Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 100, NO. B2, PAGES 2261-2278,FEBRUARY 10, 1995 Extinct spreading center in the Labrador Sea: Crustal structure from a two-dimensional seismic refraction velocity model John C. Osier• and Keith E. Loudcn Departmentof Oceanography,DalhousieUniversity,Halifax, Nova Scotia,Canada Abstract.TheLabradorSeacontains a rareexampleof anabandoned mid-ocean ridgewhere activeaccretion of oceanic crustceased dueto a change in thespreading geometry of lithospheric plates.Seismic refraction datawerecollected alonga linewhichtransversely crosses theextinct spreading center.Two-dimensional analyses of therefraction data,usingmy-tracing andsynthetic seismogram techniques revealmajorvariations in crustalthickness andvelocityin relationto the axisof theextinctspreading center.In theextinctspreading center,a crustalthickness of approximately 4 km is determined, compared with5.5km for theflanks.Substantial lateral variations in P wavev½locities of theupperandlowercrustareobserved witha markeddecrease withintheextinctspreading center.Low v½1ociti½s arealsoobserved in theuppermost manfie underlying theextinctspreading centerandareinterpreted asbeingtheresultof hydrothermal alteration. The anomalously low crustalv½1ociti½s andcrustalthinningareattributed to a decreasing supplyof partialmeltandincreasing degreeof tectonism at theslowspreading rates preceding extinction. Introduction syntheticseismogram techniques,showsthe crustwithin the extinctspreading centerto be thin andof anomalously low P Mid-ocean ridgesmarktheboundaries between diverging wave velocity relative to its flanks. Likewise, low velocities lithospheric plateswherenewoceanic crustandlithosphereare observedin the uppermostmantle,underlyingthe crust areformed.Extinctspreading centers aremid-ocean ridges within the extinct spreading center. Two potential wherethe active accretionof oceaniccrusthasceaseddue to a mechanismsfor forming these crust and upper mantle structuresare explored.The preferredinterpretation is that decreasesin spreadingrate led to an enhancedconductive cooling of the crust renderingit brittle and subjectto tectonism. The tectonism produces low crustalvelocitiesby fracturingthe crust and may causesomecrustalthinning throughthe rotationof crustalblockson major faults.The tectonismpromotesdeeppenetrationof hydrothermal fluids reducingpartial melt supplywhichyieldsthinnercrust.The circulation penetrates throughthecrustallowing theCentralBasinFault[LewisandHayes,1980];theShikoku hydrothermal Ridge [Tomodaet al., 1975];andridgesin the Tasman serpentinizationof the upper mantle therebyreducingits change in thespreading geometry of thelithospheric plates. Theyarelarge-scale features formed byridgejumpsin excess of 400 Ion betweentheextinctspreading centerandthenew locus of rifting[Batiza,1989].Theycanbeidentified bytheir morphological expression [Mammericloc andSandwell,1986] and magneticanomalypattern.Extinctspreading centers identified onthisbasisaretheMathematician Ridge[Klitgord andMammerickx, 1982];theGalapages Ridge[Anderson and Sclater,1972];theAegirRidge[TalwaniandEMholm, 1977]; [Weissel andHays,1977],Coral[Weissel andWatts,1979] velocity. The anomalous crustat the extinctspreading centerformed andLabradorSeas[Srivastava et al., 1981].Few refraction studies havebeenconducted at these extinct spreading centers. in the time interval from Chron 21 until its extinction. PhilippineSea, and seismicrefractionlines have been shot protractedover the periodfrom Chron21 to 13. The variation Nagumo et al. [1980]published refraction profiles samplingModeling of magnetic anomaly patterns [Osler, 1993] but was theuppercrestin thevicinityof theShikoku Ridgein the suggeststhat spreadingdid not ceaseinstantaneously in the observedcrustalstructurein relationto the palcorift thismaterial remains unpublished (R.B. Whitmarsh, personal axis supportsthe idea of a systematicrelationshipbetween acrossthe Aegir Ridgein theNorwegian-Greenland Sea,but slowspreading ratesand(1) increasedtectonism;(2) decreased partial melt supply; (3) hydrothermal alteration of the Thispaperpresents a two-dimensional analysis of seismic uppermost mantle; and (4) crustal thickness. refractiondata collectedon a line which traversesthe extinct communication,1992). spreading center in theLabrador Sea.Thetwo-dimensional (2- D) velocity-depth structure, developed by ray-tracing and GeologicalSetting •Now atDefence Research Establishment Atlantic, Darmouth, Nova Scotia, Canada. Copyright1995by theAmericanGeophysical Union. Paper number 94JB02890. 0148-0227/95/94JB-02890505.00 The Labrador Sea is a small ocean basin conf'med between the coasts of Labrador and Baffin Island to the west and Greenlandto the east (Figure 1). The presenceof a buried extinctridgein the LabradorSeawasfirst suggested by Drake et al. [1963] basedon seismicreflectionprofiles,althoughat that time the oceanicaff'mity of the crust in the basin was unclear[Van der Linden, 1975]. From detailedanalysisof the 2261 2262 OSLER AND LOUDEN: EXTINCT SPREADING CENTER IN LABRADOR SEA 60 55 65 60 55 50 45 Figure 1. Locationmap for the seismicrefractionexperimentin the LabradorSea.The thick solid lines are refractionprofiles; open crossesare locationsof oceanbottom seismometer(OBS) receivers.Thinner solid lines are fracturezones[Roestand Srivastava,1989b].Selectedbathymetrycontourlevels(as indicatedin meters) are shadedin grey. Selected magnetic anomaly linearionsand correspondingChronological identifiersare from Srivastavaet al. [1988]. OceanDrilling Program(ODP) Sites646 and 647 are indicated by stars. magnetic anomalies in the basin, its seafloor spreading history has been most recently documentedby Roest and $rivastava [1989a], as follows.After a periodof continental stretching,spreadingwas initiated southof the Cartwright FractureZone at Chron34 (84 Ma, followingthe timescaleof Kent and Gradstein[ 1986]) andwas established in the northern LabradorSea, southof Davis Strait, by Chron 31 (69 Ma). Spreadingcontinuedin an east-northeast directionuntil Chron [Kristoffersenand Talwani, 1977]. Spreadingrate historiesare providedby Roestand Srivastava[1989a] for a sectionof the extinct spreadingcenternorthwestof the Snorri fracturezone (Figure 1) and by Osler [1993] for the segmentof the extinct spreading centerunderconsideration in thisstudy.Roestand Srivastava [1989a] proposehalf spreadingratesof 18.2, 9.4, and 2.7 mm/yr for Chrons25 to 24, 24 to 21 and 21 to 13, respectively. Osler [1993] proposeshalf spreadingrates of 25 (59 Ma) whena majorreorientation of spreading occurred. 6.9 and 7.0 mm/yr for Chrons 25 to 24 and 24 to 21, The reorientation to a moreobliquespreading direction(north- respectively,followedby a lineardecreasein spreadingrate as northeast)was coincidentwith the onsetof hot spot activity a function of time from 5.01 to 0 mm/yr between Chrons21 in Davis Strait [Hyndman, 1973;$rivastavaet al., 1989] and and 13. The modelingof spreadingrate historiesis hampered the separationof Greenlandfrom Eurasia.A triple junction by the low amplitudeof the magneticanomaliesfrom Chron south of Greenland existed during simultaneousseafloor 24 to the extinct spreadingcenter.They may be muted due to spreadingbetweenNorth AmericaandGreenlandand Greenland the probability of rocks with mixed geomagneticpolarities andEurasia,betweenChron24 (56 Ma) andChron20 (45 Ma) occurringin a verticalcolumnat slowspreading rates[Denham Spreading continued in theLabradorSeauntilChron20 (45 and Schouten,1979] or greatersegmentation of the ridge axis Ma, Figure 2) and was abandonedsometimebeforeChron 13 during obliquespreadingjuxtaposingcrustalblocksof normal (36 Ma) which is the first anomalyin the North Atlantic andreversedpolarityacrossmany smalloffsettransformfaults parallelto andcontinuous withthepresent Mid-AtlanticRidge [Roots and Srivastava, 1984]. OSLER AND LOUDEN: EXTINCT SPREADING CENTER IN LABRADOR SEA 2263 1978]. A reexaminafionof the magneticanomalies[Roestand Srivastava, 1989a], following the collectionof the refraction Two longrefractionlineswereshotat theextinctspreading data, alteredthe strike of this flow line by approximately20ø centerin the LabradorSea(Figure2) in 1987.Line R1, 150km to the north. Hence line R2 does not exactly follow a flow in length,was situatedto samplecrustalongits strike,and line. The refraction lines were positioned to avoid the known line R2, 220 lcmin length,crossedit transversely to sample crustawayfrom the areaof the extinctspreading centerout to fractureszonesin the area(Figure 1): the SnorriFractureZone Experimental Procedure Chron 24 in order to measure variations in crustal thickness as to the northwest and the Minna and Julianhaab fracture zones the extinct spreadingcenteris approached.R2 was oriented to the southeast[Srivastavaet al., 1988]. The determination parallel to the generaldirectionof spreadingfollowing the of fracturezone locationsin the LabradorSea is guidedchiefly major reorientation at magnetic anomaly 25 [Srivastava, by their signaturein the free-air gravity field, where they 56 W 52W 54 6O N ....... ...... ......... _ - .... :::::::::::::::::::::::::::::::: :::- .. _: ................ "':.::'(i .. 7 :i. . -..-... :•.."• ... ..... ::.. '-".'2:::: ........ ....... •, .'2':':'.'::: . :-:-_.:.' .............................. ß ;:-::..{:'!.'" % ..... .- -.:•..;..:........... .-i'i!ii.' .............. .......... :-. .............. :;iii:ii!ii•iE';:: .... •.<;:i•:.•::::"-w:ii•:::i.::':... -.--•-•-•.•:.. ,!•!•i•...'..i•i.,'.i•!•.•::.:...•..::-.,::.:..{-::•.:.-•R::!'i: I '?:: :'"i.'•:.'::•;{:.:::: ...:: :.-::::::::::: i.::':!-!.:::• ....... ! :•.:-•.-':•ii•:! ... 59 .. :..:... .......... . ::.:..:.: ':'!lil...:.:.:.:.:.:.:.:.:.:.:.:iR.:.:...-_-.: .:: if::.: .:. '.:::::::2::::' ......... ... ß \,. .. -.:..-.:..- \ \.\ 58 ::.:.:.:::.:.:::. =========================================================== ......................... '2 .......................................... 57N -40 -30 -20 -10 0 10 20 30 mGals Figure 2. Shot point and receiverlocations(OBSs, solid letteredcircles,and sonobuoys, opennumbered squares) for the seismic refraction experiment. The axis of the extinct spreadingcenter is along the prominent low in the free-air gravity anomaly [from Woodside, 1989]. Magnetic lineafions and chronologicalidentifiersfollow the interpretationof Roest and $rivastava [1989a]. 2264 OSLER AND LOUDEN: EXTINCT SPREADING CENTER IN LABRADOR SEA componentmay be up to 1/4 of a seismic period between successiveshots,but its effect is removed by trace mixing. Shot point locations were obtained using a combinationof Loran C and transit satellite fixes, smoothedby a running seismically determined crustalstrdctures alongtherefraction mean applied along a 6-min window. Instrument locations lines is thusconsideredminimal but carmotbe excluded. were determinedby creatinga referencegrid andthensearching Free fall/pop-up ocean bottom seismometers (OBSs) this grid to find the location where there is a least squares minimumbetweenrangesfrom water wave arrival times,ray [Loncarevic, 1983] from Dalhousie University and the traced through a water velocity structureto their respective Atlantic Geoscience Center (Bedford Institute of Oceanography)were deployedat locations(Figure 2) along shotpoints, and rangesfrom a positionon the referencegrid lines R1 (five OBSs) and R2 (eight OBSs). The spacing to the shot points. A detailed discussion and listing of between OBSs was approximately 20 km at the extinct uncertaintiesin OBS and shot point positionsis given by spreading center and up to 50 km on the flanks. In the Osler [ 1993]. following sections,each seismic record is referred to by the Two-DimensionalVelocity Structure of Line R2 OBS letter and the directionof shooting(e.g., OBS Msw ). An From Travel Time Modeling array of 6 x 16.4 L (6 x 1000 in3) air guns was used as the seismicsourcewith a traceseparationof approximately250 m Model Parameterization (shot point locationsin Figure 2). Single-channelreflection TheRAYINVR packageof ZeltandSmith[ 1992]wasusedin data were collected simultaneously. Disposable sonobuoys developing a 2-D seismiccrustalstructure. Computed arrival were deployedbetweenOBS positions(Figure 2) andrecorded times for user selected refracted, head wave or reflected phases on analogFM tape. Two of these,S/B 11 and 12, were usedin axedetermined by raytracingthroughan irregularnetworkof the subsequent analysisbecauseof a failure of one OBS. blockswhichareusedto parameterize the For each OBS, timing correctionsare appliedto accountfor layeredtrapezoidal The ray tracingthroughthevelocitymodel (1) offsets and drifts in the OBS clock and (2) offsetsbetween velocitystructure. usingzero-order asymptotic ray theory[Cerveny analog data channelsdue to tape head misalignment.The first is performed appear as prominent, low-amplitude lineations. In positioningthe refraction lines, theseprominentfeaturesin the free-air gravity anomaly field were avoided. The possibility that fracture zone features are influencing the et al., 1977]. Criteria which guidethe specification of nodes for eachlayer are (1) ray coverage;(2) auxiliarydatasources (e.g., 12 kHz bathymetry and single channel seismic component, whichisdetermined every24hours byapplying a reflectiondata); and (3) trade-offsbetweenRMS travel time TRMs, andresolution. The numberanddistribution calibration tone to all channels, and a dynamic component residuals, that arisesfrom irregular tape transport.The offset due to this of nodesalongline R2 areshownin Figure3. correction could leave a travel time error of no more than 12 ms, which is equivalent to one sample point and is not significant.The secondcorrectionis treatedas having a static Line R2: Parameterization of Velocity-DepthStructure Southwest 0 40 Distance (km) 80 120 Northeast 160 200 0 . 10- 15•, 20o o _Dept.h o Dept•andVelocity Above 1- Water .-. o _Dept.h andVelocityBelow 4,5,6,7- Crust * Depth andVelocityAbove andBelow 8- Mantle Figure 3. The2-Dmodel of thevelocity-depth structure along refraction lineR2isparameterized bya series of layered trapezoidal blocks. Eachlayerisdefined bya series of boundary nodes alongitsupper surface and velocitynodesalongitsupperandlowersurfaces (values for eachnodeaxelistedin Table2, electronic supplement). OBSandsonobuoy positions areindicated bystarsandcircles, respectively. OSLER AND LOUDEN: EXTINCT SPREADING Employingthesecriteriaresultsin a basicalongtracknodal spacingof 10 km, althoughthereare severalareasalongthe refractionline with a differingnodaldensity.A first example is thesediment-crust boundary(betweenlayers3 and4 or 3 and 5 in Figure 3) wherenodesare specifiedevery 2.5 km as requiredby observedtravel time variations(e.g., oscillation of the observed travel time of Pn arrivalsfor OBS Msw between60 and 80 km in Figure9a). Anotherexampleis the absenceof boundaryand velocity nodesin the crustbetween 30 and60 km (layers5 and7 in Figure3), wherethepaucityof raypaths traveling through this region precludes the specificationof nodes.A third example is the lower crust withinthe extinctspreading centerfrom 70 to 90 km (Figure 3), where lateral velocity variationsare moderateand do not CENTER IN LABRADOR SEA 2265 Vuppe r andVlowe r arethevelocities at thetopandbottomof the layer respectively,the normal incidencetravel time, AT, within a layer is Vupper (1) The layerdepthandlowerboundary velocityare AT(Vlower-Vupper ) Vupper Zlower =In(1+Vlower_Vu ) (2) requirea smallernodal spacing. Vlowe r = Vuppe r + kZlower. Initial Velocity-Depth (3) Model The water colmnnis treatedas a singlelayer with a uniform velocity of 1.49 km/s. Bottom boundary nodes were determinedfrom the 12-kHz bathymetry and specified every 2.5 km. The sedimentcolumn is divided into two layers. The farst layer is between the seafloor and an internal sediment reflector at an averagetwo way travel time of 0.68 s below the seafloor, maxking the transition from the reflective upper section to the transparent lower section (Figure 4). This For the upper sedimentlayer, the velocity log at ODP Site 646 reflector Vupper--2.2 km/sandk=0.2s-1.Thenodes computed withthese correlates with reflector R1 identified on other reflectionprofriesin the LabradorSea [Srivastavaet al., 1989] and correspondsto the bottommostoccurrenceof ice-rafted debrisat OceanDrilling Program(ODP) Site 646, datedas mid to late Pliocene. The second layer is defined between the internal sediment reflector and the sediment-crustboundary. For a linear increasein velocity as a functionof depth,wherek is the velocity gradient, Z•oweris the layer thickness, and [Jarrardet al., 1989,Figure16]is usedto assign Vupper=l.5 km/s and k=0.96 s-1. Using thesevalues, the computeddepth to the internal sediment reflector averages 0.63 km for refractionline R2, which coincideswith the depthat which the velocity gradientdecreasesmarkedlyin the Site 646 velocity log. For the lower sedimentlayer, the observedtravel times from rays refractedwithin the layer are usedto selectvaluesof values require little modification in the modeling, except in the areaof theextinctspreading center,whereVuppe r andk both increase. The specificationof boundaryand velocity nodesfor the upper crust, lower crust, and mantle layer is based on 1-D travel time solutions.For each OBS, the x-p velocity-depth structures[Osler and Louden,1992] are dividedinto upperand Line R2 reflection profile 25km SW SP lOO 200 300 400 500 600 700 NE 800 4.0 5.0 6.0 7.0 Extinct Rift Figure 4. Single-channelreflectionprofile alongline R2. Individual tracesare band-passfiltered from 6 to 36 Hz, amplitudesquaredto increasesignal to noise, and multiplied by a time-varyinggain function to compensatefor energy loss due to spreading.OBS locationsare indicatedby solid circles; pointerslocate sediment-basement (B) and upperand lower sediment(S) interfaces.Vertical exaggerationis 30:1. 2266 OSLER AND LOUDEN: EXTINCT SPREADING lower crustallayers at a depthwhere the velocity gradientmay be split into two linear segments offset by a velocity discontinuity,if required.The layer thicknesses and velocities are used to define the structureof the crust at a point offset laterally from the receiver by 10 km in the direction of shooting. In caseswhere the lateral offset causesstructures from adjacentreceiversto overlap (e.g., OBS ONE and OBS Esw ), the structuresare averagedand placed at a distance midway betweenreceivers. Mantle velocities are taken from the slope-intercept velocity-depth structures [Osler and Louden, 1992]. The mantle velocity nodesare offset laterally from the receiverby 10 km in the direction of shooting.For defining the lower boundary of the manfie layer, a velocity of 8.24 km/s is specifiedat a depthof 20 km. This introducesverticalvelocity gradientswhich range from 0.02 s-1 to 0.04 s-I dependingon the velocity at the upper boundaryof the mantle layer. For most OBSs, this gradient provides a suitable match for the critical range beyond which mantle refractedphasesbegin to be observed. The irregular distribution of boundary and velocity nodesfor the crustaland mantlelayersis interpolated to form a regulargrid with a nodal spacingof 10 km. Model CENTER IN LABRADOR SEA providing an acceptable model revision. This minimum standardis only relaxedwhenmultipleray pathsfrom different receivers overlap, but no structurecan be devised which is capable of allowing Z2 < 1 for allreceivers. Final Model In Figure5, the final velocity-depth structureis represented by smoothedisovelocitycontours,calculatedevery 1 km by a linearinterpolation betweenthe velocityandboundarynodes for the trapezoidin which they are bounded.Contourscan becomecoincident along model boundarieswhere there is a velocity contrast(e.g., the sediment-crust boundary).Figures 6 and 7 presentray paths throughthe layers of the final velocity-depth structure and their associatedobserved and computed traveltimes.Thevaluesof n, TRMs, andZ2 in the final velocity-depth structureare presentedin Table 1• (electronicsupplement). Sediment Layers. The sedimentlayers are numbered2 and 3 in the modelparameterization (Figure3). Ray pathsand associated travel times for sediment refractions are found in Figure 6a. Comparisonof computedand observedarrivalsfor OBSsL, M, and QNE exemplifythe excellentfit providedby themodel(Z2 valuesfrom0.057to 0.140).Maximumshot- Revisions receiveroffsetsfor computedsedimentrefractionscorrespond An optimal fit betweenobservedand computedtravel times closelyto observations, indicatingthat the vertical velocity is soughtby adjustingthe initial velocity and boundarynodes. gradients have been appropriatelyspecified. For sediments By assigninguncertaintiesto observedtravel time picks, the within the extinctrift valley (nodesfrom approximately80 to quality of a computed travel time fit may be assessed 120 km), the velocity of nodesalong the top and bottom of numerically (equation(5) below). For all travel time picks, a the lower sediment layer (layer 3) must be increased.The baseuncertaintyof :k 0.02 s has been chosento accountfor the requirementfor highervelocitiesis indicatedin Figure 6a by timing uncertaintiesinherent in the refraction data. A further uncertaintyhas been added to reflect the quality of the pick itself. The largestuncertaintiesof + 0.12 s have been assigned to Prop and sedimentrefractionarrivals,which are difficult to pick becauseof interference from first arrivals, and to Pn arrivals,which are difficult to pick becauseof their low signal to noise. To assessthe quality of successivemodel iterations,three parameterswere compared.First, one seeksa velocity-depth structure which maximizes the number of observed locations to which rays may be traced,n. The parametersTRMs and Z2 quantify the fit between the observedand computedarrival timesfor a givenray groupand aredeemedasfollows n(Tøbsi 1/2 TRMS i=•l -Tcalc i)2, = Z2 = /=1('•r•b•/ (4) OBS E andOBS ONE' whichhavevelocitiesexceeding 2.5 km/s, while receivers outside this region have velocities slower than 2.5 km/s. The topographyof the sediment-crust boundary(interfaces3 and 4 or 3 and 5 in Figure3) has been modified to improve the travel time fit of the crustal and mantle arrivals in regions where the basement reflector is difficult to pick. Two notable examples are (1) the shortwavelengthoscillationsintroducedon the basementhigh from 60 to 80 km to reproducethe travel time oscillationsobserved in Pn for OBS Nsw, OBS Esw, and OBS Msw and (2) the deepeningof the sediment-crustboundaryaround 150 km to delay crustalarrivals. Crustal Layers. The 2-D velocity-depth structure quantifiesthe lateral variationsin the crust acrossthe extinct spreadingcenter.The model layerswhich define the crustare numbered4, 5, 6, and7 in the modelparameterization (Figure 3). Ray pathsand travel times are presentedin Figure6b for the upper crust and in Figure 6c for the lower crust. Two principal variations in crustal structure emerge: (1) the variationin the velocity and thicknessof the crustacrossthe (5) •Anelectronicsupplement of TablesI and2 maybeobtained ona disketteor AnonymousFTP from KOSMOS.AGU.ORG. (LOGIN to AGU's FTP account using ANONYMOUS as the usemarne and GUEST as the password.Go to the right directoryby typing CD APEND. Type LS to seewhat files are available.Type GET andthe where Tobsis observedtravel time, Tcalcis calculatedtravel nameof the file to get it. Finally, type EXIT to leavethe system.) (Paper 94JB02890, Extinct spreadingcenter in the Labrador Sea: time, Uobsis uncertaintyof the observedtravel time, and n is Crustalstructurefrom a two-dimensionalseismicrefractionvelocity number of calculatedtravel times reachingobservedarrivals. model,by J. C. Osier and Keith E. Louden).Diskettemay be ordered Forvaluesof Z2 < 1, thecomputed arrivalsfit theobserved from American GeophysicalUnion, 2000 Florida Avenue, N.W., arrivals within the uncertainties of the observed travel times, Washington,DC 20009; $15.00. Paymentmust accompanyorder. OSLER AND LOUDEN: EXTINCT Southwest SPREADING CENTER IN LABRADOR SEA Distance(km) 0 40 80 2267 Northeast 120 160 200 I 01., 11 12 Line R1 O E N M , , L 10 Mantle 15 Figure 5. The 2-D velocity-depthstructureis represented by isovelocitycontours,drawnevery 0.25 km/s, which can becomecoincidentalong model boundarieswhere a velocity discontinuityis present.Layers in the model parameterization(Figure 4) are distinguishedby different grey tones. Receivers are denotedby solid circles. Arrows and correspondingchronological identifiers mark the intersection of magnetic anomalylineationswith line R2. Selectedvelocity contoursare labeled in kilometersper second. (b) Upper Crust SedimentLayers 12-- -•- E ---' •-M --• Q•.--• • 6 • •:i:• 12 %.. 11-- • 11• •o•N • • 4 ! • ........ o 3 •0 • m 6 0 40 sw 80 120 160 Distance(km) 200 NE '• 6 • 12 0 SW .......... HII IllHi [ ......... 40 80 120 160 Distance (km) 200 NE Figure6. Computed raypathsandtraveltimesusingthe2-D velocity-depth structure alonglineR2 for (a) refractions in thesediment layer(b) refractions in theuppercrust,P2 (c) refractions in thelowercrust,P3, and(d) mantlewide-angle reflections, PmP.Observed traveltimesandtheiruncertainties arerepresented by the stippledareasthroughwhichthe computed arrivaltimesare traced.Ray pathsconvergeto their corresponding receivers. For clarity,thediagrams showeveryotherOBSfor eachline so thatarrivaltimes can be discriminated. 2268 OSLER AND LOUDEN: EXTINCT SPREADING CENTER IN LABRADOR SEA (c) Lower Crust (d) PmP m 6 •, 6 12-" •, •••f • 5Q-•.•--E?...•-M-,'u•! 5 m12• .... 6 11• • 5 •O••• • 5 4 o go • 6 ............• 6 m 12 Z Z '" ' Z "''• m 12 0 40 SW 80 120 160 Distance(km) 200 NE 0 40 SW 80 120 160 Distance(km) 200 NE Figure 6. (continued) (b) Pnfor OBSsN andE (a) Pnfor OBSsO andQ 6 6 •--N .•- O --•. --,- ;%...".--,• ' .-.*,• o • 10..........:.. :,•• ............. lO • 20 6 20 , 6 ,,•- E .--,,. 0l•'ill/lllI 1/1111111111'11111 IIIII • 10 20 0 40 80 120 160 Distnnce (km) 200 0 40 80 120 160 Distance (km) 200 Figure 7. Computed raypaths andtraveltimesforrefractions in themantle, Pn,using the2-Dvelocitydepth structure along lineR2for(a)OBSs OandQ,(b)OBSs N andE,and(c)OBSs L andM.Observed travel timesandtheiruncertainties arerepresented by thestippled areas through whichthecomputed arrivaltimes aretraced. Raypaths converge totheircorresponding receivers. Forclarity, thediagrams showeveryother OBS for each line so that arrival times can be discriminated. OSLER AND LOUDEN: EXTINCT SPREADING CENTER IN LABRADOR SEA 2269 have many overlappingray paths in the lower crust and some in the upper crust.The four refractionprofilessamplingcrust (c) Pnfor OBSsL andM in thisregionhaveTRMs < 0.41s andZ2 < 0.49 (Table1, electronic supplement). On the flanks of the extinct spreadingcenter,0 to 60 km to the southwest and 130 to 220 km to the northeast, the seismic behavior o lO 20 6 I I I I I IIIIIIIIIIIIIIIIIIIIIII M Illill/Ill I1l lO 0 40 80 120 160 200 Distance (km) Figure7. (continued) extinct spreading center; and (2) the asymmetry between flanks of the extinct spreadingcenter. The transitionfrom flank to extinctspreadingcentercrustal typesis markedby a combinationof (1) thinningof the crust; (2) decrease in P wavevelocity;and(3) a morelinearvelocity gradient throughoutthe crust. Within the extinct spreading center,arrivalsat OBS ONE,OBS E, andOBS Nsw definethe upper and lower crustalstructure.Low upper crustalvelocities are readily apparentin Figure 6b by noting that the refractor velocities do not exceed 5 km/s. Further evidence is the markeddelayin lower crustal(P3) arrivalsat OBSsE, Nsw, and ONE (Figure6c) which travel throughthe regionof low upper crustal velocities within the extinct spreading center. For arrivals from the lower crust(Figure 6c), the observationsof low velocitiesat theseOBSs are evenmore pronounced.They are expressedin the final model by a downwarpingof the 6 km/s isovelocity contour as the extinct spreadingcenter is enteredand the proximity of the 7 km/s contourto the crustmantle boundary. The influenceof the extinct spreadingcenteris broad, with crustalthinning observedfrom 50 to 120 km (Figure 5). The transition to the extinct spreadingcenter crustal type, thin low-velocity crust with a uniform velocity gradient, occurs between 60 and 80 km on the southwest rift flank and 110 to 130 km on the northeastflank, approximatelycoincidentwith Chron 21. The crustal thicknesswithin the paleorift valley from 80 to 110 km varies between 3.8 and 4.2 km. The thinning is defined by (1) the short shot-receiveroffsetsover which computedand observedrefractedphasesare present;and (2) the Pn arrivals, particularly the critical range at which mantle arrivals begin to be observed.The 2-D velocity-depth structure within the extinct spreading center is well constrainedbecausereceiversin this region were situatedto of the crust differs from the crust within the extinct spreadingcenter, particularly to the northeast,with higher crustalvelocities;greaterrangesat which refractedarrivalsare observed;2 to 3 times more observed arrivals; and clearly discernibleProp arrivals.The southwestern flank is definedby arrivalsfrom OBS Q, sonobuoys 11 and 12, andOBS Osw. The range from 0 to 30 km is well definedby the ray pathsfrom OBS Q and sonobuoy 11. A typical oceanic structure is observedin this region, with a high-velocitygradientin the upper crust (isovelocitycontours4 to 6 km/s in Figure 5), a weaker velocity gradient in the lower crust (isovelocity contours 6 to 8 km/s in Figure 5), and a nominal crustal thicknessof 5.5 km as definedby Pn arrivalsfrom OBS Q, OBS Osw, OBS Nsw, andOBS Esw. For the northeastflank, the P3 (Figure6c) andProp (Figure 6d) arrivalsfrom OBS NNE, OBS M, and OBS Lsw define the featuresof this region. The arrivalsare of considerablyhigher velocity and extend to greater shot-receiveroffsets than the arrivals at OBSs samplingthe southwestflank or the extinct spreadingcenter.To model the lower crustrefractorvelocities, upper boundary velocities from 6.9 to 7.2 km/s and lower boundary'velocities from 7.5 to 7.7 krcdsare required in the fastest region from 130 to 170 km (Table 2, electronic supplement). In addition, a thickening of the crust (by approximately1 km) and a weak velocity gradientare required to match the shot-receiveroffsets to which P3 and PmP arrivals are observed. From 190 to 220 km, the crustal thickness (Figure 5) is comparableto that observedon the southwest flank from 0 to 30 km, but the velocities in the lower crustare higher. The Prop arrivalsobservedon thenortheastflank areusedto definethe velocitygradientstructureof the lower crustand,in conjunctionwith Pn arrivals,the depthof theMoho. They are second arrivals which are difficult to pick because of interference with the wavetrain from earlier arrivals. Accordingly, they are assigned the largest travel time uncertaintiesof all arrivals and are intentionallythe most poorlyfit (0.6 < Z2 < 3.8). The final velocity-depthstructure which accommodates the four setsof observedProp arrivals fromOBSM, OBSLsw, andOBSNNErepresents a compromise betweenthe modelrevisionswhich individualOBSs require. For example,in the regionof rapid crustalthinningfrom 140 to 120 km (Figure 6d), OBS Msw requiresa combinationof slope on the crust-mantleboundaryand high lower crustal velocitiesto reproduce the high (greaterthan8 km/s) apparent move-outvelocitiesof ProP. The fit of computed Prop arrivals for thisOBS (Z2=0.58)andfor OBS NNE(Z2=1.78)couldbe improvedby more rapid crustal.thinning. However,the Pn arrivalsfrom OBS NNE (Figure7b) arevery sensitiveto the slopeof this boundary,as it controlsthe extentof the shadow zone encounteredby rays enteringthe mantle. As the shadow zonebroadens,the numberof observedarrivalsto whichrays can be traced decreases. The asymmetrybetweenlower crustalstructuresobservedon southwest andnortheastflanksof the extinctspreadingcenter is consistent with previousseismicand gravityobservations. High-velocity lower crustal material has been observed at 2270 OSLER AND LOUDEN: EXTINCT SPREADING CENTER IN LABRADOR SEA widely spacedlocationsnortheastof the extinctspreading Mantle Layer. Ray paths and travel times for Pn (Figs. center(LinesA andF [Stergiopoulus,1984];ODP site 646 7a, 7b and 7c) are used to detimethe thicknessof the crust and [Srivastavaet al., 1989];andline 15/77[Hinzet al., 1979]), the velocity of the uppermostmantle. These two quantities thoughall but oneof theserefractionprofilesareunreversed exhibit significant variations along the length of line R2, and may be influencedby sedimentthicknessvariationsor primarily in relation to distance from the extinct spreading dippinginterfaces. The presenceof this materialmay be the center.In fitting the observedarrivalsfor Pn, TRMS was kept cause of the regional asymmetryin the free-air gravity within the uncertaintyof the travel time picks (i.e., Z2 < 1) for anomalynotedby Srivastavaet al. [1981]northof theMirma seven of the 10 sets of observedarrivals (layer 8 in Table 1, andCartwrightFractureZones.It is largelypositivenortheast electronicsupplement).Model revisionswere terminatedwhen of the extinctspreading centerandnegativeto the southwest reductionsin TRMSfor onereceivercausedTRMSto increasefor with the degreeof asymmetry increasing northwards.Vogtet another.For example,thelargestPn traveltimeresidual(TRMs al. [1982], in studying this regional asymmetryin the = 0.12 s), at OBS MNE (Figure7c), may be reducedby thinning Labrador and Norwegian-GreenlandSeas, find that the the crust or increasingthe mantle velocity between 170 and asymmetryis most pronouncedfor crust formed between 220 km. However, the arrivals for OBS L shootingsouthwest magneticanomalies 25 to 29 andweakerfor theyoungercrust. (Figure 7c) also travel throughthis region of the manfie but The formationof high-velocitylower crustmay be relatedto alreadyarrive too early (TaMs= 0.86 s). The remainingtravel hot spot activity from the time of anomaly24 until the time residuals may be attributed to (1) limitations of 2-D cessation of spreading [Srivastava et al., 1989], but methodology,with rays traveling outside the distance-depth knowledgeof its regionalextentwill be requiredto address plane assumedin the 2-D velocity-depthstructure;(2) variationsin the sumctureof an overlyinglayer which are not this speculation. (c) Method2: TRMS(ms) (a) Method 1' ReduceCrustalThinning ....................................... r.• I ,•, , ,/ _,•,r"• "•;// •,//l• o.o //,'7//./,/./' ,/1/if/// ,•/ /• • •. /I. /- , / ,. >-0.21 ....................................... 0.2 12 60 40 80 100 140 120 d.) Method 2:Z2 ..................................... ! I Distancealongline R2 (km) •0.0 (b)Method 1'Z2 -o.21....................................... I 0.2 e! ..................................... In Method 2:Observed arrivals reached, 0% 20% 40% 60% 80% Percentageof crustalthinningremow•d >-0.21........................ -2.0 -1.0 0.0 ,,. ............ I 1.0 Depth Perturbation(km) Figure8. Erroranalysis alonglineR2. Formethod1, (a) percentage is shownof crustal thinning removed withintheextinctspreading centerfrom40 to 140km and(b) thecorresponding influence is shownonthe qualityof fit between observed andcalculated arrivaltimesfor all mantlearrivals. Formethod 2, velocity anddepthperturbations (stars)areintroduced to nodeswhichformthecrust-mantle boundary in thefinal velocity-depth structure for line R2. The effectof theseperturbations on thequalityof themodelfit is assessed by plottingcontoursof (c) TRMs, (d) Z2, and(e) n. 2.0 OSLER AND LOUDEN: EXTINCT SPREADING CENTER IN LABRADOR SEA adequately resolved by the existing data; or (3) instrument timing errorswhich remain at variousreceivers. The isovelocitycontoursin Figure 5 show low velocities(< 8 kin/s) in the uppermost mantle at three locations. The thickest of the low-velocity regions underlies the extinct spreadingcenter and extendsto a maximum depth of 4.3 km below the crust-mantleboundary.The upper mantle velocities within the extinct spreading center are the most well constrainedin the model becauseof the numberof overlapping ray paths and the confidencein the crustal structureoverlying the mantle. The quality of the travel time fits is particularly 2271 perturbations (crustal thinning). For positive velocity perturbationsexceeding0.05 km/s, the number of observed arrivals which are reached by the rays begins decreasing markedly (Figure 8e). The increasesin the upper mantle velocities decreasethe velocity gradient of the mantle layer to the point where it is too weak to refract rays to the shorter shot-receiveroffsets.Using the dimensionsof the 80 ms TRMS contour and rejecting positive velocity perturbations,average uncertaintiesof q- 0.25 km and q- 0.08 km/s can be assignedto the depth and velocity nodes respectively which comprise the crust-mantle boundary. good(0.24< Z2< 0.58)for thereceivers whichhavePn ray paths crossing or entering the mantle within the extinct spreadingcenter(OBS E, OBS ONE,OBS Nsw, andOBS Msw). The depth of the low-velocity zone underlying the extinct spreadingcenter complementsthe results obtained by 1-D WKBJ syntheticseismogrammodeling for OBS G on line R1 along the axis of the extinct spreading center [Osler and Louden, 1992], which suggests that a relatively strong gradientexistsin the uppermostmantle and extendsto a depth of 3 km below the crust-mantleboundary. Mantle arrivals help define the transition from typical oceanic crustal structures on the flanks to the distinct crustal type within the extinct spreading center, as previously mentioned.On the southwestflank, the slope on the crustmantle boundaryis revealed by the Pn arrivals at OBS Osw (Figure7a) from 45 to 60 km, whichareup to 0.4 s earlierthan Synthetic Seismograms Synthetic seismogramswere generated for all receivers along line R2 using TRAMP, a geometrical ray theory algorithm for inhomogeneousstructures(C. Zelt, personal communication, 1992). The impulse responseseismograms were convolved with a source function, the exponentially dampedcosinefunction of Cervenyet al. [1977]: +v) (6) f(t)= exp - T t2cøs(2•fmt where fm is dominant frequency of source (6.8 Hz), T is dampingfactor (5.6), v is phaseshift (•c/3), and t is time. This thosefrom 10 to 45 km. OBS QNE(Figure7a) alsodefinesthe reproducesthe long wave train (approximately450 ms) which of the air gun sourcearray. crust-mantle boundary on the southwest flank. For this is characteristic A comparisonof synthetic seismogramscomputed for a receiver, the P n arrivals terminate abruptly at 55 km at a shadow zone generatedby the rapid crustal thinning. On the typical 1-D velocity-depth structure between TRAMP and northeastflank, the transitionis not as easily discerned.The WKBJ [Osler, 1993] is consistentwith previouscomparisons high velocity lower crust in this region counteractsthe delay of asymptotic ray theory and its extensions (the WKBJ which P n would otherwise exhibit as the crust thickens. approximation)[e.g., Chapman, 1985]. These differencesare However, there is evidence for slope on the crust-mantle most prevalent around the critical point for mantle arrivals, boundaryfrom the move-out of Prop and the shadowzone where asymptoticray theory breaks down. The result is that the peak amplitudeof PmP arrivalsis observedat 6 km greater generatedfor Pn arrivalsfrom OBS N shootingnortheast. rangein the WKBJ syntheticsandthe amplitudeof PmP builds up very rapidly in the TRAMP synthetics,while it is more Error Analysis gradual in the WKBJ synthetics.The amplitudesof refracted Error analysisof the modelparameters definingthe crust- arrivals comparefavorably except for Pn in an interval from mantle boundarywas performedby perturbingthe model just beyond the critical point to a range of 40 km, where the parameter valuesandthenray tracingthroughthe perturbed amplitudeof the TRAMP syntheticsare lower than the WKBJ model. In the first of two approaches, the degreeof crustal counterparts. thinningwithin the extinctspreading center,as definedby Refractionprofiles at two OBS have been selectedfor boundarynodesfrom 40 to 140 km along line R2, was inclusionin this paper. Observedprofiles for all OBS are progressively decreased (Figure 8a) until the crust-mantle available on microfichet (FiguresA1 to A10 and B1 to B4). boundary waslinearbetweenthe40 km and140km endpoints. The synthetic seismograms andobserved profilesfor OBSM In Figure8b,Z2 valuesusingthemantlearrivals at all OBSs locatednortheastof the extinctspreadingcenterare presented are presentedfor eachof theseperturbedmodels.Following in Figures9a and9b. Shootingsouthwest, the high amplitude the criteria of fitting computed arrivals to within the Prop arrivalsfrom 18 to 48 km in the observeddata are uncertainty assigned to theobserved traveltimepicks(Z2<1), matchedin the syntheticsfrom 18 to 39 km, at whichpoint no more than 25% of the crustal thinning may be removed theProp arrivalsterminateabruptly.The gapwith no second before the travel time fits are no longer acceptable. arrivals and weaker first arrivals from 15 to 19 km in the In the secondapproach,a constantvelocity and/or depth observeddata is also producedin the synthetics,but the perturbationwas appliedto all nodescomprisingthe crust- intervalis shiftedto 17 to 21 km. The observedPn arrivalsfor mantleboundary. Contours of TRM s,Z2,andn arepresented in this receiverare particularlystrongfrom 80 to 90 km as rays Figures8c, 8d, and 8e, respectively,as a functionof the enter the southwestflank of the extinctspreadingcenter.The velocityand depthperturbation magnitude.The TRMs = 80 ms andZ2 = 1 contoursshowa preferredorientationto plausible 1Appendix figuresareavailablewith entirearticleon microfiche. perturbations. Increasesin Pn velocity are accompanied by Orderfrom AmericanGeophysicalUnion, 2000 FloridaAvenue,N.W., positive depth perturbations (crustal thickening), and Washington,DC 20009. Document B95-001; $2.50. Paymentmust decreases in Pn velocityare accompanied by negativedepth accompany order. OSLER AND LOUDEN: EXTINCT SPREADING CENTER IN LABRADOR SEA 2273 syntheticsdo not replicatethis behavior.Shootingnortheast, the syntheticis a particularlygoodrepresentation of the data. The low amplitudessurrounding the basementhigh (shotpoint 700) are presentin both the syntheticsand observeddata from 19 to 24 km. On either side of this feature, the synthetics match the high-amplitudearrivals from 10 to 19 km and 25 to supportedby a number of studies.Huang and Solomon [1988] present observations that the maximum centtold depth of active ridge crest earthquakesincreasesfrom 2-3 km at full spreadingrates of 4045 mm/yr to 5-6 km at full spreading rates of 5-10 mm/yr. Purdy et al. [1992] show that a systematicrelationshipexists between spreadingrate and the 45 kin. depth to which a P wave velocity inversion is observed at The syntheticseismograms andobserved profilesfor OBS E active spreadingcenters. Their preferred explanation is that locatedwithin the extinct spreadingcenterare presentedin the "major faults that extend to greater depths on slowerFigures10a and 10b. The amplitudeof the upper and lower spreadingridges [Huang and Solomon,1988] and allow a more crust refracted arrivals is strong and well modeled in the vigorous circulation of cooling sea water to reach greater synthetics,as the energy is refractedby the nearly linear depths,blocking the upward migration of large melt bodies" velocity gradient throughout the crust within the extinct [Purdy et al., 1992]. The maximum depth to which faulting spreadingcenter.The high amplitudesin the observedcrustal extends is an indication of the thicknessof the mechanically arrivalsare followedby strongPn arrivalsfrom 20 to 30 km. stronglayer which is prone to brittle failure in responseto the The synthetics put very little energyinto thesePn arrivalsjust extensionalstressescausedby the divergingplates. Through of the throw on normal faults at slow beyondthe criticalpoint, despitea relativelystrongvelocity an examination spreadingcenters,Macdonald and Luyendyk [1977] suggest gradient(0.04 s-l) for theuppermantlein thisregion. The syntheticand observedrefractionprofiles compare that up to 20% of the plate divergencemay be accommodated favorablyas the major amplitudevariationsfor P2, P3, and by these faults. Last, there are compilations of seismic Prop arrivalsare matched,althoughthe patternsof amplitude refraction data which document that crustal thickness becomes variationsare shiftedby severalkilometersin someinstances. a function of spreadingrate when the spreadingrate is less The greatest shortcomingsof the syntheticsare the low than 15 mm/yr [Reid and Jackson, 1981; Bown and White, amplitudeof thePn arrivalsandthevery rapidterminationof 1994]. A systematicrelationshipbetweenspreadingrate, faulting, the PmP arrivals.However,basedon the comparison of the TRAMP and WKBJ algorithms,theseshortcomings stemin hydrothermalcirculation,and the depthbelow the spreading part from limitationsof the asymptotic ray theorywhichthe centerat which an appreciablepartial melt fractionis observed TRAMP algorithmemploys,andnot solelyfrominadequacies couldbe readilyappliedto the spreadingcenterin the Labrador Sea precedingits extinction.The zone in which crustal and in the f'malvelocity-depthstructure. Discussion Nature and Origin of the Crust The seismicrefractionresultsalongthe strikeof the extinct spreadingcenter [Osler and Louden, 1992] and crossingit transversely (this paper) find a prevalence of anomalously thin and low velocity crust in the extinct spreadingcenter, which is underlain by low velocities in the uppermostmantle. There are two frameworksby which theseobservations may be explained. These consider the crust as (1) having formed through igneous extrusive and plutonic processes, then subsequentlyaltered by tectonism;or (2) being composedof hydrothermally altered upper mantle. Oceanic crest, in the prevalentview, is igneousrock whichsolidifiesfrom partial melt after it migratesto the accretionary axis.The detailsof this process remain equivocal, but analogies between ophiolitestudies[e.g., Salisburyand Christensen,1978] and marine seismic studies [e.g., Spudich and Orcutt, 1980] suggestthat the structureof matureoceaniccrestis primarily vertical, with (1) the upper crust (layer 2) composedof extmsive volcanics (pillow basalts), underlain by a dense seriesof vertical dikes (sheeteddiabasedikes) stemmingfrom upper mantle velocities are anomalouslylow and the crustal thinning occursis boundedwithin anomalies21 west and 21 east(Figures1 and 5) wherethe slowestspreadingratesare encountered. It is within this region that the crest shouldbe most susceptibleto brittle failure, thinn2ngby tectonism,and a reducedpartialmelt supplydueto conductive coolingof melt bodies. This hypothesis is supportedby a multichannel seismic reflection profile crossing the extinct spreading center to the south of line R2. Srivastava et al. [1993] find evidence for crustal extension by rotated crustal blocks between anomalies 21 west and 21 east. A second view of the anomalous crustal structure within the extinct spreadingcenter is one in which it is formed by the serpentinizationof upper mantle peridotitc [Hess, 1962; Lewis and Snydsman,1979]. The Moho in this casewould not representa compositionalchangebetween crest and mantle but the depth to which upper mantle was hydrothermally altered. The crustal P wave velocities observed would be a linear function of the volume percentageof serpentinite, velocities of approximately 5 km/s for 100% serpentinite increasingto 8.0 km/s for unalteredperidotitc [Christensen, 1966].Serpentinite hasa Ve/Vsratiowhichis muchhigher than other rocks likely to constitutethe oceaniccrest. It is on the basisof Vt•/Vsstudies[Christensen,1972] that the episodesof injectionfrom an underlyingmagmachamber;and serpentinitemodel of normal oceanic crest has been largely (2) the lower crest (layer 3) consistsof plutonicallyeraplaced rejected. However, it remains a viable alternative for the gabbros. Assuming that the crust at the extinct spreading center formedfollowingthis generalmodel,a pervasivefracturingof the entire crest is a mechanism for thinning the crest and lowering its velocity. In this scenario,decreasesin spreading rate precedingextinctionwould be associated with a decrease in magmatismand an increasingrequirementfor tectonismto accommodatethe divergenceof the plates.This associationis structure of the crust at oceanic fracture zones [Calvert and Potts, 1985; Louden et al., 1986] and has been applied as an alternativemodel for the East Pacific Rise basedon gravity data and thermal models [Lewis, 1983]. In the serpentinitemodel, the variable thicknessof the crust along the strike of the extinct spreading center, the linear velocity gradients and the low crustal velocities (line R1, 1-D travel time solutionsand z-p inversions[Osler and 2274 OSLER AND LOUDEN: EXTINCT SPREADING CENTER IN LABRADOR SEA (a) Line R2: ObservedandSyntheticRefractionProfilesfor OBS E ShootingSW Shot Point 100 200 300 I I-',.... ,._-:._,.;,:,, ?,,..-..,..:• ,,,;..,:,:..:.., ;;....,•.... ...... ,. ,, ,:" ,'.,, ,,. , ",.... . ,.,.,"". • ,,.,' '.''.,,.,"c..,.r';.;•.',,.;.,... ',;',',:,;.,'-'.,, ;...... ,a..... ,,.., 'f:.,., ....... F:..,.,.,,,;:,,•,.,.,..,.,•,.z,:';,,,,Z. ,.,"r, .,,;.,,,..,;•., ::....... ,:,,-,,,.-1 ...... -,.,',•'.", .,•,. ,,,, ,,,' ,,, ,-,., ........ ,,', ,,, .,,.,.' '{-•-:,;.,:-d"Z::;"-•:-":'•Z :";•" '-':":'•';".:'":; '. :"'T..,-'..,•: ,, ,: ','.,,":.'.,,,-,.,, ' -.,,, ,-: '• • ,' 'i) ) w,)' ',,)•.' ,,, ',. "' . '), k)n) )• ',"" ,,,...... • ), ,,*' ,","•,, • ß ','..,.,.,,':.,..,,," ,,. ")'•"• "..,,--'-,.,'"•.,':.,,..,.,') { ;, 'M ;,,d,,)'.."'•?,,•_ ,.",'' ,•, , ))'•) ' ,,,,.. M,.' "7•' ,'.•,, ,',,•),',,,,", ,.,,...., 'I,'.'..' ,j ,,, , ,, ,)•k• ,,,,, ,,,,. ,..,') ,'• ,.,,,', ,),•) ") ,. '"--- -' • •"ik.'• -. i,,.,-'.,, "" "2 . :' '" .,_ , ,-. "' •-"')"'• • ,, ., ' ,, ,,. ' :.,::,, II '...,,_.. ['""-;";• •/ •,1,, ....• ,'.,._ •'.,_ •,,•,,, ' ,-;..,.',,•,,. ,_.,:,,' ,,..,:,.,;;. -'..r, ,, ," , ' , , ,, , ,,,o I me) ,,,),,, ), ),•)l) ,,) ),, M,)t) i) , ) w , ,w , • {] •/•j,• IL )'/% Ih I1/ • If--','"• % • I "."'.."-•-' ;'•., - ,-.:...' ,:"'",.*""'.".< ....... '.::."-',,.,' ,:;', "',,.. ";',, .............. J ;•, -'":-'•, "..' ..',,,•• ".,' ...:;. •..' ,• !;',. ;, e':.',--' ", ,- ", ......... -L.... '...,,',i,•.,,,,"' •., ._., ,:_'./•'..',,•,,,,, ,,,,.' ,,,•,.,'" ,,,,; ,,, '.-'"' ,,';,, ,', ,' ',, ,L"'" '.,....••", ".,'. '-"•' ;',.•"''' .-':, ' -•.,.' '_':.,',':3 ,..,.'..... .•.,....',, .... '• '-i" ',..,•,., '.._.t_•_ ß ',., ',"'• '-."' •.,'.:.,-_•L,'.. ',T,-,;";,:'.,;,,',5.,:, ".'..'..',. •-• -,so , , I ' ..... :.... ,, :' ,. , , ,, '";, , , "" Di•..,•(kin) -so 7 3 I I -90 I 450 Distance0an) -30 0 (b) Line R2: ObservedandSyntheticRefractionProfilesfor OBS E ShootingNE 500 ,..::. ..... . .... . 600 ,,,•,-;......•,...,;, .,,,J,........ ß ,... , ,,,,,.,,...,.,. ,, ', ,, 6 ,•...), ....... ',;......... M)) ' ,' .............. • ), , )' ., , . , , , , . ' .. , , •, 0 )') )) ) ) , , , ,. I'l)n•,)I )))'*' ,. ) )), ) ))" ;..., '.;•., -. '.. .,.,•.. ,.:',.,,. , ,; ",.' .:'•" ") ',:"'" M))1'):-i ! i1....) ) .i.14,1, i• ,))) ) • ,...... ' ." '' :.r, . , '. ,',, .....:';..'.; :;'....: ........ i."!'.-:... ....... ".....; .... '"::"•'"::' :.:.:.,.,,,, ....... : ,,•....., r'•.., ,, .;:,.,;. ""• .... . .....:,,,,,.,:,:'.:,, .:,..:"•:• ))) ß )L ' ) ) ) .),)) , .,,, . ,.,. 20 ,,,,•. :.. ,,....... ,..w ',: ,,,', ,. ) ..... ß ..' . ) ) ,•).1,) )1. )) ) ) ) )M)w) ) )14. • .')) ) ,)• ' , ) )) ,• )) ,. Shot Point ß .::; Distance 0ma) , ) P ) ) )')))'•) ')))) ..... . ::...'; 40 60 6 I 0 20 I Distance (kin) I I 40 Figure10. Observed andsynthetic refractionprofilesfor (a) OBSEsw and(b) OBSENE. Observed tracesare band-pass filteredfrom 4 to 12 Hz; threetracesaremixed.Gain is appliedas a functionof rangeandtraces are plotted at a reducingvelocityof 8 km/s. Arrivalsfrom differentphasesare identifiedon the synthetic profiles. 60 2275 OSLER AND LOUDEN: EXTINCT SPREADINGCENTER IN LABRADOR SEA Chron (a) 21 20 13 20 21 ; . ; . ; ; : : ; ; Thin. rang iRegion. orCru!tal (b) • i o o I ; ; ; : : :; : Southwest i.; : ; o : ; : : Northeast 0 i•::•::::i•i:i•i•i•:::::•ii?:•::•:::• ................ •i!iiii::iii!!::•::!::iiiii,i::??:::!:' '!::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: ....... 'i":;:;:; ....................... ii::•::!::i::•??• ....... •.................. iil!!iiiiiiii::.. ':i':i,.'::::.. "i:i:i:i;i ..:ii!!i :•(• :' 10 iii,,i:ii!ii :.:.:.:.::::::::::: •::iii'i::i:: •!!!!!!!• - i 15 Mantle I I 40 ! i '*•-...... • 'F,% PervaSive Faulting oftheCrust I ! I ! I "• Serpebtinized Upper mantle 80 I I 120 I I 160 I 200 Distance (km) Figure11. Summary of results alonglineR2 transversely crossing theextinctspreading centerin the Labrador Sea.(a) Crustal thickness at eachmantleboundary nodein the2-D seismic modeling and(b) the suggested relationship between thecrustal thinning, low crustal P wavevelocities andhydrothermal alteration of theuppermantledueto faultingandreduced partialmeltsupplyat theslowspreading rates precedingthe extinctionof the spreadingcenter. Louden, 1992]) may be explainedin termsof the regional Nature and Origin of the Upper Mantle variabilityin the depthto whichhydrothermal circulation has penetrated andalteredtheuppermanfie. However, thepresence Low velocitiesare observedin the uppermantle in four of of velocitygradients in the uppermantlebeneaththe Moho the sevenrefractionprofilesshoton line R1 alongthe strike requiresa furthercomplexity. Two stagesof hydrothermalof the extinctspreadingcenter[Oder andLouden,1992].The alterationwouldneedto havebeenoperative:onethatis quite averageuppermantlevelocityis 7.74 km/s, whichis lessthan vigorousandshortlived to establish the distinctboundary the velocities of 7.92, 7.82, and 7.94 km/s determined for the interpreted astheMohoanda second thatis moreprotracted to threeuppermantlenodeswithin the extinctspreadingcenter alter rocks below this boundary.While this mechanismis in this study(Table 2, electronicsupplement). This difference largelyspeculative, multiplestages of hydrothermal alteration in velocity may be attributedto azimuthalanisotropyin the arerequired to explain thepetrology of gabbroic rocksdredged mantle [MacKenzie, 1972; Bibee and Shor, 1976]. The low velocities are interpreted as resulting from the from the failed MathematicianRidge in the easternPacific Ocean[Stakesand Vanko,1986]. An initial high-temperature hydrothermal alteration of upper mantle peridotire. The vaporphaseis purported to be followed by a morepervasive pathway for hydrothermalalteration of the upper mantle (Figure 11) couldbe providedby faultsextendingthroughthe lower-temperature brinephase. A crustallayerwhichis discretefrom the mantleis more easilyexplained by theigneous modelin whichthevelocity of the crustis loweredby pervasive fracturingandthinnedby increasedtectonismand decreasedpartial melt supply.The faults providea pathwayfor fluids to penetrateinto the uppermost mantleand alter the peridotite(Figure11). A preexisting Moho is thusoverprinted by one episodeof hydrothermalalteration.Discriminationbetweenthese crust. No more than 10% of the peridotire must be serpentinized to producethe observedvelocities[Christensen, 1966]. There is evidence for hydrothermal alteration and serpentinization at oceanicfracturezones[e.g., Calvert and Potts, 1985; Christensen, 1972; Minshull et al., 1991]. Hydrothermalalterationis called upon by Calvert and Potts [1985] to explain the widespreadoccurrenceof low upper mantle velocities in old Atlantic fracture zones and an alternatives wouldrequireVp/Vsratios for thecrustal level apparentdecreasein upper mantle velocity in fracturezones with increasingage. Given the similarity betweenthe crustal structures observed within the extinct spreading center to arenot well developed andonlyshowarrivalsfor the mantle those at oceanic fracture zones [Oder and Louden, 1992] and refraction, Sn. material. S-waves are observedon some OBSs; however, they 2276 OSLER AND LOUDEN: EXTINCT SPREADING the likelihood of faults penetratingthe entire crust at the slow spreadingrates preceding extinction, the extinct spreading center is an environmentin which uppermantle peridotitcare liable to be serpentinized. Comparison to Crustal Spreading Centers Processes at Active Slow At the Mid-Atlantic Ridge, recent studies suggest that within a ridge segment there is considerable along-axis variation in the thermal and mechanical properties of the lithosphere.The studiesinclude along axis seismicrefraction [Purdy and Derrick, 1986], high-resolution multibeam bathymetry mapping [Sempere et al., 1990], multichannel seismic reflection [Derrick et al., 1990], 3-D gravity studies [Kuo and Forsyth, 1988; Linet al., 1990], 'microearthquake source locations and delay time tomography[Tooracyet al., 1988; Kong et al. 1992], and geochemicalstudies[White et CENTER IN LABRADOR SEA extinctspreadingcenterin the LabradorSea.Informationfrom 12 kHz bathymetrydata, seismicreflection data, and plane layerand'r-p 1-D traveltime solutions wereusedto developan initial velocity-depth structure. The final velocity-depth structurewas obtainedby revising this initial velocity-depth structureto improve(1) the fit betweenarrivaltimescalculated by ray tracingand their observedcounterparts; and (2) the fit between amplitude patterns in synthetic and observed refraction profiles. Crustwithin the extinctspreadingcenteris foundto be thin andof anomalously low P wavevelocitywhencomparedwith mature oceanic crust. It is underlain by low upper mantle velocities which may exhibit some anisotropy. The low crustalvelocities are attributedto an increasein the degreeof tectonismprecedingextinction, which pervasivelyfractured the crust(Figure 11). This faultingalsoprovidesa pathwayfor hydrothermalalterationof the uppermantleperidotitcand is al., 1992]. the manner by which the low velocities in the uppermost An important consensusemerging from these studies at mantle are produced.The lateral extent of the thinning and slow spreadingcentersis that crustalaccretionis an episodic occurrenceof low-velocity crust are coincidentwith the time and not a steady state processand crustal accretionmust be interval in which spreadingrate in the LabradorSea decreased focused at the center of the ridge segment [Ballard and most rapidly (Chron 21 until the cessationof spreading).This Francheteau, 1983; Solomonand Toorney,1992; White et al., observation supports a systematic relationship between 1984]. As there is no seismic evidence supporting the spreadingrate, faulting, hydrothermalcirculation,and partial presence of steady state axial magma chambers at slow melt supply at active spreadingcenters.A subsequentpaper spreadingcenters [Derrick et al., 1990], magm•ifismat the will extend the seismic results presentedin this paper by crustal level is consideredto be an intermittent process. consideringgravity models which are basedon the proposed Consequently, plate divergenceat slow spreadingcentersis seismicvelocity-depthstructureand plausiblespreadingrate partially accommodatedthrough lithospheric stretching historiesfor the Labrador Sea basedon modelingof magnetic [Macdonaldand Luyendyk,1977]. Along-axis,ridge segment anomalies. ends appear to experiencingmore tectonismthan the axial high wherecrustalaccretionis focused[Huangand Solomon, Acknowledgments. We would like to thank the officers and crew of 1988; Lin and Bergman, 1990]. The ends of the ridge segmentsare cooler than any other portion of the ridge segmentbecauseof their proximity to the cold lithosphere againstwhich they are juxtaposed.Consequently, they are mechanicallystrongerare more prone to brittle failure than otherportionsof the ridge crest.The endsof ridge segments are also areas where the production of partial melt supply and/or its transportis most severelylimited [Phipps Morgan and Forsyth, 1988]. the C.S.S.Hudson,BordenChapman,Guy Fehn,andNeil Hamilton,for their technicalexpertiseduringthe courseof the seismicrefraction experiment.The analogrefractiondata werereplayedand digitized usingcomputing facilities attheBedford Institute of Oceanography with the assistance of StevePearyand Ian Reid. The development of processing algorithms for therefraction datawasassisted in itsinfancy by GregLegerandthestaffof theDalhousie University Computing and Information Servicesandmorerecentlyby DepingChianandSaeid Yazdanmehr.The researchwas supported by Energy,Mines and Resourcesgrant 30/04/87 and Natural Scienceand Engineering ResearchCouncilof CanadagrantsOGP0008459,EQP0027418,and INF0001469.J.C.O.acknowledges financialsupport providedby a fellowshipfrom DalhousieUniversity.Comments on the original manuscript by J. Austin,D. Hutchinson, andananonymous reviewer are The processeswhich occur during the extinction of a spreadingcenter have an analog in the spatial relationship betweenthe center of a ridge axis and its fracturezones.The parallels are (1) the presenceof brittle crust due to thermal regime, hot and cold lithospherejuxtaposedat a fracturezone greatlyappreciated. and a dissipatingthermalanomalyat a dying spreadingcenter; References (2) the effectsof tectonismwhich thin the britfiecrustdirectly through crustal extension accommodatedby rotated crustal Anderson,R.N., andJ.G. Sclater,Topography andevolutionof the East PacificRisebetween5øSand20øS,EarthPlanet.Sci.Lett.,14, 433blocks or indirectly by reducingpartial melt supply;and (3) 441, 1972. the provision of a pathway for hydrothermalalterationof the Geologicprocesses of the mid-ocean upper mantle. At the extinct spreadingcenter,the decreasesin Ballard,R. D., andJ. Francheteau, ridge and their relation to sulfide deposition,in Hydrothermal spreadingrate which precede the extinction of a spreading Processes at SeafloorSpreadingCenters,editedby P. A. Ronaet al., center enhancecooling of the crust by conduction.This leads to an increase in tectonism as the crust becomes britfie. The increased tectonism promotes deeper penetration of conductive cooling which restricts partial melt supply. Serpentinizafionof the upper manfie ensuesonce the crusthas been faulted and deephydrothermalcirculationoccurs. Conclusion A 2-D seismic velocity-depth structure (Figure •) was developed for refraction line R2 transverselycrossingthe pp. 141-168,Plenum,New York, 1983. Batiza,R., Failed rifts, in The Geologyof North America,vol. N, The EasternPacific Oceanand Hawaii, editedby E. L. Winrefer,D. M. Hussong,and R. Decker, pp. 177-186, Geological Society of America,Boulder,Colo., 1989. Bibee,I• D., andG. G. J. Shot,Compressional waveanisotropy in the crustanduppermanfie,Geophys. Res.Lett.,3, 639-642,1976. Bown,J. W., andR. S. White, Variationwith spreading rateof oceanic crustalthickness andgeochemistry, EarthPlanet.Sci.Lett., 121,435449, 1994. Calvert, A. J., and C. G. Potts, Seismicevidencefor hydrothermally OSLER AND LOUDEN: EXTINCT SPREADING CENTER IN LABRADOR SEA alteredmantle beneaththe old crustof the Tydemanfracturezone, Earth Planet. Sci.Lett., 75, 439-449, 1985. 2277 seismotectonicsof the Vema fracture zone, Atlantic Ocean, J. Geol. Soc.London,143,795-805, 1986. Cerveny,V., I. Molotkov,andI. Psencik,Ray Methodin Seisinology, Macdonald, K. C., and B. P. Luyendyk, Deep-tow studiesof the Universityof Karlova,Prague,Czechoslovakia, 1977. structure of the Mid-AtlanticRidgecrestnearlat 37øN,Geol. Soc. Am. Bull., 88, 621-636, 1977. Chapman, C. H., 1985,Raytheoryanditsextensions: WKBJandMaslov seismograms, J. Geophys., 58, 27-43,1985. MacKenzie, D. B., Peridotitefabricsand velocity anisotropyin the Christensen, N. I., Elasticityof ultrabasicrocks,J. Geophys.Res., 71, Earth's mantle, Mem. Geol. Soc. Am., 132, 593-603, 1972. 5921-5931, 1966. Mammerickx,J., andD. Sandwell,Riftingof old oceaniclithosphere, J. Geophys.Res.,91, 1975-1988, 1986. Christensen, N. I., The abundance of serpentinites in the oceaniccrust, J. Geol., 80, 709-719, 1972. Denham,C. R., andH. Schouten,On the likelihoodof mixedpolarityin oceanicbasementdrill cores,in Implicationsof Deep Drilling Results in the Atlantic Ocean: Ocean Crust, Maurice Ewing Ser., vol. 2, editedby M. Talwaniet al., pp. 160-165,AGU, Washington,D.C., 1979. Derrick, R. S., J. C. Mutter, P. Buhl, and I. Kim, No evidencefrom multichannel reflection data for a crustal magma chamberin the MARK areaon the Mid-Atlantic Ridge,Nature, 347, 61-64, 1990. Minshull, T. A., R. S. White, J. C. Mutter, P. Buhl, R. S. Detrick, C. A. Williams,andE. Morris,Crustalstructure at theBlakeSpurFracture Zone from expandingspreadprofiles,J. Geophys.Res.,96, 99559984, 1991. Nagumo,S., T. Ouchi,andS. Koresawa,Seismicvelocitystructurenear the extinct spreadingcenterin the ShikokuBasin,North Philippine Sea, Mar. Geol.,35, 135-146, 1980. Osier, J. C., Crustal structureof the extinct spreadingcenterin the LabradorSea:Implicationsfor dynamicmodelsof flow beneathmid- ocean ridges, Ph.D. thesis, Dalhousie Univ., Halifax, Nova Scotia, Drake, C. L., N.J. Campbell,G. Sander,and J. E. Nafe, A mid1993. LabradorSea ridge,Nature, 200, 1085-1086,1963. Hess,H. H., Historyof oceanbasins,in PetrologicStudies:Buddington Osler, J. C., and K. E. Louden, Crustal structureof an extinct rift axis in the Labrador Sea: Preliminary results from a seismic refraction Volume,pp.599-620,Geological Societyof America,Boulder,Colo., 1962. Hinz, K., H. U. Schlfiter,A. C. Grant,S. P. Srivastava,D. Umpleby,and J. Woodside, Geophysical transects of theLabradorSea:Labrador to southwest Greenland,Tectonophysics, 59, 151-183, 1979. Huang,P. Y., and S.C. Solomon,Centroiddepthsof mid-oceanridge earthquakes: Dependence on spreading rate,J. Geophys.Res.,93, 13,445-13,477, 1988. Hyndman,R. D., Evolutionof the LabradorSea,Can.J. Earth Sci.,10, 637-644, 1973. Jarrard,R. D., K. A. Dadey, andW. H. Busch,Velocity and densityof sediments in the Eirik Ridge,LabradorSea:Controlby porosityand mineralogy, Proc. OceanDrill. Program,Sci.Results,105, 811-835, 1989. Kent, D.V., and F. M. Gradstein,A Jurassicto recentchronology,in The Geology of North America,vol. M, The WesternNorth Atlantic Region, edited by P.R. Vogt and B. E. Tucholke,pp. 45-50, GeologicalSocietyof America,Boulder,Colo., 1986. Klitgord, K. D., and J. Mammerickx, Northern East Pacific Rise: Magnetic anomalyand bathymetricframework,J. Geophys.Res., 87, 6725-6750, 1982. survey,Earth Planet. Sci. Lett., 108, 243-258, 1992. Phipps Morgan, J., and D. Forsyth, Three-dimensionalflow and temperatureperturbationsdue to a transform offset: Effects on oceaniccrustaland upper mantle structure,J. Geophys.Res.,93, 2955-2966, 1988. Purdy, G. M., and R. S. Detrick, Crustal structureof the Mid-Atlantic Ridgeat 23øNfromseismic refraction studies, J. Geophys. Res.,91, 3739-3762, 1986. Purdy, G. M., L. S. L. Kong, G. L. Christeson,and S.C. Solomon, Relationshipbetween spreadingrate and the seismicstructureof mid-oceanridges,Nature,355, 815-817, 1992. Reid,I., andH. R. Jackson, Oceanspreading rateandcrustalthickness, Mar. Geophys.Res.,5, 165-172, 1981. Roest,W. R., andS. P. Srivastava, Sea-floorspreading in the Labrador Sea:A new reconstruction, Geology,17, 1000-1003, 1989a. Roest,W. R., and S. P. Srivastava,Seafloorspreadinghistory,I, Magneticanomaliesalongtrack, in East CoastBasinAtlas Series: LabradorSea,coordinated by J.S.Bell, p. 98, GeologicalSurveyof Canada, Atlantic Geoscience Centre, Dartmouth, Nova Scotia, 1989b. Kong, L. S. L., S.C. Solomon,and G. M. Purdy, Microearthquake Roots,W. D., and S. P. Srivastava,Origin of marinemagneticquiet characteristics of a mid-oceanridge along-axishigh, J. Geophys. zonesin the Labradorand GreenlandSeas,Mar. Geophys.Res.,6, Res.,97, 1659-1686, 1992. 395-408, 1984. Kristoffersen,Y., and M. Talwani, Extinct triple junction southof Greenlandand the Tertiary motion of Greenlandrelative to North Salisbury,M. H., andN. I. Christensen, The seismicvelocitystructure of a traverse through the Bay of Islands ophiolite complex, America, Geol. Soc.Am. Bull., 88, 1037-1049, 1977. Newfoundland,an exposureof oceaniccrustand uppermantle,J. Geophys.Res.,83, 805-817, 1978. Kuo, B., and D. Forsyth, Gravity anomaliesof the ridge-transform systemin the SouthAtlanticbetween31 and34.5øS:Upwelling Sempere,J., G. M. Purdy, and H. Schouten,Segmentationof the MidAtlanticRidgebetween24øN and30ø40'N,Nature,344, 427-431, centersandvariationsin crustalthickness, Mar. Geophys.Res.,10, 205-232, 1988. Lewis, B. T. R., The East Pacific Rise and the thermal model, J. Geophys.Res.,88, 3348-3354, 1983. Lewis, S. D., and D. E. Hayes, The structureand evolution of the 1990. Solomon,S.C., and D. R. Toomey,The structureof mid-oceanridges, Annu.Rev. Earth Planet. Sci., 20,329-364, 1992. Spudich,P., andJ. Orcutt,A new look at the seismicvelocitystructure of the oceaniccrust,Rev. Geophys.,18, 627-645, 1980. CentralBasinFault, West PhilippineBasin,in The Tectonicand GeologicEvolutionof Southeast AsianSeasand Islands,Geophys. Srivastava,S. P., Evolutionof the LabradorSea andits beatingon the earlyevolutionof the North Atlantic,Geophys.J. R. Astron.Soc.,52, Monogr.Ser.,vol. 23, pp.77-88,AGU, Washington, D.C., 1980. 313-357, 1978. Lewis, B. T. R., andW. E. Snydsman,Fine structureof the lower crust Srivastava,S. P., R. K. H. Falconer, and B. MacLean, Labrador Sea, on the CocosPlate,Tectonophysics, 55, 87-105, 1979. Davis Strait,Baffin Bay: geologyand geophysics-A review,Mem. Lin, J., and E. A. Bergman,Rift grabens,seismicity,and volcanic Can. Soc.Pet. Geol., 7, 333-398, 1981. segmentationof the Mid-Atlantic Ridge: Kane to Ariantis fracture zones(abstract),Eos Trans. AGU, 71, 1572, 1990. Lin, J., G. M. Purdy,H. Schouten, J. C. Sempere,and C. Zervas, Evidencefrom gravitydatafor focusedmagmaticaccretionalong theMid-AtlanticRidge,Nature,344, 627-632, 1990. Loncarevic,B. D., Oceanbottomseismometry, in CRC Handbookof Geophyscial Exploration at Sea,editedby R.A. Geyer,pp.219-263, CRC Press,BocaRaton, Fla., 1983. Louden,K. E., R. S. White,C. G. Potts,andD. W. Forsyth,Structure and Srivastava, S. P., C. G. Powell, A. G. Nunns, L. C. Kovacs, D. G. Roberts,M. T. Jones,C. Uruski,andD. Voppel,Magneticanomalies of total intensity,in GeophysicalAtlas of theNorth AtlanticBetween 50øto 70•Bland0 øto 65øW,editedby S. P. Srivastava, D. Voppel, and B. Tucholke,pp. 3-4, DeutschesHydrographisches Institut, Hamburg, 1988. Srivastava, S. P., K. E. Louden, S. Chough, D. Mosher, B. D. Loncarevic, P. Mudie, A. de Vernal, and B. MacLean, Results of 2278 OSLER AND LOUDEN: EXTINCT SPREADING CENTER IN LABRADOR detailedgeologicaland geophysical measurements at ODP sites645 in Baffin Bay and 646 and 647 in the LabradorSea,Proc. Ocean Drill. Program,Sci.Results,105, 891-922, 1989. Srivastava,S. P., C.E. Keen, and P. Potter, Crustal extensionin the Labrador Sea: Evidence from multichannel seismic reflection data (abstrac0,Eos Tram. AGU, 74, Fall Me•ting suppl.,605, 1993. Stakes, D., and D. A. Vanko, Multistage hydrothermalalterationof gabbroicrocksfrom the failed MathematicianRidge, Earth Planet. Sci. Lett., 79, 75-92, 1986. Stergiopoulus, A. B., Geophysical crustalstudiesoff the SW Greenland margin,M. Sc. thesis,DalhousieUniv., Halifax, Nova Scotia,1984. SEA Weissel,J. K., and A. B. Watts,Tectonicevolutionof the Coral Sea Basin,J. Geephys. Res.,84, 4572-4582,1979. White,R. S., R. S. Detrick,M. C. Sinha,andM. H. Corntier,Anomalous seismiccrustalstructureof oceanicfracturezones,Geephys.J. R. Astron.Sec., 79, 779-798, 1984. White, R. S., D. P. McKenzie, and R. K. O'Nions,Oceancrustal thickness from seismic measurements and rare earth element inversions, J. Geephys. Res.,97, 19,683-19,716,1992. Woodside, J.,Geophysics II, Gravityanomaly, in EastCoastBasinAtlas Series:Labrador Sea, coordinated by J.S. Bell, p. 94, Geological Surveyof Canada,AtlanticGeescience Centre,Dartmouth, Nova Scotia, 1989. Talwani, M., and O. Eldholm, Evolution of the Norwegian-Greenland Zelt, C. A., andR. B. Smith,Seismictraveltime inversion for 2-D crustal Sea, Geol. Sec. Am. Bull, 88, 969-999, 1977. velocitystructure, Geephys. J. Int., 108, 16-34,1992. Tomoda, Y., K. Kobayashi,J. Segawa,M. Nomura, and K. KJmura, Linear magnetic anomalies in the Shikoku Basin, northeastern PhilippineSea,J. Geomagn.Geoelectr.,28,47-56, 1975. K. E. Louden,Departmentof Oceanography, DalhousieUniversity, Toomey, D. R., S.C. Solomon,and G. M. Purdy,Microearthquakes Halifax, Nova Scotia,CanadaB3H 4J1. (e-mail: keith•papa.ocean. beneaththe median valley of the Mid-Atlantic Ridge near 23 øN: dal.ca). Tomographyandtectonics,J. Geephys.Res.,93, 9093-9112,1988. J. C. Osier, Defence ResearchEstablishmentAtlantic, Dartmouth, Van der Linden,W. J., Crustalattenuation andseafloorspreading in the Nova Scotia,CanadaB2Y 3Z7. (e-mail: osler•maggie.drea.dnd.ca) Labrador Sea, Earth Planet. Sci. Lett., 27, 409-423, 1975. Vogt, P. R., L. C. Kovacs,C. Bernere,and S. P. Srivastava,Asymmetric geophysicalsignaturesin the Greenland-Norwegianand southern LabradorSeas and the Eurasiabasin, Tectonophysics,89, 95-160, 1982. Weissel, J. K., and D. E. Hays, Evolution of the Tasman Sea reappraised, Earth Planet.Sci.Lett.,36, 77-84, 1977. (Received February 21, 1994; revised October 31, 1994; accepted November 2, 1994.)