Mantle Shear-Wave Velocity Structure Beneath the Hawaiian Hot Spot

Mantle Shear-Wave Velocity Structure Beneath theHawaiian Hot SpotCecily J. Wolfe, et al.Science 326, 1388 (2009);DOI: 10.1126/science.1180165The following resources related to this article are available online atwww.sciencemag.org (this information is current as of December 4, 2009 ):Updated information and services, including high-resolution figures, can be found in the onlineversion of this article at:http://www.sciencemag.org/cgi/content/full/326/5958/1388Supporting Online Material can be found at:http://www.sciencemag.org/cgi/content/full/326/5958/1388/DC1A list of selected additional articles on the Science Web sites related to this article can befound at:http://www.sciencemag.org/cgi/content/full/326/5958/1388#related-contentThis article cites 27 articles, 3 of which can be accessed for free:http://www.sciencemag.org/cgi/content/full/326/5958/1388#otherarticlesThis article appears in the following subject collections:Geochemistry, Geophysicshttp://www.sciencemag.org/cgi/collection/geochem_physInformation about obtaining reprints of this article or about obtaining permission to reproducethis article in whole or in part can be found at:http://www.sciencemag.org/about/permissions.dtlDownloaded from www.sciencemag.org on December 4, 2009Science (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. Copyright2009 by the American Association for the Advancement of Science; all rights reserved. The title Science is aregistered trademark of AAAS.

REPORTSMantle Shear-Wave Velocity StructureBeneath the Hawaiian Hot SpotCecily J. Wolfe, 1 * Sean C. Solomon, 2 Gabi Laske, 3 John A. Collins, 4 Robert S. Detrick, 4John A. Orcutt, 3 David Bercovici, 5 Erik H. Hauri 2Defining the mantle structure that lies beneath hot spots is important for revealing their depthof origin. Three-dimensional images of shear-wave velocity beneath the Hawaiian Islands, obtainedfrom a network of sea-floor and land seismometers, show an upper-mantle low-velocity anomalythat is elongated in the direction of the island chain and surrounded by a parabola-shapedhigh-velocity anomaly. Low velocities continue downward to the mantle transition zone between410 and 660 kilometers depth, a result that is in agreement with prior observations of transitionzonethinning. The inclusion of SKS observations extends the resolution downward to a depth of1500 kilometers and reveals a several-hundred-kilometer-wide region of low velocities beneathand southeast of Hawaii. These images suggest that the Hawaiian hot spot is the result of anupwelling high-temperature plume from the lower mantle.Hawaii is the archetypal hot spot and hasbeen suggested to be the surface expressionof a mantle plume: a localized upwellingof hot buoyant material from Earth’s deepmantle (1, 2), although such an origin has beendebated. The Hawaiian-Emperor chain of islandsand seamounts spans thousands of kilometers;records age-progressive volcanism for >75 millionyears (3); and exhibits a broad ~1000-kmwideregion of elevated topography, known asthe Hawaiian Swell (Fig. 1A), that surrounds thelocus of current volcanism. Estimated to transporta larger buoyancy flux than any other active plume(4, 5), the Hawaiian hot spot has been the focusof numerous geochemical and geodynamical studies[for example, (4–10)] that attempted to resolveits origin. One of the most straightforwardindications of whether a hot spot is the result of aplume is the presence of a narrow, vertically continuouszone of low seismic velocities in the underlyingmantle, indicative of higher-than-normaltemperatures, anomalous mantle composition, ormelt (11, 12).Global seismic tomography suggests that lowseismic velocities may extend from the upper tothe lower mantle beneath the Hawaiian hot spot(13–15), but there remains doubt about how well,and to what depth, narrow plumes may be resolvedby such methods (16, 17). Moreover, global modelsof mantle structure are limited in their resolutionnear Hawaii by the sparseness of wave paths,given the lack of seismic stations on the oceanfloor and the great distances between the islandsand most circum-Pacific earthquake sources. Previousinvestigations of mantle structure near Hawaiiusing island stations (18–20) or sea-floor deploymentsof instruments (21, 22) have been limitedin geographic coverage and numbers of instruments.Such experiments have not yielded the highresolutionregional tomographic models neededto settle the debate over even such a basic questionas whether the upper mantle beneath theHawaiian Islands is marked by low seismic velocitiesand higher-than-normal temperatures, assuggested by some studies (13–15, 19), or not, asimplied by others [for example, (23)].Here we report results from the HawaiianPlume-Lithosphere Undersea Melt Experiment(PLUME), which was designed to determine athigh resolution the mantle seismic velocity structurebeneath the Hawaiian hot spot, to assess thehypothesis that the hot spot is the product of anupwelling plume, and to determine how mantleflow may interact with the overlying lithosphereA25˚20˚−5000 −2500 0 2500mto generate the Hawaiian Swell. The experimentfeatured a dense, large-aperture seismic networkconsisting of two year-long deployments of threedozen broadband ocean-bottom seismometers(OBSs) and the concurrent deployment of 10portable broadband seismometers on the HawaiianIslands (Fig. 1A), with all instruments operatingcontinuously to record teleseismic and localearthquakes. During the experiment period, 2146S-wave relative arrival times (direct S and SKSphases) were collected from 97 earthquakes (24)(fig. S1) and corrected for estimated variations instation elevation and crustal thickness. Correctedmean station delay patterns reflect upper-mantleheterogeneity and indicate relatively low velocitiesbeneath and to the west of the HawaiianIslands and high velocities east of Hawaii and atdistant stations around the margins of the swell(Fig. 1B).The arrival times were inverted for S-wave velocityheterogeneity, damped earthquake relocations,and origin time terms using finite-frequency(13, 25) methods. Because of the large stationspacing, crossing wave paths were lacking, andvertical resolution was limited in the uppermostmantle. One strategy to mitigate the effect of smearingstrong, shallow heterogeneity deeper into amodel is to include station terms to partially absorbthe effect of shallow structure, but this approachcomes at the expense of decreased resolutionat shallow depths, and tests indicate that stationterms may not successfully absorb structure atdepths of 100 to 200 km if such variations areextremely strong. We therefore performed inversionsboth with (Fig. 2 and fig. S6) and without(Fig. 2 and figs. S7 and S8) station terms. In bothcases, a low-velocity anomaly was well resolvedin the upper mantle beneath Hawaii and waselongated in a southeast-northwest direction parallelto the Hawaiian Islands. As expected, theB25˚20˚Downloaded from www.sciencemag.org on December 4, 20091 Hawaii Institute of Geophysics and Planetology, University ofHawaii at Manoa, Honolulu, HI 96822, USA.2 Department ofTerrestrial Magnetism, Carnegie Institution of Washington,Washington, DC 20015, USA.3 Cecil H. and Ida M. GreenInstitute of Geophysics and Planetary Physics, Scripps Institutionof Oceanography, San Diego, CA 92093, USA.4 WoodsHole Oceanographic Institution, Woods Hole, MA 02543, USA.5 Department of Geology and Geophysics, Yale University, NewHaven, CT 06520, USA.*To whom correspondence should be addressed. E-mail:cecily@soest.hawaii.edu15˚−165˚ −160˚ −155˚ −150˚15˚−2 s+2 s−165˚ −160˚ −155˚ −150˚Fig. 1. (A) Locations of seismometers used in S-wave tomography. Land stations are indicated by bluetriangles and OBSs are indicated by red (first deployment; 2005 to 2006) or yellow (second deployment;2006 to 2007) circles. Only stations that successfully recorded two horizontal components are shown.Bathymetry is taken from (30). (B) Corrected mean station delays, adjusted to vertical incidence. Earlyarrivals are shown by blue circles and late arrivals by red triangles, with symbol size scaled linearly to themagnitude of the delay (see scale at lower left).13884 DECEMBER 2009 VOL 326 SCIENCE www.sciencemag.org

REPORTSamplitude of upper-mantle heterogeneity was largerin inversions without station terms. A parabolashapedregion of higher-than-average seismic velocityalong the edge of the Hawaiian Swell wasalso observed to extend downward to 300 kmdepth, although patterns outside its boundarieswere constrained only to the southeast, where stationcoverage extends beyond its edge. Low velocitiesbeneath Hawaii continue into the transitionzone, and inversions resolved a broad region ofdepth = 100 kmstation terms (s )-2-1 1 2-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5S-wave % velocity anomaly164 162 160 158 156 154 152 150 148 146Longitude Westdepth = 300 km164 162 160 158 156 154 152 150 148 146Longitude Westdepth = 400 km262420 22Latitude1816-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5S-wave % velocity anomaly164 162 160 158 156 154 152 150 148 146Longitude West262420 22Latitude1816-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5S-wave % velocity anomaly262420 22Latitude1816low velocities in the lower mantle from 700 to1500 km depth southeast of Hawaii (Fig. 3).The vertical extent of upper mantle structureis not well constrained by our S-wave data alone,as illustrated by a two-step inversion (fig. S9). Inthe first step, we inverted for a model in which allstructure was “squeezed” between 50 and 250 kmdepth. In the second step, we corrected the observationsfor the effects of the squeezed modeland inverted for an additional residual model.depth = 100 km-4.0 -2.7 -1.3 0.0 1.3 2.7 4.0S-wave % velocity anomaly164 162 160 158 156 154 152 150 148 146Longitude Westdepth = 300 km164 162 160 158 156 154 152 150 148 146Longitude Westdepth = 400 km262420 22Latitude1816-2.0 -1.3 -0.7 0.0 0.7 1.3 2.0S-wave % velocity anomaly262420 22Latitude1816-2.0 -1.3 -0.7 0.0 0.7 1.3 2.0S-wave % velocity anomaly164 162 160 158 156 154 152 150 148 146Longitude WestFig. 2. Upper-mantle velocity heterogeneity at (top)100,(middle)300,and(bottom)400kmdepth.The scale of heterogeneity is indicated in the upper right corner of each panel. The left column displayssolutions from an inversion with station terms; the right column shows the results of an inversion withoutstation terms.262420 22Latitude1816The two-step model continued to recover a lowvelocityanomaly in the transition zone, and themodel deeper than 500 km remained virtually unchanged,but much of the heterogeneity at 300 kmdepth can be squeezed to shallower depths. Giventhe tradeoffs between the amplitude and the depthinterval of heterogeneity, however, lateral variationsat 100 to 200 km depth in this two-stepmodel were extremely large at T8%. Although reconciliationwith surface-wave observations isneeded to provide firm limits on the lateral variationin upper-mantle seismic velocities and helpresolve these ambiguities, prior surface-wave observations(21) and analysis of PLUME data todate indicate mantle S-wave velocity heterogeneityof no more than T4% in the depth intervalfrom 60 to 140 km and are thus inconsistent withthe large variations recovered in the two-stepmodel.SKS waves have steeper arrival angles thandirect S waves do and thus provide wave pathsthat sample the lower mantle beneath Hawaii (fig.S2). To test how the subset of SKS data contributesto the resolution of lower-mantle structure,we performed inversions in which SKS phaseswere excluded (figs. S11 and S12). Direct S wavesfrom the first deployment dominantly constrainedthe low-velocity anomaly in the upper mantle beneathand around the main Hawaiian Islands, butthese inversions had less structure in the mantletransition zone (410 to 660 km depth) and negligiblelower mantle structure (fig. S11). Includingdirect S-wave data from the second OBS deploymentextended the spatial region of the model andimproved resolution of the low-velocity anomalyin the upper mantle and transition zone (fig. S12),but lower-mantle heterogeneity remained small.These tests indicate that low velocities in the lowermantle beneath Hawaii are not artifacts of verticalsmearing of strong upper-mantle heterogeneity,but instead are driven by the resolving power ofSKS arrival times. The existence and southeasternposition of the lower-mantle anomaly constrainedby SKS data were affected, however, by the solutionfor shallower upper-mantle structure (24).Heterogeneity deeper than 1500 km is not requiredby our data alone, but the eventual incorporationof PLUME data into global models may extendthe resolution to greater depths. Montelli et al.(13, 14) imaged a broad region of low velocitiesbeneath Hawaii that extended downward to 1900 kmdepth in their global S-wave model and to thecore/mantle boundary in their P-wave model. Ourlow-velocity anomaly is narrower than the ~10°wideanomaly displayed by Montelli et al.(13, 14),which is probably an indication of the improvedhorizontal resolution provided by the PLUMEdata. Resolution tests conducted with a wide varietyof cylindrical and checkerboard structures(figs. S16 to S28) demonstrate that the PLUMEdata set resolves both upper- and lower-mantlestructure at the level of our interpretations. Onepossible biasing effect on our solutions in thelower mantle, however, may come from undeterminedstructure near the core/mantle boundaryDownloaded from www.sciencemag.org on December 4, 2009www.sciencemag.org SCIENCE VOL 326 4 DECEMBER 2009 1389

REPORTS-1.0 -0.7 -0.3 0.0 0.3 0.7 1.0-1.0 -0.7 -0.3 0.0 0.3 0.7 1.0A B Cdepth = 600 kmS-wave % velocity anomalydepth = 900 kmS-wave % velocity anomalydepth = 1200 km-1.0 -0.7 -0.3 0.0 0.3 0.7 1.0S-wave % velocity anomaly12121214141416161618 20 22 24Latitude18 20 22 24Latitude18 20 22 24Latitude262626282828303030168 166 164 162 160 158 156 154 152 150 148 146 144 142Longitude West(24), where steep lateral velocity gradients havebeen observed at the edges of the Pacific largelow-shear-velocity province (26).Our results improve the fidelity of S-wavemantle imaging around Hawaii as compared withprevious efforts and suggest that the upper mantlebeneath the islands is characterized by lowseismic velocities and, by inference, anomalouslyhigh temperatures. The peak-to-peak 3% velocityvariations that we imaged at 300 km depthin inversions with station terms corresponds to atemperature variation across the model of 250°C,and the 1.5% contrast near 900 km depth correspondsto a temperature variation of 300°C(24). These values are consistent with prior petrologicand geodynamic inferences of a hightemperatureplume beneath the Hawaiian hotspot (4–6, 10, 27).In terms of plume dynamics, the broad, lowvelocityanomaly at shallow mantle depths (