, Eos Trans.A GU, 72, 441,445-446, 1991. Applicationto uppermantleanisotropy, J. Geophys. Res., 79, 1175-1182, 1974. Raitt, R. W., G. G. Shot Jr., T. J. G. Francis,and G. B. Morris, Anisotropyof the Pacificuppermantle,J. Geophys. Res., 7•, 3095-3109, 1969. Zoback,M. L., First- and second-order patternsof stressin the hthosphere:The World StressMap Project, J. Geophys. Res., 97, 11,703-11,728,1992. Zoback, M. L., and M.D. Zoback,State of stressin the GAO ET AL.: SKS SPLITTING BENEATH CONTINENTAL conterminous United States, J. Geophys.Res., 85, 61136156, 1980. Zonensha.in, L. P., and L. A. Savostin, Geodynamics of the Baikal rift zone and plate tectonics of Asia, Tectonophysics, 76, 1-45, 1981. Zorin, Y. A., and S. V. Osokina, Model of the transient temperature fields of the BaikM rift hthosphere, Tectonophysics, 103, 193-204, 1984. RIFT ZONES 22,797 hliu@dtm.ciw.edu; pslack@ess.ucla.edu; arigor@ess.ucla.edu) S.Gao, Department of Terrestrial Magnetism, Carnegie Institution of Washington,5241 Broad BranchRoad, N.W., Washington,DC 20015. (e-mail:sgao@dtm.ciw.edu) Y. A. Zorin, V. V. Mordvinova, V. M. Kozhevnikov, and N. A. Logatchev, Institute of the Earth's Crust, Siberian Branch• Russian Academy of Sciences, 128 Lermontov Street, Irkutsk 664 033, Russia. (e-mail: zorin@crust.irkutsk.su) P.M. Davis, H. Liu, P. D. Slack,and A. W. Rigor, De(ReceivedDecember3, 1996; revisedMay 30, 1997; nia, Los Angeles,CA 90095. (e-mail: pdavis@ess.ucla.edu;acceptedJune 24, 1997.) partment of Earth and Space Sciences,University of Califor-