Skip to content


    Skip to content

    RESEARCH: GEOSCIENCES | Contents | Next

    What is Under the East Pacific Rise?

    FEATURED
    Donald W. Forsyth
    Chairman, Department of Geological Sciences, Brown University
    Shu-Huei Hung
    Postdoctoral Researcher, Princeton University

    W hen scientists accepted in the 1960s the ideas of plate tectonics--that the Earth's surface was composed of several slowly moving pieces, or plates--they also recognized that new crust forms at mid-ocean ridges, places where the plates spread apart on the sea floor. Understanding how energy and material are transferred from the Earth's interior to the surface at mid-ocean ridges is thus central to plate tectonics and other Earth sciences, yet it is difficult to observe processes that occur under the ocean floor. Now, observations and computational experiments are helping scientists determine whether crust formation at the ridges is passive, dynamic, or both. In other words, do the spreading plates permit a broad zone of melted rock to form and be drawn upward, or do plumes of magma rise to the spreading centers?

    For example, seismological observations--signals from earthquakes all over the world--have demonstrated that basaltic melt is present in a large region beneath the ridge known as the East Pacific Rise (EPR), at the junction of the Pacific and Nazca plates about 4,000 kilometers off the coast of South America.

    "As the oceanic plates separate at this spreading center, partial melting of the upwelling mantle creates enough magma to form a layer of basaltic crust six to seven kilometers thick," said Donald Forsyth, chairman of the Department of Geological Sciences at Brown University. "At the EPR axis, the crust appears to form within one or two kilometers of the sea floor. A major question has been how melt is transported from the distributed region of melt production to this narrow zone at the axis."

    Forsyth is a principal investigator for the NSF-sponsored Mantle ELectromagnetic and Tomography (MELT) Experiment, involving 14 investigators from seven institutions. "Our experimental work has begun to give us answers to some of our questions, and our computational work, done at SDSC, is now helping us to refine those answers," Forsyth said.

    PASSIVE SPREADING
    COMPUTATIONAL CONSTRAINTS

    forsyth1

    Figure 1: Ocean Bottom Seismometer Arrays

    Deployment sites of ocean bottom seismometers in the MELT Experiment area along the East Pacific Rise. (Figures 1 and 2 reprinted with permission from Science, 280 (22 May 1998): 1215-1216. Copyright 1998 American Association for the Advancement of Science.)

    PASSIVE SPREADING

    In passive flow models, viscous drag from the separating plates induces a broad zone of upwelling and melt production. "The melt must then migrate to the ridge axis in a horizontal direction, despite the tendency of its own buoyancy to drive it vertically," Forsyth said. In dynamic flow models, by contrast, low viscosity in the melting regime and buoyancy from mantle depletion and melt retention force a focused upwelling that may be quite narrow, only a few kilometers. Melt transport is primarily vertical, with most of the melting beneath the ridge axis. Some models postulate several centers of upwelling, with material redistributed along the ridge axis into a uniform layer.

    The passive models, according to Forsyth, simply tap a well-stirred, nearly isothermal asthenosphere. Proponents of more dynamic models are suggesting that upwelling beneath ridges is a more active part of the whole mantle convection system. The primary goal of the MELT Experiment is to distinguish between the two scenarios by searching for the narrow upwellings predicted by the dynamic models.

    The experimenters aboard the Scripps Institution of Oceanography ship R/V Melville deployed two arrays of ocean-bottom seismometers at the EPR in November 1995 (Figure 1). The instruments recorded seismic waves from earthquakes around the world for six months (Figure 2), then were retrieved from the sea floor. Because seismic waves travel more slowly in basaltic melt than in solid mantle, delays in the arrival of waves are signs that melt is present in the wave path. The investigators' analyses of the seismic data appear in the May 22, 1998, Science.

    "Overall, a region of low velocities several hundred kilometers across and perhaps as deep as 150 kilometers is clear in the data, corresponding to the passive scenario," Forsyth said. "The only evidence for a narrow zone of partial melt was in delays at the station located directly on the ridge axis, and this anomaly was probably caused by magma within the crust, extending only a few kilometers below the sea floor. The observations support the hypothesis that ridges passively tap the asthenosphere and do not represent buoyant, upwelling limbs of whole mantle convection cells."

    forsyth2Figure 2: Sources of Signal in MELT Experiment

    A map centered on the experimental area shows the locations of several large earthquake events measured at the East Pacific Rise during the six-month experimental period.

    9808143-2-outlines

    Figure 3: Structure Under an Ocean Ridge System

    Ocean ridges like the East Pacific Rise are the source of the oceanic crust (lithosphere). Subsurface structure can be detected by examining velocity changes and angular displacements in arriving seismic waves.

    COMPUTATIONAL CONSTRAINTS

    But might a narrow upwelling zone be hidden beneath the resolution limits of the data? "The shortest-period waves from distant seismic events recorded in the experiment corresponded to a minimum wavelength of about 25 kilometers, which might be greater than the expected width of a dynamic upwelling zone," Forsyth said. "Such a zone might be so narrow that we could not detect it."

    A computational exploration of this possibility became a dissertation topic for one of Forsyth's graduate students, Shu-Huei Hung, now a postdoctoral researcher at Princeton University. "We used the IBM SP at SDSC to carry out numerical simulations of finite-frequency seismic waves propagating through postulated heterogeneities that corresponded to the narrow upwellings of the dynamic models," she said.

    Her multi-domain pseudospectral model for 3-D anisotropic seismic waveform modeling is designed to run on a high-performance parallel computer. "Time constraints on the IBM SP at Brown made it advantageous for us to use the SP at SDSC, which has higher-performance nodes," Hung said. The model solves acoustic and elastic wave equations formulated by velocity and stress for fluid layers and solid spaces, respectively. Wavefields and the structure are described by Fourier series in the horizontal directions and Chebyshev polynomials in the vertical direction.

    "Seismic wave propagation in structures that are heterogeneous and contain anisotropies is complicated," Hung said, "but our method accounts for and detects some fairly subtle effects." These include diffraction, shear wave splitting, wavefront healing (which can mask the signals of heterogeneities), and anisotropies caused by the preferred orientations of the crystalline structure of different kinds of rock. "This has only become possible in the last few years with the advent of high-performance parallel machines," Hung said.

    Hung's model represents a hypothetical upwelling zone through which a seismic disturbance propagates (Figure 3). The model output demonstrates the capacity of the program, given data like that acquired in the MELT Experiment, to detect a very narrow (four kilometers across) low-velocity structure running from 10 to 60 kilometers beneath the sea floor (Figure 4). "The synthetic calculations show that diffraction and wavefront healing do not hide the travel-time delay signature of a narrow, vertical, low-velocity channel," Hung said. "Any anomaly this large or larger should have been detected in the EPR data, so our modeling puts some tough constraints on the dynamic hypothesis."

    "While it is possible that the threshold detection limit of the MELT Experiment is just large enough to allow the existence of a narrow upwelling zone beneath the ridge axis," Forsyth said, "such a zone, while marginally possible, seems physically improbable to us on the basis of the modeling work." --MM

    undersea-snap1-cmykFigure 4: Detecting a Small Low-Velocity Zone

    Shu-Huei Hung and Donald Forsyth modeled the East Pacific Rise experiment on the IBM SP at SDSC. The model low-velocity zone (white rectangle) is less than four kilometers wide, running from about 10 to 60 kilometers deep. As it is crossed by a synthetic seismic disturbance, the model registers the complex energy interference of multiple reflected and converted waves from the water and crust layers. A low-velocity zone narrower than that shown would probably not be detectable.