Difficulties in Seismically Imaging the Icelandic HotspotFigure Captions____________________________________________________________Keller et al. - IMAGING ICELAND HOTSPOT - 03/08/00 - page 1 of 12Difficulties in Seismically Imaging the Icelandic HotspotWilliam R. Keller, Don L. Anderson, Robert W. ClaytonDepartment of Geological and Planetary Sciences, California Institute of TechnologySubmitted to Science 02/28/00Correspond with: William R. KellerSeismological LaboratoryCaltech, MC: 252-21Pasadena, CA 91125Ph: (626) 395-6932Fax: (626) 564-0715e-mail:[email protected]____________________________________________________________Keller et al. - IMAGING ICELAND HOTSPOT - 03/08/00 - page 2 of 12AbstractThe locations of volcanic islands may be controlled by thin or extending parts of thelithosphere over a partially molten asthenosphere (1, 2), by edge effects near the boundaries ofthick cratonic lithosphere (3), or by narrow jets of hot mantle rising from deep within the mantle(4-6). Many hotspots are found on or near ridges, at lithospheric discontinuities, or inextensional environments, so high resolution seismic images are required to determine whether itis lithospheric structure, stresses in the lithosphere, or the deep mantle that is the controllingfactor for the location of active volcanoes. In this study, we perform a simple experiment inwhich we use basic geometrical arguments and idealized experimental parameters in order tounderstand the resolution of tomographic images of the upper 400 km of the mantle underIceland. Our results indicate that a narrow, deep seated mantle plume is not required in order toexplain the observed travel time delays. Results of tomographic inversions are often viewed asunique; however, recent seismic studies of the Icelandic Hotspot have illustrated the non-uniquenature of these models.The geometry of plumes in laboratory and computer simulations is a narrow cylindercapped by a bulbous head that flattens beneath the lithosphere, giving an overall mushroomshape to the upwelling (7-9). Deep mantle upwellings are also expected to broaden beneath the650 km endothermic phase change. On the other hand, the geometry of upwellings driven by____________________________________________________________Keller et al. - IMAGING ICELAND HOTSPOT - 03/08/00 - page 3 of 12plate divergence or by lateral changes in lithospheric thickness are expected to be focused at thesurface toward the thin or extending regions. Iceland is in a particularly complex region,different from other volcanic islands, because it is located on a very slowly spreading ridge in theyoungest, narrowest part of the Atlantic Ocean and is bounded by thick cratonic lithosphere. Theseparation of thick cold cratonic lithosphere will generate a deep upwelling which focuses towardthe surface to fill in the newly formed gap. Passive steady-state upwellings, such as those foundat mature ridges away from thick cratonic lithosphere, will exhibit a similar geometry but will nothave as deep of an expression. Ribe et al. (10) showed that a hot, narrow, rising plumeunderneath Iceland would produce a bathymmetric signature that is inconsistent withobservations (11) and suggested that the anomaly must be cooler and wider than would beexpected from a hot rising plume. Using seismic methods, it is theoretically possible todistinguish between a narrow plume upwelling, passive effects due to plate divergence, anddynamic upwelling between two cratons; however, distinguishing between these three scenariosis problematic with real data.Using data from a regional broadband seismic experiment (ICEMELT), Wolfe et al. (12)produced three dimensional tomographic images of the mantle beneath Iceland which show aΑcone shaped≅ low velocity zone beneath the island that is approximately 150 km wide at thesurface and is inferred to extend to at least 400 km depth. They suggest that this low velocityzone is the expression of a plume that is rising from deep within Earth=s mantle. However, thisΑcone shaped≅ geometry is not consistent with published images of plumes that suggest theexistence of a cylindrical plume conduit which feeds a broadening plume head in the uppermost____________________________________________________________Keller et al. - IMAGING ICELAND HOTSPOT - 03/08/00 - page 4 of 12mantle (7-9). The Αcone shaped≅ tomographic appears to be defined by the cone of incomingrays, and most of the rays are traveling nearly vertically in the upper 400 km beneath Iceland. Because of the lack of crossing rays, the structure described might be explained by the smearingout of a shallow (<200 km depth) low velocity anomaly instead of the effect of a deep mantleplume. This is the well known parallax problem and is not unlike the problems encounteredwhen a light is shone on an object and one attempts to reconstruct the shape of the object fromthe shadow it forms on the wall. For instance, a disc, a sphere, an ellipsoid, a cone, and acylinder will all cast a circular shadow on the wall when oriented in the proper way. The onlyway to determine the three dimensional shape of the object is to observe the shadow when thelight source is shone on the object at many different angles. We show that the uniqueness andresolution problem encountered when imaging the Icelandic mantle is due to the geometry of theexperiment and the lack of crossing ray information. Other tomographic studies in areas nearhotspots have found that it is impossible to distinguish between a shallow anomaly in the upper200 km of the mantle and a narrow deep seated plume. In a recent study of the YellowstoneHotspot, Saltzer and Humphreys (13) found that both scenarios fit their tomographic inversionresults equally well.We perform simple tomographic resolution tests in which we calculate idealized syntheticdelay times for S-waves through a variety of velocity anomalies in the upper 400 km of themantle, and then invert these delays for structure in order to understand how well theseanomalies may be resolved in a tomographic inversion. Raypaths from sources and receivers____________________________________________________________Keller et al. - IMAGING ICELAND HOTSPOT - 03/08/00 - page 5 of 12along the two dimensional profile discussed in Wolfe et al. (12) were calculated to determine theray coverage, number of ray crossings, and angle of ray crossings along this cross section of themodel (figure 2a). The sources and receivers used in this