The Department of Geology and Geophysics (G&G) conducts research into a wide variety of topics aimed at furthering our understanding of the dynamic processes of the Earth/Ocean/Atmosphere system. Our research spans across land and oceans as we seek to understand connections between the continents and oceans, ice-sheet dynamics and the formation and evolution of the Earth as a whole. We study the structure and evolution of the oceanic crust from its formation at mid-ocean ridges to consumption at subduction zones, coupled with the dynamics of the mantle that drives seafloor spreading. We study a wide range of fluid-mediated processes, including those occurring at hydrothermal vents, at shelf-edge seeps and in subduction zone settings. Included in these processes are links to seismicity, fluxes of chemicals to the ocean and mantle, microbial activity and the subseafloor biosphere. We study the role of oceans both in relation to past climate change and as a driver of present day climate dynamics, and use natural archives like from sediments, corals, and tree rings to understand past climate. We study a wide range of coastal processes including the impacts of climate change and storms on coastal regions.
The Department today consists of about 30 Ph.D. level Scientific Staff and another 16 Technical Staff (many of whom hold Ph.D. degrees). In addition there are about 25 graduate students pursuing their Ph.D. through the WHOI/MIT Joint Program and roughly 8 Postdoctoral Scholars, Fellows and Investigators.
The Scientific and Technical staff carry out research that involves sea-going deployments of instruments built in house; laboratory studies using high precision analytical facilities; and theoretical and computational studies of ocean and climate processes and geodynamics. Examples of the facilities within the department include the National Ocean Sciences Accelerator Mass Spectrometry Facility (NOSAMS) and the Northeast National Ion Microprobe Facility (NENIMF). We now run the national Ocean Bottom Seismograph Instrument Center (OBSIC).
Geology & Geophysics Department
Spring 1998 — Long before the plate-tectonic revolution began in the 1960s, scientists envisioned drilling into the ocean crust to investigate Earth’s evolution. As early as 1881, Darwin suggested drilling Pacific atolls for scientific purposes. From the end of the nineteenth into the first half of the twentieth century, drilling was used to penetrate the reef and uppermost volcanic foundation of several oceanic islands, and these glimpses of oceanic geology whetted the scientific community’s appetite for deeper and more complete data.
The idea of ocean drilling gained momentum in the 1950s and resulted in the “Mohole project,” whose objective was to sample the material beneath the Mohorovicic discontinuity, or “Moho,” the boundary between Earth’s crust and mantle. The Mohole project proved to be ahead of its time, but the program’s test drilling in the deep eastern Pacific Ocean proved that ocean crust could be successfully drilled and cored from a dynamically positioned ship in several thousand meters of water. Mohole was succeeded beginning in the late 1960s by the hugely successful Deep Sea Drilling Project, the International Phase of Ocean Drilling, and the current Ocean Drilling Program. (For “An Abridged History of Deep Ocean Drilling” by Arthur E. Maxwell, see Oceanus Vol. 36, No. 4.)
A major scientific goal of all these efforts has been to recover a complete section of normal ocean crust and uppermost mantle. The lithology, properties, and geological relations of these rocks are key to understanding such varied phenomena as convection in the earth’s mantle, melting and transport mechanisms in the upwelling asthenosphere beneath mid-ocean ridges, rock deformation and alteration, the sources of magnetic anomalies in ocean basins, and hydrothermal circulation and formation of ore deposits.
Poking at the Oceanic Lithosphere
The outstanding obstacle to drilling through the entire ocean crust is that normal crustal thickness is on the order of 6 to 7 kilometers. Penetrating it beneath several thousand meters of water is presently at the limits of drilling technology as well as beyond the financial boundaries of scientific funding. Currently, drilling to depths of 1.5 to 2 kilometers has yielded upper ocean crust samples, as well as sections of deeper crust and upper mantle where these rocks are exposed by natural tectonic processes. Pieced together, these sections are in many ways similar to our conception of what ocean crust should look like, based on studies of ophiolites, old seafloor parcels that have been exposed above sea level by various tectonic mechanisms. (See map in Peter Kelemen article.)
At its base, the “standard ophiolite model” (see top figure) includes an “ultramafic” layer (made up of dense rock rich in iron and magnesium) that represents the oceanic mantle. The overlying crust is formed from melt that rose buoyantly at a mid-ocean ridge from the hot, upwelling mantle. From bottom to top, the crust consists of layers formed by crystallization in presumed magma chambers or zones of hot crystal mush, a shallower section of vertically oriented, intrusive dikes,* and a volcanic surface layer of pillow basalts extruded onto the seafloor and capped by sediments. Laboratory analyses show that each of these sequences transmits sound waves with characteristic velocities, and seismologists studying sound propagation in the deep ocean basins have roughly equated observed crustal velocity layering with rock type in the standard ophiolite model. From mapping of seismic velocities, the subseafloor depth of the top of the mantle (the Moho), and thus the apparent thickness of normal ocean crust, appears to be remarkably uniform at about 6 to 7 kilometers.
However, there are several problems with relying too heavily on the ophiolite and velocity models. First, while ophiolites offer the convenience of being able to walk the outcrop and directly study geological relationships, there is considerable debate as to their value for interpreting the structure of normal ocean crust. Many ophiolites are known to have originated in unusual tectonic settings such as behind island arcs or above subduction zones, and it is uncertain how representative they are of crust beneath most large ocean basins. Second, it is likely that the standard ophiolite model is overly simplistic. Spreading rates and the associated levels and constancy of magma production vary dramatically among mid-ocean ridges, so the magmatic and tectonic structure of deep ocean crust probably seldom approaches the ideal model, particularly where magma supply is limited. Finally, our suppositions about the relative uniformity of crustal thickness and composition, as inferred from velocity-lithology correlations, could be seriously in error. In particular, normally high-velocity ultramafics of the upper mantle may become altered by seawater that penetrates deeply along faults and fractures. Such alteration can dramatically reduce sound velocity in these rocks, making them appear seismically to be part of the ocean crust and thus causing us to overestimate true crustal thickness.
To resolve such uncertainties, we are still faced with the need systematically to sample the entire crustal and uppermost mantle sequence within the ocean basins. To date, the best path to this goal has been to sample local tectonic exposures, but because of the intermittency and incompleteness of outcrops, this has been considered a piecemeal solution at best. Recent discoveries, however, indicate that complete, naturally occurring crust-mantle cross sections may be exposed on the seafloor by unusually long-lived faults. These sections have the potential to yield long-sought answers to questions about the structure and origin of oceanic crust.
Nature Opens the Door
To understand how tectonism exposes cross sections of the ocean crust, we must first consider the typical along- and across-axis structure of mid-ocean ridge (see second figure from top). The spreading axis, where magma wells up to form new ocean crust, is offset by discontinuities to form numerous individual spreading segments that extend from a few tens of kilometers to more than a hundred kilometers along the ridge axis. At the axes of slow-spreading ridges (where total spreading rate is less than about 35 to 40 millimeters per year), seismic studies of spreading segments, together with gravity studies of density distribution and direct seafloor sampling by dredging and submersible, indicate that crustal thickness is greatest near segment centers and that it thins markedly toward the discontinuities at segment ends. From this, we infer that upwelling of magma from the mantle is relatively focused near the centers of slow-spreading segments. At faster-spreading ridges, in contrast, crustal thickness tends to be more uniform along the length of spreading segments, and upwelling of magma therefore appears to be more evenly distributed along a segment.
Another significant difference between slow- and fast-spreading ridges is in magma supply. The rate of magma supply at slow-spreading ridges is relatively low compared to the rate of axial crust extension, while the rates are more comparable at fast-spreading ridges. The result is that tectonic extension, expressed in the form of large normal faults, is much more profound at slow-spreading ridges, particularly at segment ends. These faults cut deeply into and sometimes through the ocean crust. If extension occurs for a long enough period on a fault, then a significant portion of the ocean crustal section, and possibly even upper-mantle rocks, could be exhumed along the fault plane. Do these geological cross sections actually exist on the seafloor, and, if so, under what conditions are they formed?
Recent discoveries suggest an affirmative answer to the first question. In 1996, both British and US expeditions to the slow-spreading Mid-Atlantic Ridge found remarkable seafloor edifices, each of which appears to have originated by long-lived slip on an individual normal fault. These features, which we term “megamullions,” have two distinctive characteristics: first, a domelike or turtleback shape extending over a diameter of some 15 to 30 kilometers, and second, conspicuous grooves or corrugations (mullions) that formed as part of the faulting process and that parallel the direction of fault slip over the domed surface. (See third figure from top.) We have interpreted the mega-mullions as footwall blocks, that is, blocks exhumed from beneath normal faults. (See figure above right.) In the case of megamullions, these faults have been unusually long-lived and areally extensive, and we refer to them as “detachment faults.” Once megamullions were recognized, our subsequent study of existing, detailed multibeam bathymetric data has identified a total of 17 of these features between about 21° N and 31° N on the Mid-Atlantic Ridge, and a number of other megamullions have been recognized where strong tectonic extension occurs on slow- to intermediate-spreading mid-ocean ridges in the Indian Ocean.
Each detachment fault that forms a mega-mullion has three notable features: a breakaway zone where the fault began, the exposed fault surface that rides over the megamullion dome, and a termination, which usually is marked by a valley and adjacent ridge. By identifying breakaway and termination ages from dated seafloor magnetic anomalies, we can establish that the faults forming the North Atlantic megamullions accommodated slip for periods between 1.0 and 2.6 million years, with an average period of 1.5 million years. The original dip of the faults is uncertain, but they probably dipped at up to about 45°, much like most seismically active normal faults presently observed at the axes of mid-ocean ridges. With this dip, the faults would have exhumed a full 6-kilometer-thick crust and even exposed the underlying mantle within less than a million years. As these rocks are drawn out from beneath the fault, the footwall “rolls over,” laying out the geological cross-section across the surface of the megamullion. The rollover also flexes the brittle footwall much like bending a wooden ruler—the upper part of the footwall block is under tension and new normal faults break through the detachment surface.
Under what conditions do these long-lived faults develop? In terms of their position on the seafloor, all megamullions identified thus far appear near segment ends at inside-corner (IC) locations (that is, within the bights between the actively slipping sections of ridge-axis discontinuities and the spreading axis). Geophysical data and recovered rock samples show that ocean crust at inside corners of slow-spreading ridges is intermittently thin or missing compared to relatively normal thickness, outside-corner (OC) crust on the opposite side of the spreading axis. This seems to be best explained by consistent orientation of faults dipping from inside corners toward outside corners; in this way the upper crust (hanging wall) is frequently stripped from the inside-corner footwall. However, this process cannot be continuously occurring on a single fault. If it were, all inside corners would exhibit megamullions (but only a small percentage do), and they would almost exclusively expose mantle rocks (which they do not). Rather, it appears that a single fault is normally active for only a short period of time (a few tens to hundreds of thousand of years) before it is abandoned and replaced by a new fault closer to the spreading axis.
What, then, promotes slip on a single fault for periods of up to 2 million years or more? And what eventually causes the fault to be abandoned? Clearly, for a fault to remain active it must be weak in comparison to adjacent crust where another fault might otherwise nucleate. Recent laboratory studies have demonstrated changes in deformation mechanism at sub-seafloor depths where faults in the brittle lithosphere flatten out into zones of plastic deformation in hotter, deeper rocks. These changes weaken the rock by a factor of ten or more compared to surrounding, unfaulted rocks, and they thus promote slip in the shear zone. Slip probably is also enhanced by the occurrence of weak serpentinites along faults at shallower levels within the brittle litho-sphere. Serpentinites form as sea-water percolates down fractures and reacts with ultramafic rocks in the lower crust and upper mantle at temperatures ranging from about 100° to 500°C.
These kinds of weakening, however, are not the entire answer to the question of fault longevity because they probably occur on many (if not most) faults in slow-spreading crust, whereas truly long-lived faults are few and far between. A more complete explanation can be devised if we consider the temporal variability of magmatism at the spreading axis. Episodes of magmatism are known to occur at time scales of tens to hundreds of thousands of years at slow-spreading ridges, and recent analysis of gravity data over ridge flanks also indicates a predominant cycle at a period of two to three million years. Megamullions consistently correlate with the parts of these longer cycles where the gravity data indicate the presence of thin crust and predominantly amagmatic extension. Thus, it appears that persistent slip on a fault occurs while the spreading axis is relatively cold, but that when magmatism is renewed, it heats and weakens the lithosphere, new faults form, and the long-lived fault is abandoned. The relative rarity and the locations of megamullions suggest that completely amagmatic extension is not common even in slow-spreading crust, and when it occurs it is restricted mostly to segment ends near ridge-axis discontinuities.
Windows of Opportunity
The cross sections through the crust and into the upper mantle that appear to be laid out across the surface of megamullions offer exciting new windows of opportunity finally to sample the oceanic lithosphere in detail. Marine geologists are now proposing to drill a series of half- to one-kilometer-deep holes, aligned in the direction of fault slip across the surfaces of megamullions. From these cores, we should be able to construct a composite, but relatively complete, picture of the entire crust-mantle section, without ever having to drill the 6 kilometers or more that would be required to reach the mantle beneath normal-thickness ocean crust.
There is, however, much to be done before the drilling occurs. A few samples have been obtained from megamullions, and as expected they include lower-crustal and upper-mantle rocks. Nonetheless, most of our interpretations are based on remotely sensed data such as gravity and bathymetry, together with models of the faulting process. Detailed surveys and sampling with deep-towed instruments and submersibles like Alvin are needed fully to document the geology of the outcrops and to select optimum locations for drilling. This mapping and sampling will also provide critical information for correlating data between drill holes and to understand the three-dimensional internal structure of the crust and mantle.
In addition to providing lithospheric cross sections, megamullions have the potential to yield significant insights into other outstanding geological problems. For example, it has long been thought that the primary source of marine magnetic anomalies lies in the upper, extrusive section of the oceanic crust. Yet, at megamullions, where this layer is thought to be missing, perfectly normal magnetic anomalies usually are developed. Either we are grossly mistaken in our interpretation of these features, or else the lower part of the ocean crust contains a substantial magnetic signature. Near-seafloor magnetic studies and laboratory analyses of magnetization in recovered rock samples will resolve this question.
Although the focus here has been on oceanic lithosphere, we also can benefit from and contribute to the study of tectonics in continental crust (see two photos above). Megamullions have dimensions, shape, and an apparent mode of origin via long-lived detachment faulting that are very similar to those of “metamorphic core complexes” exposed in extensional mountain belts such as those found in the southwestern United States. In these areas, domed core complexes have exhumed rocks from some 15 kilometers deep in the continental crust, and their structure and deformation history have been studied extensively for more than two decades. Cross-pollination of insights gained from both the continental and oceanic realms will allow us to develop a much more comprehensive understanding of extensional tectonism and the intriguing exposures of deep lithosphere both on land and in the ocean basins.
This mid-ocean ridge research has been supported by the National Science Foundation, the Office of Naval Research, and WHOI’s Andrew W. Mellon Foundation Endowed Fund for Innovative Research. The author is indebted to colleagues Jian Lin, Marty Kleinrock (Vanderbilt University), Greg Hirth, Henry Dick, and Joe Cann (University of Leeds, UK) for stimulating discussions about tectonism in slow-spreading crust, and to Eric Frost (San Diego State University) and Kip Hodges (MIT) for superb field exposition of extensional tectonics in the Basin and Range.
Born and raised in the natural geological laboratory known as the Black Hills of South Dakota, Brian Tucholke has been a geologist from the time he first crawled off his baby blanket and grabbed a fistful of dirt. As one of the Ph.D. students entering the MIT/WHOI Joint Program in its inaugural year, he (mostly) gave up the bedrock stability of the western mountains for the rolling sea, and he has since been on more than 25 oceanographic research cruises to investigate the flow of abyssal currents, seafloor sedimentation patterns, the paleoceanographic history of ocean basins, the structure and evolution of continental margins, and the tectonics of mid-ocean ridges. The mark of his passage can often be seen in the imprint of his cowboy boots, several decrepit pairs of which he has buried at sea in oceans ranging from the North Atlantic to the Bellingshausen Sea off Antarctica.
Spring 1998 — Our knowledge of the physical characteristics of Earth’s deep interior is based largely on observations of surface vibrations that occur after large earthquakes. Using the same techniques as CAT (Computer Aided Tomography) scans in medical imaging, seismologists can “image” the interior of our planet. But just as medical imaging requires sensors that surround the patient, seismic imaging requires sensors surrounding the earth.
At the end of 1997 there were 105 high quality seismic stations on continents and islands around the globe as part of the US sponsored Global Seismic Network. In addition, smaller networks of land stations are sponsored by Japan and France. Ideally, seismologists would like to establish 128 seismic stations uniformly distributed over the surface of the globe. Why 128? Seismic processing, like processing on a personal computer, is more efficient if carried out with binary, that is, base 2, numbers. 128 is 2 to the seventh power. The next smallest power of 2, 64, would not give sufficient resolution of earth structure. The existing land-based networks are more dense than this. The next biggest power of 2, 256, would be simply too expensive. The 128 stations would be separated by about 2,000 kilometers. Since over two-thirds of the earth’s surface is covered by water, “uniformly distributed” implies that about 20 stations need to be located on the deep ocean floor, far from continents or islands. The goal of the Ocean Seismic Network (OSN) program is to develop the methodology and instrumentation necessary for continuous, high quality, seismic observations in the deep oceans.
There are three major challenges in establishing a seafloor seismic observatory: installing the sensor to obtain comparable quality to land or island stations, providing power to the instrument, and retrieving the data.
A ‘modern’ seismic installation will faithfully measure accelerations of the solid earth in three dimensions (vertical, north-south, and east-west) down to the quietest earth motions over a broad band of frequencies. In the absence of earthquakes and such human activity as traffic, shipping, and explosion tests, the earth’s surface vibrates in response to wind, ocean currents, and ocean waves. While the human ear responds to vibrations in air spanning frequencies from 20 hertz to 20 kilohertz, seismometers used to study whole earth structure respond to lower frequencies, from 0.002 hertz to 10 hertz. At 0.002 hertz the seismometer will rise and fall once in about 8 minutes.
Useful signals for tomographic studies of the earth have frequencies around 1 hertz. To get a feeling for the sensitivity of a modern seismometer, consider an earthquake and a receiving station that are 20° apart (about 2,200 kilometers or the distance from Woods Hole to New Orleans). A very large earthquake, with a magnitude of about 8, will cause displacements at the seismometer of about 1 centimeter. A small earthquake, with a magnitude of about 4, will cause displacements of about 10 millimicrons. (A millimicron is a billionth of a meter.)
The seismometer will also be responding faithfully to true earth ambient noise with frequencies around 1 hertz and displacements less than 1 millimicron. These correspond to accelerations of about one-tenth of a billionth of the acceleration due to gravity. Directly above a large, shallow earthquake, the surface of the earth will be displaced over a meter, but a high quality seismometer can measure displacements as small as a few millimicrons. The ability to faithfully measure small variations occurring at the same time as large variations is called dynamic range. A modern digital seismometer uses 24 bits to represent a number and has a “nominal” dynamic range of 144 decibels. If the full dynamic range were available, this would correspond to displacements ranging from a millimicron to about a centimeter. It is rare for a seismic station to be directly over a large earthquake. Special “strong motion” sensors are used for the very large displacements expected in earthquake-prone areas such as Japan and California.
A seismometer in Montana can measure ground vibrations that are excited by storms over the North Pacific and North Atlantic Oceans. Wave interaction during the storm at sea creates low frequency “sound” that travels through the solid earth to stations deep within the continents.
Experience on land with Global Seismic Network stations indicates that the best results are obtained by placing the seismic sensor in a borehole at about 100 meters depth. Although expensive, this has two advantages. First, a sensor set firmly in bedrock moves faithfully with the true motion of the solid earth. (Layers of soils, gravels, sands, and muds that lie on the hard rock basement distort the true earth motion.) Second, the ambient seismic noise levels, which effectively limit our ability to observe small earthquakes, are reduced considerably by getting away from the surface. The ambient noise, which is predominantly generated from atmospheric and oceanographic effects at the surface, is trapped in interface waves that attenuate rapidly with distance away from the interface. By getting below the region of interface waves, the ambient noise is considerably reduced. The International Ocean Network and Ocean Seismic Network communities are actively advocating the drilling of boreholes on the seafloor for seismic installations, and they are modifying the high quality borehole seismometers used on land for operations in the deep sea.
The digital data acquired by the borehole seismometer can be recorded on the hard drive of a seafloor computer. Data is acquired at 20 samples per second on each of three axes. At this data rate, a seismometer would fill the 1 gigabyte drive on a personal computer in about two months. Three approaches are being considered for data retrieval. Where retired submarine telephone cables cross an area, seismometers could be tapped into a seafloor junction box. Data could be retrieved and power could be supplied over the cable continuously in real time, an ideal situation. For remote areas without cables, the “brute force” approach is to record data autonomously in the seafloor package and grapple for the package periodically, say, once per year from a supply vessel. Batteries could be changed out at the same time. Though technically the simplest approach, the major disadvantage of this system is that the data would not be available until many months after the events of interest. The third idea is to moor, in the deep ocean near the seismometer, a buoy with a transmitter that can communicate via satellite to shore. This increases the power requirements considerably, but data could be available in near real time. However, maintaining a mooring continuously in the deep ocean is an expensive and challenging task.
A state-of-the-art broadband borehole sensor for deep sea applications draws between 4 and 12 watts of power continuously. (Compare this to leaving a 100-watt light bulb on in your home.) Improving the power consumption of deep sea seismometers and recording systems is an active engineering development task. A pressure housing 4 feet long and 12 inches in diameter filled with lithium batteries can run a 4-watt system for four months for about $10,000. Batteries alone for a 12-watt system would cost about $100,000 per year. Not inexpensive, but at least it can be done.
The Ocean Seismic Network Pilot Experiment
The primary goal of the Ocean Seismic Network (OSN) Pilot Experiment is to learn how to make high quality broadband (0.003–5 hertz) seismic measurements on the seafloor in preparation for extending the Global Seismic Network to the ocean basins. This experiment was carried out at a drill site (Ocean Drilling Program Hole 843B) 225 kilometers southwest of Oahu. In January and February 1998, we deployed three broadband seismic instruments: a borehole seismometer, a sensor buried surficially in the sediments, and a sensor resting on the seafloor. The seismometer was placed in the borehole using a Wireline Reentry System (see figure opposite). The instruments were recovered in June 1998.
While at the site in February, we acquired some sample time series data from each of the broadband seafloor installations. While tethered to the borehole sensor we recorded surface waves from the magnitude 6.1 event that occurred in Afghanistan on February 4 (top figure above). (The seafloor and buried sensors were not yet in place during this event.) Ambient noise spectra of the three components of the borehole sensor show that the horizontal components are noisier than the vertical components below about 0.04 hertz and that the horizontal components are quieter than the vertical components above the microseism peak at about 0.1 hertz (lower figure at left). Comparison of ambient noise in the borehole on the seafloor with a state-of-the-art sensor on the island of Oahu shows that seafloor stations have comparable noise levels to island stations (figure below) from about 0.02 hertz to 1.5 hertz. Scientists use these ambient noise curves and observations of signals from different sizes of earthquakes to evaluate the performance of a station.
The international community plans to install six OSN stations in the next five years. The Ocean Drilling Program has just completed a borehole on the Ninety-East Ridge in the Indian Ocean that will be used by the French for a borehole seismic station. At least one borehole station is planned near Japan. In fall 1998 a WHOI-led team installed a junction box on the retired AT&T cable between Hawaii and California, and ODP plans to drill a hole for the borehole seismometer at the site. There are also plans for two boreholes in the equatorial Pacific. The sixth site is on the Mid-Atlantic Ridge where a borehole already exists.
The goal of uniform seismic coverage of planet Earth is within sight, but we have a long way to go.
This work was supported by the National Science Foundation with additional support from Incorporated Research Institutions for Seismology, Joint Oceanographic Institutions, Inc., Scripps Institution of Oceanography, and a Mellon Independent Study Award from Woods Hole Oceanographic Institution.
Acknowledgments: The Ocean Seismic Network Pilot Experiment was a collaborative effort involving about 20 scientists, engineers, and technicians from Woods Hole Oceanographic Institution (WHOI) and Scripps Institution of Oceanography (SIO). In addition to the author, the principle investigators from WHOI were John Collins in the Department of Geology and Geophysics and Ken Peal in the Department of Applied Ocean Physics and Engineering. At Scripps the principal investigators were Fred Spiess and John Hildebrand from the Marine Physical Laboratory and John Orcutt and Frank Vernon from the Institute of Geophysics and Planetary Physics. Of course, the operations at sea also involved the officers and crews of the University-National Oceanographic Laboratory System research vessels. The deployment cruise was carried out on R/V Thomas G. Thompson (University of Washington) under the command of Captain Glen Gomes. Recovery was accomplished with R/V Melville (Scripps Institution of Oceanography) under the command of Captain Eric Buck.
Ralph Stephen’s Ph.D. thesis at the University of Cambridge was based on the first borehole seismic experiment carried out in the deep ocean in 1977, and he has been pursuing marine borehole seismology at WHOI ever since. In addition to this field work, Ralph develops synthetic seismogram methods for predicting sound propagation in the seafloor. His hobbies include being a husband and father.
Spring 1998 — I never imagined I would spend six weeks of my life “wandering around” the seafloor exploring an 11 million year old beach, and it never occurred to me to look for a fossil island. But that’s what I did, and that’s what we found on two research voyages separated by more than a decade.
But, first, a bit of history: During World War II, Princeton mineralogist Harry Hess convinced US officials to keep Navy ships’ echo sounders running as they crisscrossed the Pacific. Hess, a captain and later admiral in the Navy, even managed to have his men bring back rock samples when they returned to the ship after Pacific Island landings. We not only won the war, but also identified the mid-Pacific mountain range now known as the East Pacific Rise—a discovery that proved fundamental to the then developing theory of plate tectonics.
Thus began modern marine geology and geophysics. Prompted by Hess’s discoveries, when peace returned, American scientists initiated a systematic study of this little known region. One of the first results was the discovery that the ocean crust appeared to present a consistent, layered seismic structure composed of sediments (layer 1 at the seafloor) followed by two layers with rapidly increasing sound velocity, and then a reflector, dubbed the Moho after the its discoverer, a Czechoslovakian seismologist named Andrija Mohorovicic. What surprised scientists was the Moho was found at a very shallow, nearly constant depth of 6 to 7 kilometers, and that seismic layers 2 and 3 above also appeared uniform in the oceans, whereas this boundary ranges from 35 to 60 kilometers deep beneath continents.
This led to a great debate about the nature of the Moho beneath the oceans and the composition of the layers above it. Hess proposed in his now famous 1962 paper The History of the Ocean Basins * that the Moho was an alteration front created by hydrothermal solutions converting the earth’s mantle to a lighter rock known as serpentine. Serpentine forms by hydration of the primary mantle rock peridotite, which is named for the gem stone peridote (the most common mineral, known to scientists as olivine). The uniform depth of the Moho, then, represented the upper temperature and pressure limit of peridotite and olivine as stable elements beneath the ocean ridges. Hess thought Layer 2 was the volcanic carapace formed as lavas erupted above the hydrothermally altered mantle rock.
This theory was rejected, however, because laboratory experiments on serpentine failed to produce the exact acoustic velocity characteristics inferred for seismic layer 3. Instead a model where the Moho was the crust-mantle boundary, with layer 3 composed of the rock known as gabbro overlying the unaltered mantle peridotite has become the generally accepted paradigm. Gabbro is a coarse grained rock similar in composition to the basaltic lava that erupts from submarine volcanoes. This rock is found exposed on land where deep layers of the earth have been uplifted onto mountain ranges and exposed by faulting and erosion. It is believed to represent the frozen remains of magma chambers where lava rising out of the earth’s interior pooled deep in the crust. Gabbro is also found in “ophiolite” complexes, bits of fossil ocean crust and mantle commonly found on land in tectonic plate collision zones and in island arcs (see map in Peter Kelemen article). There it underlies extrusive pillow lava and sheeted dikes and overlies mantle peridotite; it is also believed to represent the remains of magma chambers. The ophiolite sequence of pillow lavas–dikes–gabbros–periodite provides a convincing match to the seismic layering.
However, ophiolite crust is generally much thinner than the seismic crust in the oceans, and the lavas there often have quite different chemical compositions than those on the seafloor. Do ophiolites represent broader ocean crust characteristics? Hess and geophysicist Walter Munk (Scripps Institution of Oceanography, University of California, San Diego) cooked up the idea of drilling a hole into the earth’s mantle and sold it to a group of their colleagues, known as the “American Miscellaneous Society,” who in turn sold it to Washington: Drill a hole into the mantle all the way through the ocean crust. Dubbed “Project Mohole,” the idea eventually foundered on the rocks of the congressional budget office in cost overruns and ballooning budgets. It was later reborn, however, as the successful Deep Sea Drilling Project and its successor, the Ocean Drilling Program. (See Oceanus, Vol. 36, No. 4 for “An Abridged History of Deep Ocean Drilling.”) These programs have incrementally sought to drill deeper, first through the sediments, then through layer 2, which proved to be, as thought, pillow basalts and sheeted dikes, the remains of the lava that erupted onto the seafloor and its feeder channels. But there the drilling stopped. The material being recovered was too brittle, too broken, and it seemed too difficult to go deeper.
In 1984 my colleague Jim Natland (Scripps) and I proposed using tectonic windows in the ocean crust to get to the deeper rocks. Simply go someplace where Mother Earth had done a lot of work for us by tectonically stripping off the shallow layers, and start deep. In 1986, we surveyed the seafloor south of the island of Mauritius at a great oceanic transform fault named Atlantis II Fracture Zone, looking for a place to drill. The survey produced a map of a 200-kilometer long, mid-sea mountain range flanking the fault. In the middle of these mountains was a remarkable flat-topped peak rising up some 5 kilometers from the seafloor. On the walls of Atlantis Bank, as we named the feature, we sampled gabbro and peridotite, proof that drilling here should be productive.
Then, in 1987, Jim and I returned to Atlantis Bank aboard the drill ship JOIDES Resolution along with several WHOI colleagues including co-chief scientist Dick Von Herzen, who had backed the project from the beginning. Over 16 days of drilling we recovered 437 meters of gabbro from a 500 meter hole, making Hole 735B the most successful hard rock hole ever drilled in the oceans. Although we expected to find gabbro in the lower ocean crust, these rocks still produced a lot of surprises. Scientists had long debated whether the lower crust was crystallized from one great steady-state magma chamber many kilometers across lying beneath the ocean ridge. There was no proof for that here—rather there were many small intrusions, hundreds of meters in dimension, not kilometers. Crystallization here was ephemeral and short-lived. And, a great surprise—the magmas deep in the earth were moved around as the rocks deformed when the ocean crust stretched and ripped apart during seafloor spreading. The crust simply looked a lot different than we expected.
The Indian Ocean is a tough place to get to. It took 10 years to get back. Finally, Jim (now at the University of Miami) and I returned last fall aboard JOIDES Resolution as co-chief scientists and started drilling again to deepen Hole 735B. We drilled another kilometer, and surprise—everything changed again.
After careful study, the results showed that the deep ocean crust is compositionally zoned. Its upper layers are enriched in iron and titanium. In the lefthand figure on page 30, this is reflected in a change in the amounts of different rock types. Many large gabbro intrusions on land are also compositionally layered, but not this way. In this piece of ocean crust, the layers are created as tectonic forces drive the last liquid, with iron and titanium in solution, out of numerous small magma intrusions when they solidify—pushing it from the lowermost layers upward to “freeze” at the top of the section. Our discovery of this “synekinematic igneous differentiation” reveals a whole new process never before dreamed of—unique to ocean ridges.
In the 10 years we had been absent, British colleagues led by Tim Minshull (Cambridge University) had been by to conduct a seismic experiment, which showed that the Moho beneath Atlantis Bank was some 5 to 6 kilometers down. This was much deeper than expected or seemed reasonable in an area where the ocean crust was believed considerably thinner than the usual 6 or 7 kilometers, and where shallow seismic layer 2 (1.5 to 2 kilometers) was already gone.
Well, where was the mantle here anyway? If it were on the walls of Atlantis Bank, then the Moho beneath wasn’t the crust-mantle boundary. That’s what we set out to find in spring 1998. To do this, Paul Robinson (Dalhousie University), Chris MacLeod (Cardiff University), Simon Allerton (University of Edinburgh), and I organized an international expedition on the British Antarctic Survey ship James Clark Ross. Using British Geologic Survey rock drills and the Canadian remotely operated vehicle ROPOS (Remotely Operated Platform for Ocean Science), we went back to see if we could find the crust-mantle boundary on the wall of the transform, and to map out the great gabbro massif through which we were attempting to drill. What we found was more surprises. We did find the crust-mantle boundary—outcropping on the side of Atlantis Bank—only a few hundred meters deeper than where we stopped drilling. So it appears the Moho is an alteration front, as Harry Hess once supposed, at least in this one spot in the ocean. And we did, with our British and Canadian colleagues, map out an enormous block of gabbro over the top of the bank—proving that the crust here is relatively intact and not a great tectonic jumble. But we also found a fossil island.
And what about that fossil beach? About two- thirds of the bank is covered by limestone, with ripple marks, just like those in the sand at a modern beach. However, these were “frozen,” lithified as what was once an island sank beneath the waves millions of years ago. There are little pot holes ground into gabbro rock, still partially filled with pebbles and sand, and headlands and fossil seastacks (isolated erosional remnants of the island). No beach umbrella, but an old rotting fishing net—the place had a very large population of lobsters, crabs, sharks, sea fans, siphonophores, sponges, and other critters. It seems that long ago, the whole massif popped up out of the earth some 6 or 7 kilometers at the intersection of the Southwest Indian Ridge and the Atlantis II Fracture Zone to rise above sea level. This great tectonic island with an area of at least 25 square kilometers then slowly subsided back beneath the waves to its present position some 700 meters below the sea surface. It is a remarkable place, both for its modern biologic community and its unique fossil island characteristics. Flat topped seamounts called guyots are well known in the Pacific. They are submerged volcanic islands whose tops are worn away by wave action as they sink beneath the sea surface. But this is the first tectonic guyot anyone has ever studied. And with it comes a unique paleontological record. Between the many seastacks, boulders, and ledges of gabbros, we deployed our short, over-the-side rock drills—and came back with a lot of limestone in addition to the gabbro we sought. There were fossil clams, snails, a limestone made of sea urchin spines, and a kind of limestone known as oolitic, something that forms in lagoons. But they’re all now 700 meters down. Jim, formerly a Moho chaser like myself, found it so fascinating that he has converted himself into a carbonate sedimentologist to explore the mysteries of that fossil island.
This research was funded by the US National Science Foundation with additional support for the author from WHOI’s Van Allen Clark, Jr., Chair. International support was provided by the Canadian Natural Sciences and Engineering Research Council and the British Natural Environment Research Council for remotely operated vehicle and rock drilling operations at Atlantis Bank aboard RSS James Clark Ross.
Henry Dick first came to Woods Hole Oceanographic Institution from Yale University in 1975 as a postdoc with Bill Bryan of the Geology and Geophysics Department. He has been a Senior Scientist since 1990. He has a remarkably dedicated wife named Winifred and three small children, Helene, Spencer, and Lydia, who think their daddy goes to sea too much. He’s promised them he won’t even look at the ocean for at least a year, much less get his feet wet—after his next cruise this fall!
1998— The MELT Experiment was the largest seafloor geophysical experiment ever attempted, and one of its major components was MT, the magnetotelluric technique. MT offers a valuable tool toward the MELT Experiment’s goal of probing the earth’s inaccessible deep interior. But the technique remains something of a mystery even to many marine scientists. It has been used widely on land, particularly for regional-scale surveys, but only a few full-scale MT surveys have been carried out on the seafloor.
The primary data collected by marine MT experiments are measurements of changes in the earth’s electrical and magnetic fields at the seafloor. These fields are affected by electromagnetic currents within the earth, and here’s where MT’s apparent complexity starts—because the source of these currents is not within the earth, but rather in the ionosphere.
Charged particles, emitted from the sun as a solar wind, become trapped in the ionosphere by the earth’s magnetic field. These moving charges essentially create a variety of electric currents encircling the earth. If the earth were a perfect insulator, like space, that would be the end of the story. But the earth can conduct electricity. As these ionospheric currents flow around the earth, they generate a response within the planet itself. More specifically, the pattern of ionospheric currents induces almost a mirror-image pattern of currents within the earth.
These so-called “induced image currents” cause changes in the earth’s electric and magnetic fields. These changes depend on the conductivity of the earth’s interior, which, in turn, is determined by the composition and structure of the materials that constitute our planet’s interior. Thus, by measuring changes in Earth’s electric and magnetic fields at the surface, we can effectively deduce its electrical conductivity and reveal its interior structure. As CAT scans reveal images and frameworks that enable us to learn about the workings of the human body, MT experiments similarly provide essential cutaway views that allow us to learn about processes taking place within our planet.
Like standard alternating currents in most households, which have a frequency of 60 Hertz, or one cycle per 1/60 of a second, induced image currents also alternate—though they do so over a wide range of frequencies. The variations, or frequencies, we use in seafloor MT range from periods of about 100 seconds to several hours. These variations are caused by the chaotic nature of the events that entrap ions from the solar wind, as well as by more regular events, such as the earth’s daily orbit around the sun. The important point is that different frequencies penetrate the earth to different depths. If induced image currents came in only one flavor, we would be able to image the earth’s interior at only one depth. As it is, higher-frequency currents (with one cycle per 100 seconds, for example) don’t penetrate deeply and can tell us about structure 10 to 15 kilometers deep; the lowest-frequency currents (with one cycle per several hours) can tell us about depths of several hundred kilometers.
The goal of the MELT Experiment was to map basaltic melt, from its source within the mantle to the base of the oceanic crust at the mid-ocean ridge crest. While the earth can conduct electrical currents, most rocks, including those comprising the mantle, do not conduct particularly well. This situation changes considerably when melt is present: Pure basaltic melt is several orders of magnitude more conductive than olivine, a common mantle mineral. In the mantle melting column, we do not expect to see pure melt, nor anything like it, but rather some distribution of streams and pools of liquid melt within a matrix of solid mantle rocks. In this case, how the melt is distributed is important. It is possible to think of the melt as a network of wires that connect parts of the mantle. If the melt forms a well-connected network through the rock, electric currents can flow and the mantle will be electrically conductive. Of course, reality is more complicated and other factors, such as water dissolved in the mantle rock, can affect conductivity. These other factors are also important for understanding the whole process of melt production.
The MT component of the MELT Experiment was a truly multinational effort involving more than a dozen scientists from Woods Hole Oceanographic Institution and Scripps Institution of Oceanography in the US, and from France, Japan, and Australia. Each group contributed instruments to the array and played a role in the data analysis. From June 1996 to June 1997, 47 instruments were deployed at 32 seafloor sites to measure the time variations of the electric and magnetic fields. Two lines were set out. The main southern line had 19 sites and crossed a magma-rich segment of the East Pacific Rise ridge crest, extending 200 kilometers on either side of the crest. The second line of 13 sites crossed the ridge to the north on a magma-starved ridge segment, extending 100 kilometers on either side of the axis.
Each group’s instruments essentially did the same thing: measure changes in the electric and magnetic fields at the seafloor. But each group accomplished this in slightly different ways, deploying very different-looking instruments. As in all marine experiments, the environment makes seafloor MT measurements more difficult to make, but in one way nature helps us. The ocean is electrically very conductive and acts as a screen against electromagnetic noise—extraneous signals from other sources that would confuse interpretation of the data. On land, power lines, for example, can be a nuisance. The seafloor, however, is electrically quiet, making it possible to measure very small electric field variations. The other part of the MT signal is the seafloor magnetic field—not the steady field trapped in lavas and used to identify magnetic reversals, but the magnetic field variations linked to ionospheric currents.
To a first order, the ratio of the electric to the magnetic field at the earth’s surface is a direct measure of the earth’s electrical conductivity. We calculate this ratio for a range of current frequencies using modern processing techniques. To produce a model of the earth, data from all instruments have to be examined through a process of numerical inversion. The interaction of induced currents in the earth with the conductive bodies we hope to image (such as the melt column) affects the electric and magnetic fields over a wide region of seafloor. Generally, it is not possible to look at data from a single instrument and interpret the underlying structure. Instead, we have to use computer modeling to predict the fields that the mantle would create and compare these answers to data from all the instruments. The model is updated to improve the agreement and the process is repeated until a satisfactory model is found. There are many pitfalls involved in this process, as well as different ways of carrying it out. The groups involved in the MELT Experiment have been using a variety of methods over the past few months, and we are in the process of comparing results and discussing their implications.
The MT analyses are still in their early stages, but some first-order results are beginning to come through. The MT data show an asymmetrical distribution of melt between the areas west and east of the ridge crest, with a more extensive region to the west. The melt column also appears to be a broader feature, with a low percentage of melt in it, rather than a narrow vertical column of melt directly beneath the ridge. This indicates a more passive flow of mantle toward the ridge crest. Deeper, we see some evidence for a conductive mantle at depths greater than 150 kilometers. If this proves to be true, it could be evidence for deeper melting—deeper than the part of the mantle generally believed to be responsible for most melt generation. However, in the final analysis, water dissolved in the mantle rock may prove an important factor in mantle conductivity at this depth.
Funding for the MELT Experiment was provided by the National Science Foundation through the RIDGE Program. The many people involved in the MT component of MELT include: Alan Chave, Bob Petitt and John Bailey (WHOI), Jean Filloux and Helmut Moeller (SIO), Pascal Tarits (UniversitÉ de Bretagne Occidentale), Martyn Unsworth and John Booker (University of Washington), Graham Heinson and Anthony White (Flinders University, South Australia), and Hiroaki Toh, Nobukazu Seama and Hissashi Utada (University of Tokyo).