The Great Heat Engine: Modeling Earth's Dynamics
The Earth embodies the paradox at the heart of geology: a seemingly solid, relatively fixed subject for study, it is actually a dynamic planet whose seething, continuous undercurrent of activity is manifest in the earthquakes and volcanoes that science has only begun to be able to model and predict. These and other cataclysmic symptoms of the Earth's internal dynamics command attention, in part, because they transpire within human time frames—minutes, hours, and days. They make news. Geologic time, by contrast, is measured in millions and billions of years and yet marks the unfolding of an even more dramatic scenario—the movement of entire land masses over thousands of kilometers, though at a rate of only a few centimeters a year, and the creation of mountains, ridges, valleys, rivers, and oceans as a consequence. Further, since most of the Earth's internal processes develop hundreds and thousands of kilometers below the surface, scientists have only indirect evidence on which to base their theories about how these many phenomena tie together.
The search is on for a harmonious model of geologic evolution that will explain and embrace all of the Earth's internal features and behavior (largely inferred) and its tectonics—the deformations of its crust that form the Earth's surficial features. And while no such comprehensive synthesis is in sight, a number of illuminating models accounting for many of the observable phenomena have been developed in the last 30 years, including models explaining the thermodynamics and magnetohydrodynamics of the Earth's core. A number
of the other models and theories of how the Earth works are constructed atop the reigning paradigm of geology, plate tectonics, which has only been widely accepted for less than three decades. Plate tectonics has grown from an intuitive suspicion (when maps produced by explorers showed the uncanny complementary shape of the continental shores on either side of the Atlantic, as if the continents had been unzipped), to an early-20th-century theory called continental drift, to the major set of ideas that unify today's earth sciences.
Plate tectonics is no causal force, but merely the surface manifestation of the dynamics of this "great heat engine, the Earth," said presenter David Stevenson. These surface manifestations concern civilization in an immediate way, however: earthquakes, volcanoes, and unpredicted disasters continue to cause tragic loss of life and property.
Stevenson, chairman of the Division of Geological and Planetary Sciences at the California Institute of Technology, described the scientific unifying power of plate tectonics and how it has transformed geology into a brace of strongly interdisciplinary approaches: "Geologists look at morphology, at the surface of the Earth, to try to understand past movements. Petrologists examine the rocks found at the surface of the Earth, trying to understand the conditions under which that material formed and where it came from. Volcanologists try to understand volcanos. Geochemists," he continued, anticipating the discussion of recent diamond-anvil cell experiments, "look at traces of elements transported upward by the forces within the Earth, trying to discern from them the history of the planet, its dynamic properties, and the circulation within. Seismologists look at the travel times of sound and shear waves for variations in signature that provide clues to both lateral and radial structure within the Earth," one of the primary sources of geological information on the deep planetary interior. ''Geodesists look at the motions of the Earth as a planetary body, constructing comparisons with distant reference frames, for example, by radioastrometric techniques, and the list goes on and on,'' Stevenson added. Most of these specialists feel the crucial tug of one another's insights and theories, since the Earth as a coherent system of features and phenomena may be best explained with reference to a variety of probes and explorations.
The Frontiers of Science symposium was host to a solid contingent of such earth scientists, many of whom have been major contributors to the lively debate surrounding the newer theories arising from plate tectonics. Stevenson's presentation, "How the Earth Works: Techniques for Understanding the Dynamics and Structure of Planets," served as a broad overview of the field. Emphasizing the lively premise that "ignorance is more interesting than knowledge," he re-
turned often to the theme of describing Earth processes that "we don't understand."
Marcia McNutt, a session participant from the Massachusetts Institute of Technology, has led numerous marine geophysical expeditions to the South Pacific and has contributed significant new additions to the theory of hot spots. Hot spots, upwelling plumes of hot rock and magma that emanate from the deeper layers of the Earth, may provide important clues to one of the basic features of the Earth's dynamics, mantle convection, a process modeled by another session participant, Michael Gurnis from the Ann Arbor campus of the University of Michigan. Another contributor, Jeremy Bloxham from Harvard University, has conducted an exhaustive study of the Earth's magnetic field by reexamining worldwide historical data spanning centuries and has made some strong inferences about the Earth's core. This deep region has also been explored by simulating the conditions of high pressure and temperature that prevail there, an approach described for the symposium's audience by Russell Hemley of the Carnegie Institution of Washington, where pioneering work has been accomplished in the Geophysical Laboratory through the use of diamond-anvil cells.
The session was organized by Raymond Jeanloz of the University of California, Berkeley, and Sue Kieffer from Arizona State University, whose departmental colleague Simon Peacock was also present. The presenters were also joined in the audience by a pair of prominent geologists from the University of California, Santa Cruz—Elise Knittle, a collaborator with Jeanloz, and Thorne Lay, who has provided pioneering seismological insights about the boundary between the Earth's central core and the mantle above it, a region often referred in its own right as the core-mantle boundary. Together these geologists from many allied disciplines ranged over many of the issues in modern earth science, touching on such basic questions as how the Earth came to be, what it looks like inside, how it works, what early scientists thought about it, and how such views have evolved, as well as what theories now dominate scientific thinking and how scientists develop and test them. In considering these and other questions, they provided a vivid picture of a science almost newly born, one in which the technological and conceptual breakthroughs of the latter half of the 20th century have galvanized a truly interdisciplinary movement.
HOW THE EARTH WORKS
The universe was created at the Big Bang, probably about 13 billion years ago, but more than 8 billion years passed before the
cosmic junk cloud where we now reside—itself a residue of the explosive formation of our Sun—began to coalesce into what has become a dynamic and coherent planet, the Earth. As ever larger pieces of the debris called planetesimals aggregated, their mutual gravity and the orbital forces of their journey around the Sun became more attracting, and the Earth grew to what would prove to be a crucial size. During this process, meteorites of all sizes crashed into and became a part of the surface, and much of the energy of such collisions was converted to heat and was retained in the growing mass, or heat sink. Beyond a certain point where size, mass, and heat reached critical dimensions, the collected heat began to generate an internal dynamics, one that continues to this day. The Earth's internal temperature reached a point comparable to that of the Sun's outer regions, and a central core developed that 4.6 billion years later is still about 20 percent hotter than the Sun's surface.
This central furnace probably melted everything, and the iron then sank, relative to lighter material such as silicates, which rose toward the surface, hardened, and became rock. This intense heat energy continues coursing outward through the 6370-kilometer radius of the planet. The Earth also has a second source of energy, the decay of radioactive materials deep within. This atomic process is also converted to heat, and most geophysicists believe this the greater of the two sources of energy powering the heat engine. Regardless of its source, however, it is the behavior of this heat that determines the dynamic fate and future of the planet.
A New Model of the Earth
Stevenson surveyed a broad picture from his position in the earth sciences community and provided a cutaway snapshot of the Earth as scientists now believe it to be (Figure 1.1). Although slightly oblate, the Earth may be viewed as a large sphere consisting of more or less concentric layers (although deformities of this regularity exist at the boundaries between the major regions and are of especial interest). The heat engine aspect of the Earth corresponds to this diagram: the closer to the center of the Earth's core, the hotter the temperature and the greater the pressure, reaching a peak of perhaps in excess of 6600°C and over 3.65 million times the atmospheric pressure found at the surface. Thus, at any specific depth, local environmental conditions facilitate and to a limiting extent determine the presence, phase state, and "movability" (or its inverse, viscosity) of particular components of the Earth's interior. Geologists taking seismological readings are most often deducing temperature and pres-
sure conditions at particular locations within the Earth. However, since all such markers are inferred rather than measured, kilometer readings of depth render a more graphic image. Stevenson reinforced the importance of the underlying regular phenomena of temperature and pressure. Together with measurements of the Earth's magnetic field, they provide baseline information for modeling that allows scientists to search for anomalies.
At the center of the Earth, 6370 kilometers from its surface, is its inner core. "The solid inner core has seismic properties consistent with those of metallic iron," said Stevenson, who added that "one suspects there is some nickel mixed in." Although it is the hottest region within the Earth, the inner core is solid due to the astounding pressures it is subjected to. Upward from the center toward its 2900-kilometer-deep outer edge, the core changes to a liquid. Slightly less pressure and—some believe—possibly higher proportions of other materials permit this change. This liquid is in motion, due to the heat convecting from beneath, and may be said to be sloshing around a million times more rapidly than material in the inner core beneath it. The motion of this material, said Stevenson, "is the seat of the dynamo action, where the Earth's magnetic field is formed.'' Thus more than half of the Earth's diameter consists of the inner and outer core. Next comes a very small region rich with phenomena and
controversy, the core-mantle boundary (CMB) between the core and the next major section above, the mantle. A number of regional structures of as much as several hundred kilometers seem to arise discontinuously near this depth and are referred to as the D-double prime (D") region. The temperature here may be perhaps 3500°C, but is quite uncertain.
Above the CMB (and—where it is found—the D") is the other major region of the Earth's interior, the mantle, which rises from a depth of 2900 kilometers through a transition zone at about 670 kilometers to its upper edge, variably between depths of about 50 to 150 kilometers. Thus the core and the mantle together make up over 99 percent of the Earth's volume. "The mantle is mostly solid, composed of materials that are primarily magnesium, silicon, and oxygen—the silicates and oxides," said Stevenson. The final of the Earth's three layers is the crust, whose thickness ranges from about 6 kilometers under the oceans to 50 kilometers under certain continental regions. Dividing the Earth into core, mantle, and crust is traditional, and these three regions are distinct in their rock chemistry.
Alternatively, geologists tend to divide the outermost part of the Earth into regions that reflect differences in the ways materials behave and how matter deforms and flows—the study of rheology. The uppermost region, the lithosphere, is rigid, is on average about 150 kilometers thick (thereby embracing the crust and part of the upper mantle), and actually slides around on the surface of the top zone of the upper mantle, known as the asthenosphere, which convects and is considered to be less viscous than the zones immediately above and below it, and may be partially melted in some places. Generally about 200 kilometers thick, the asthenosphere is the earthly "sea" on which float the lithospheric plates, atop which civilization and the Earth's oceans and its visible surface deformations perch. The lithosphere is not a solid, continuous shell but consists rather of at least a dozen identifiable plates (and more, when finer distinctions are made) that quite literally slide over the surface of the asthenosphere.
How the Model Came to Be
People have no doubt always wondered about the Earth on which they stood, but the science of geology was born from the observations and ideas of European and North American naturalists near the turn of the 19th century. When British geologist Charles Lyell (1797–1875) in his classic Principles of Geology developed what he called uniformitarianism, he struck a new course from the prevalent view
that the Earth was formed by cataclysmic events like Noah's flood, and other biblical occurrences. He suggested that fundamental ongoing geologic processes were the cause of the Earth's primary features—its mountains, valleys, rivers, and seas—processes that he believed had been occurring gradually and for much longer than was then thought to be the age of the Earth. The prevalent view that followed for nearly a century and a half, however, was that the Earth's surface was fairly rigid.
The notion that the Earth's solid surface might be mobile—" although on geologic time scales of hundreds of millions of years," said McNutt—had been a distinctly minority opinion, first proposed by biologists like Charles Darwin who could in no other way account for similarities of flora and fauna at widely separated locales. Then another science, physics, began to chip away at the theory of a rigid Earth. The development of radioactive dating methods provided a way of discerning the age of a material by its rate of atomic decay, and suddenly the assumption made by Kelvin and others that the Earth was at most 100 million years old was obliterated, and along with it mathematical calculations of the cooling rate. Rocks could now be dated and showed age variations with consistent pattern over hundreds of millions of years. Suddenly a whole new set of data was presented to theorists, and geologists now had a time line for the Earth's history that was consistent with mobilism, the theory of dramatic—albeit slow—movements within the Earth over long spans of time.
Plate tectonics, like many another revolutionary idea, has had a long genesis. As did their colleagues in biology, late-19th-century geologists found striking similarities in structure and materials throughout the Southern Hemisphere and theorized about an erstwhile super-continent, which Austrian geologist Eduard Suess named Gondwanaland. Looking at all of these clues, German geologist Alfred Wegener (1880–1930) proposed a formal theory of continental drift in 1915. Although many observed, based on evidence provided by ever better maps, that the outlines of the continents seemed to dovetail—as if they were separate pieces broken off from one original continuous land mass Wegner called Pangaea—Wegener's ideas were not embraced for decades, and he died on expedition in Greenland, in search of corroborating evidence that that island was indeed drifting away from Europe.
By the 1960s, observations of a great Mid-Atlantic ridge on the ocean floor almost precisely midway between Europe and America revealed it to be a crack in the Earth, from which spewed molten rock. A magnetic profile of the material away from the crack toward
the respective continents shows a steadily older rock, indicating that the plates on which the continents rest are moving apart at a rate of about 2 centimeters per year. Scientists now reason with much greater confidence that 130 million years ago the two continents were indeed one, before a series of volcanoes and earthquakes developed into a rift that filled with water from the one primordial sea, giving birth to distinct Atlantic and Pacific oceans. The ridge has since been recognized as the longest structure on the Earth, over 75,000 kilometers, winding from the Arctic Ocean through the Atlantic, eastward around Africa, Asia, and Australia, and up the Pacific along the West Coast of North America. "Actually a system of mid-ocean ridges," said Stevenson, it provides geologists with a major geological feature that McNutt believes deserves more exploration.
The Earth's dozen major lithospheric plates illustrate demonstrably at their boundaries their underlying movement with respect to one another (Figure 1.2). At divergent boundaries such as the Mid-Atlantic ridge, as the plates move apart in more or less opposite vectors, molten material erupts in the rift and forms a mountainous ridge, actually adding to the Earth's crust. Conversely, at convergent
boundaries such as the Marianas Trench, the plates move directly toward one another, and as they collide, one slides directly above the other, the latter undergoes subduction, and crustal material is destroyed by being recycled into the mantle. If the vector directions of the plates are neither in collision nor opposition, the plates may be said to slide past one another in what is called a transform fault. The Earth's total mass and its surface area are conserved in the plate tectonic process.
Rather than describing it as a full-blown theory, Stevenson preferred to label plate tectonics as "a description of what happens at the Earth's surface—the motion associated with continental drift and the generation of new ocean floor. It is a phenomenological description." He did not diminish its power, however. Clearly, he said, "the paradigm governing much of earth science since the 1960s is mobility. It is a picture particularly pertinent to an understanding of the ocean basins. A variety of measurements, initially paleomagnetic and subsequently geodetic, have confirmed the picture rather well."
NEW METHODS AND TOOLS FOR WORKING GEOLOGISTS
The rapid progress in modeling of the Earth in the last 15 years owes a heavy debt to growing computer power, especially in the analysis of seismic waves, which have been collected and studied without the aid of sophisticated computers since the 1920s. Closely related to this improvement have been advances in the data collection itself, with more—and more accurate—measuring instruments and seismic wave collection centers being established all over the world, and a better and more elaborate campaign of magnetic and gravitational data surveys as well. Moreover, simulations are not limited to the realm of the computer: a new process called diamond-anvil cell technology has been developed to recreate the high-temperature and high-pressure conditions thought to exist in the lower mantle and the core itself.
As suggested by Stevenson's catalog of allied disciplines, geology employs a number of distinct scientific tools and methods to model and probe the Earth's interior. One of the most valuable, barely a decade old, is seismic tomography, which exploits the sound and shear waves created each year by many global earthquakes of varying intensities. An earthquake, although difficult to predict, is not hard to rationalize. Again, it is the Earth's dynamism, specifically its property of imperfect elasticity, that gives rise to seismic waves. An elastic medium will resist deformation, such as that promoted by the moving plates. This resistance to being compressed or sheared takes
the form of a restoring force, which, when the break finally occurs because of imperfect elasticity, is released to emanate through the medium in the form of a seismic wave, i.e., a strong, low-frequency sound wave. These waves take several different forms, but each has a recognizable signature and behaves with distinguishing characteristics according to the principles of physics. Surface waves, called Love and Rayleigh waves, travel circular paths along the Earth's surface and extend deep enough into the Earth to interact with the upper mantle. So-called body waves forge curved paths through the Earth's depth and deliver information from their journey all the way down to the core.
Seismic tomography employs heavy computer power in a process closely analogous to computerized tomography in medicine. Instead of undergoing variable absorption, as do x rays in the body, seismic waves confer information by altering their speed when they encounter differing materials. As with computerized tomography, the results provide a three-dimensional picture of the waves' journey from their source at the epicenter of a quake to the many measuring devices spread all over the globe. Since the journey's distance can be computed precisely, any delay in a wave's arrival at a given site can be attributed to the nature and condition of the media through which it just passed. A given wave may penetrate thousands of kilometers of the Earth on its journey and arrive with but one piece of data, its travel time, which yields only an average for the trip, not data indicating how the wave may have slowed down and speeded up as it encountered different conditions along its path. Thousands of such readings all over the planet for similar waves traveling along different, cross-cutting paths make it possible to perform extremely sophisticated analyses of the data, since each separate measurement can provide a restraint on the generality of each of the others. Elaborate mathematical analyses combine all of these data into hitherto unglimpsed, three-dimensional views of the inner Earth, all the way down to the center.
Seismic tomography exploits the fact that a medium's physical properties—its density, composition, mineral structure, mobility, and the degree of melt it contains—will determine how fast waves of a particular type are transmitted through it. The liquid material in the Earth's outer core does not transmit shear waves but does transmit compressional waves. Properties such as a medium's rigidity and its compressibility are in turn affected by its temperature and its pressure. Knowing the type of wave being measured and the distance it travels, seismologists can hypothesize based on what they believe they know about temperature, pressure, mantle convection, and the
composition of the Earth at different depths. When the tomographic readings vary from the prediction, anomalies can be inferred and theories developed to explain them that nonetheless fit all of the constraining data. Colder material tends to be stiffer and a faster transmitter of waves than hotter material.
Another property providing seismologists with valuable data is how a material's mineral crystals are aligned. Flowing material in the mantle tends to orient the crystals in nearby rock in the direction of the flow. Crystals have three different axes that affect the speed of seismic wave propagation differently. If randomly aligned, the crystals produce an average wave transmission speed, compared to a faster speed if a crystal's fast axis is parallel to a given wave's (known) direction of propagation. The complexity of this picture does not yield perfectly transparent readings, but the power of the computer to filter the data through a web of interrelated equations provides seismologists with a powerful modeling tool.
Geological models of the deep Earth must incorporate the lessons of solid-state physics, said Stevenson, which indicate that most materials will behave differently under the temperature and pressure extremes thought to prevail there. Some of these assumptions can be tested with a new technology that promises to open fertile ground for theorizing. A diamond-anvil cell consists of two carefully cut and polished diamonds mounted with their faces opposite each other in a precision instrument capable of compressing them together to produce extremely high pressures, creating an "anvil" with a surface less than a fraction of a millimeter across on which samples can be squeezed (Figure 1.3). Hemley, who uses a variety of such devices at the Carnegie Institution's Geophysical Laboratory in Washington, D.C., explained that the transparent nature of diamonds allows the experimenter to further probe the sample with intense laser and x-ray beams. Lasers can also be used to increase the temperature. Together, these techniques can simulate temperatures approaching those found at the surface of the Sun and pressures over 3 million times that at the Earth's surface, up to and including the pressures at the center of the Earth.
"Under these conditions, planetary constituents undergo major structural and electronic changes," explained Stevenson. The Earth's most common mineral is probably magnesium silicate perovskite, a denser structure type than most other constituents and one that is not thermodynamically stable for silicates at normal atmospheric pressure. The normal olivine structure—in which a silicon atom is surrounded by four oxygen atoms—has been shown by diamond-anvil cell experiments to undergo a series of transformations, ultimately
taking on the perovskite structure, in which the silicon is surrounded by two additional oxygen atoms. This occurs at just about the pressures that have been determined by seismology to prevail at the 670-kilometer boundary region between the upper and lower mantle (Knittle and Jeanloz, 1987). Said Stevenson, ''The reduction of bandgaps and preference for compact phases may greatly change our view of the differences between and interrelationships among silicates, oxides, and metallic oxides."
Hemley believes that discovery of these phase transitions and measurements of properties of the perovskite minerals favor the mantle convection models that employ more complicated patterns than the single, deep-convection theory. "But," he cautioned, "more measurements are required to be certain." Hemley's laboratory and others continue to probe the atomic behavior of materials under conditions more like the extreme ones found deep inside the Earth. One such material is hydrogen, interest in which, said Hemley, ''dates back to the 1935 calculation by Wigner and Huntington that showed that a monotonic lattice of hydrogen atoms will be a high-density metal. Hundreds, if not thousands, of papers have been written on this topic in the intervening 55 years." Hemley explained that until recently,
direct tests were not possible, and so he and his colleague H.K. Mao wanted to test their prediction about where "the transition from the low-density, molecular, insulating phase to this metallic phase" would occur.
Although exploring the metallic phase of hydrogen might seem to be a concern primarily of condensed-matter physics, this question has great relevance to planetary geology as well and illustrates, said Hemley, the interdisciplinary nature of this work. Jupiter and Saturn are composed largely of hydrogen and are believed to contain metallic hydrogens because of the enormous pressures reached inside. "Understanding the metallization phenomenon is therefore crucial for modeling their dynamics," explained Hemley. So far, their results indicate an intriguing series of phase transitions rather than simple conversion of an insulating molecular solid to an atomic metal. These new phenomena are challenging them to invent new variations on the basic diamond cell techniques and are driving the development of new theories in condensed-matter and planetary sciences.
Gravity provides another lens into the Earth. Since Newton's time, the understanding of gravitational forces has provided basic constraints as well as a realm of useful data for scientists in many fields, for geologists perhaps more than others. Scientists refer to the geoid as the equipotential surface normal to the Earth's gravity. "A buried mass excess," explained McNutt, "will raise the geopotential surface, that is, cause it to move further from the Earth's center. Conversely, a mass deficit will lower the equipotential."
"Geodesy remains a powerful method," said Jeanloz, and together with seismology offers a powerful way of looking inside the Earth, notwithstanding the recent applications of tomography. All of these methods provide excellent and often distinct supplementary perspectives to corroborate or constrain theories based on data gathered with yet other methods. In a science for which the deepest mine penetrates barely 2 miles and the deepest hole less than 10, inferences assume greater importance because most measurements are necessarily indirect. And while geophysics is providing richer data with each year of technological progress, geologists searching for the big picture must bring together a wealth of distinct, sometimes contradictory perspectives. Theorizing about how the Earth works is inviting and yet risky. But the ferment and excitement surrounding the endeavor are palpable, as the Frontiers session on geology illustrated.
SHARPENING THE FOCUS ON THE EARTH'S DYNAMICS
A central problem for earth scientists, said Stevenson, is to reconcile plate tectonics and mantle convection. Both phenomena occur,
but exactly how convective currents inside the mantle may produce the motions of the plates has not been fully clarified. Because the Earth's internal heat source has been generally verified, he continued, our knowledge of the physical principles of convection makes this explanation appealing on the theoretic level. Undoubtedly the mantle convects; the question is what patterns the convection follows. The question is by no means trivial, for if there is one large convective pattern circling throughout the mantle—the deep-convection model—the fate of the Earth and its internal heat will run differently than if there are at least two layers more or less independently convecting. On the other hand, while plate tectonics has been demonstrated by myriad data, measurements, observations, and chemical analysis, a full theoretical explanation has yet to be offered. Geologists have not been able to come up with a comprehensive model that incorporates either mantle hypothesis with plate tectonics.
As Stevenson put it, "The biggest problem confronting a synthesis of plate tectonics and mantle convection lies in the inability to incorporate the extremes of rheology," the study of the deformation and movement of matter. A problematic element for the mantle convection models is the quasi-rigid nature of the plates. Those models posit that temperature has a great effect on viscosity, but that would lead, he said, "to a planet completely covered with a surficial plate that has no dynamics at all," despite the convection beneath the surface, instead of the planet we actually observe. Thus the appealing theoretical success of mantle convection does not begin to explain much of the data. Stevenson called it ''a profound puzzle," and concluded, "We simply do not understand why the Earth has plate tectonics.''
Yet the plate tectonics model also embodies a dilemma: not only is the cause of plate tectonics elusive, but conventional models also do not even fully describe the phenomenon. As Stevenson said: "An interesting question is the extent to which plate tectonics is even correct. That is to say, are those plates actually rigid entities rotating on the surface of the Earth about some pole? The answer," he continued, "is yes, to perhaps a 90 percent accuracy. You might say that that is quite good, but that 10 percent failure is important because it leads to the formation of many of the Earth's mountain ranges and other prominent geological features." Scientists modeling plate tectonics treat the plates as rigid and develop predictions about how they move over the Earth's surface. The 10 percent failure to predict what is observed points to "deformation primarily at plate boundaries," said Stevenson, and explains why so much attention is paid to regions of the world where the plates are colliding. Another phe-
nomenon that may provide a link between convective and plate tectonic theories is the world's 40 or so active hot spots.
Hot Spots, French Polynesia, and the "Superswell"
Hot spots present scientists with what seems to be another main ingredient in the Earth's recipe for its simmering equilibrium. These plumes emanate from deep within the mantle (or some from not so deep) and consist of slender columns, each about 300 kilometers in diameter, of slowly rising rock hotter than their surroundings (Figure 1.4). Although the temperature differential may be as small as 100°C, its impact on viscosity and therefore on the rise of the convective plume can be dramatic. Most hot spots are believed to begin as deep-mantle phenomena and provide a benchmark for plate movement because of their small lateral motion. Since the lithospheric plates move across and above the asthenospheric point(s) where a plume emerges from its long journey through the mantle, they get marked by a trail of volcanoes that leave a permanent record of the direction and speed of the plate itself.
The first coherent theory of hot spots was proposed by the University of Toronto's J. Tuzo Wilson in 1963 after he studied the Hawaiian volcanic fields. In the intervening years, the plume model has been elaborated and now may provide an explanation of how convection delivers heat to the surface, although its most recent proponents hasten to admit that no plumes have been observed directly. Vink et al. (1985) have explained that "ultimately the energy that
drives plate motion is the heat released . . . deep in the mantle. The plumes provide an efficient way of channeling the heat toward the surface. . . . Less viscous material produced by variations in temperature or volatile content tends to collect and rise toward the surface through a few narrow conduits, much as oil in an underground reservoir rises through a few bore-holes" (p. 50–51). Their theory does not contend that the plumes actually propel the plates, but rather that "the two are part of the same convective cycle" (p. 51). The asthenosphere must be heated in order to provide a less viscous surface for the plates to slide across, and without the plumes to provide this heat the plates would grind to a halt.
Certain regions of the Earth's surface are pregnant with these hot-spot eruptions from the mantle beneath. One such region is the seafloor beneath French Polynesia in the South Pacific, explained McNutt, whose career has been devoted to looking more closely at the hot spot phenomenon there. That region differs from the standard oceanic lithospheric profile "in almost all respects. Its depth is too shallow, its flexural strength too weak, the velocity of seismic waves through it too slow, and geoid anomaly above it too negative for what has been determined to be its lithospheric age," she pointed out. Together with MIT co-worker Karen M. Fischer, McNutt named this geologic curiosity the South Pacific "Superswell."
Theorists wonder if this area might yield clues to a unified theory. By positing a thinner plate in this region, McNutt can account quantitatively for the anomalous seafloor depths observed and qualitatively for the low velocities of seismic surface waves, the weak elastic strength of the lithosphere, and the region's high vulnerability to what is distinguished as hot-spot volcanism. The thinner plate hypothesis does not, however, account for other observed features, such as why the plates move faster and spread faster over the mantle in the Superswell region than anywhere else on the Earth. "And most importantly," emphasized McNutt, "a thinner plate in the absence of other explanatory features would produce a geoid high," not the low that is actually observed.
Polynesia presents another mystery to earth scientists. Seismic studies confirm that these volcanoes are of the hot-spot variety. Although constituting only 3 percent of the planet's surface, the region accounts for 30 percent of the hot-spot material—the flux of magma—that appears all over the Earth. But why there, and why now? The answer, or at least another important piece of the puzzle, may lie thousands of miles to the northwest in an area of ocean floor southeast of Japan known as the Darwin Rise. More hot-spot volcanoes are clustered there, although these are now extinct and lie beneath
the surface of the sea. What is compelling about these two fields of volcanism—one active, the other extinct—is their location. The Darwin Rise tests out at an age of 100 million years. McNutt was able to "backtrack the movement of the Pacific plate in the hot-spot reference frame to its position at 100 million years ago," and she demonstrated that when the hot-spot volcanoes of the Darwin Rise were created—when material welling up from within the Earth formed a ridge with active volcanoes—it was located above the mantle where the French Polynesian volcanoes are now erupting. She argued from these data that hot spots must be driven by processes within that part of the mantle and, further, that they occur episodically.
To account for these observations, McNutt and co-workers at MIT have developed a theory. Seismic wave data show that the lithosphere beneath the Superswell has been thinned to about 75 kilometers, 40 percent less than the usual lithospheric thickness under the North Atlantic and North Pacific oceans. This thinner plate, she believes, is not the cause of the Superswell; rather it is the surficial manifestation of forces below, specifically a low-viscosity area under the plate and a plume of heat convecting through the mantle. To simplify somewhat, the MIT team has inferred from the data that a large hot blob of material periodically rises through the mantle from near the CMB in the Superswell region. Once the blob arrives near the upper reaches of the mantle, one would expect the geoid above to reflect it. Since the data show a depression—not a rise—in the geoid, something else must be happening. McNutt has inferred the presence of a low-viscosity zone near the upper reaches of the mantle. Such a phenomenon would account for the geoid anomaly and further might explain why the lithosphere in the Superswell region moves so rapidly.
Is Volcanism an Essential Part of the Cycle?
Hot-spot volcanoes may well be an indigenous phenomenon, pointing to something unique in the Earth below them. But volcanoes erupt all over the Earth, and Stevenson believes it important to understand the role volcanism plays in the interaction between the tectonic plates and the mantle. He said that thermal convection theories alone are not rich enough to explain all the complexities of the Earth's dynamics. "It seems unlikely that we can claim any deep understanding of how the Earth works if we confine ourselves to the behavior of a single working fluid of very high viscosity, and restrict ourselves to density differences that arise only from the small effects of thermal expansion," he explained. While these considerations are not fully
accepted tenets in earth science, they have deflected inquiry away from another phenomenon Stevenson thinks suggestive: differentiation of the Earth's constituents and thus the possibility of two-phase flow. In theory, convection could be occurring up through the Earth's profile with no actual melting into liquid form. It is known that partial melting is consistent with the fact that some of the components melt at lower temperatures than do the dominant constituent. From such melting—at least in localized regions—result significant changes in fractional density, up to as much as 10 percent.
Now the deductions start to line up: changes in density from melting or partial melting produce significant changes in viscosity; the melting process can be enhanced by water or other volatiles; the result, said Stevenson, echoing the plume model, may be a kind of "lubrication of plate tectonics." That is, the presence of partially melted lower-viscosity material at the boundary may effectively "de-couple the mantle from overlying plates."
So, according to this hypothesis, partial melting helps lubricate the sliding plates, which at moments and points of impact may spew volcanoes. What, then, would happen if there were insufficient heat in the Earth to accomplish this melting? If the temperature of the mantle were reduced by only 5 percent, said Stevenson, volcanism would largely cease: "If that occurred, what then would the Earth do? The Earth is a heat engine. It could certainly keep convecting, but could it have plate tectonics?" He reported that many of his colleagues suspect "that the melting process, the formation of basalt, may be an important factor keeping the plate tectonics system going and may be a part of the difference between a planet that has a single plate completely encompassing it and a planet, like the Earth, which is tessellated into these many plates of many different sizes." For Stevenson, therefore, volcanism is essential: "It is not a sideshow. To understand mantle convection, you have to take into account the fact that material melts to some extent.''
Dynamics in the Core and the Magnetic Field
Looking deeper yet, beneath the mantle, scientists find that the material in the outer core is highly conductive and low in viscosity. The forces of convection moving through this region have somehow set up a continuously regenerating dynamo, a massive natural version of the man-made power systems up above that produce predictable magnetic fields. Exactly how this dynamo works is not known, although a semiquantitative picture has developed over the past few decades. Stevenson pointed out that scientists attempting to under-
stand the Earth's magnetic field have one advantage over those trying to understand the interaction between the mantle and tectonic plates: the material parameters of the core may be simpler and less variable. Nevertheless, Stevenson emphasized, "This is a difficult problem. People are tackling it now on supercomputers and by approximation techniques."
The magnetic field profile of the Earth shows a familiar dipole. The geomagnetic field actually observed, however, contains a number of other, more complex components. Often in geology scientists look to the edges, the boundaries, for their cues, and it seems likely that some of the magnetic field anomalies are related to what is happening at the CMB. "There must be a bumpy and fuzzy interface between the core and the mantle," Stevenson said, rather than "the fluid dynamicist's ideal world of a smooth, isothermal, equipotential yet rigid bounding surface." In fact, the terrain at the bottom of the mantle known as the D"—a region, where it occurs, of a few hundred kilometers with characteristic, perhaps even continental-scale features—could be comparable in general shape to the topography seen at the surface. If this region is truly rich with such features, Stevenson continued, it follows that heat flow between various regions of the mantle and the core would be nonuniform. And the effects would not stop there.
These features at the interface most likely have a significant influence on heat flow in the outer core, perhaps on the heat flow that generates the dynamo itself. This raises the exciting but complex prospect of relating the observed field geometry and how it has varied throughout a couple of hundred years of recorded history to the distribution of irregularities in the lowermost mantle. For example, take a region of mantle that has substantial downflow because it is colder than average. What can be inferred about the CMB in this local region? Stevenson believes that the core probably contains "a very-low-viscosity fluid that may vigorously convect on a small-length scale . . . in the same way as air shimmers in the desert in the boundary layer region between the hotter Earth and the cooler atmosphere." This border region draws contending theories as honey does bees, but few as controversial as that of Jeanloz.
Jeanloz's work in probing the chemical reactions at the CMB has been provocative but has enriched the debate considerably. Along with Knittle, Jeanloz believes that there is considerable chemical interaction between the silicates that probably make up the lower mantle and the iron of the outer core. As Hemley and the diamond-anvil cell experiments have shown, these elements remain chemically aloof at the surface, but under conditions of extreme pressure and temper
ature they intermix. Thus, Jeanloz and Knittle have reasoned, silicon or oxygen alloys from the lower mantle could actually become a part of the core's chemistry and thereby reduce its density (Knittle and Jeanloz, 1991), a finding corroborated by other independent measurements. Thorne Lay has provided a powerful set of seismic measurements that seem to corroborate material variations.
Stevenson cited recent analyses of data in ships' logs and marine almanacs suggesting that for almost 300 years the basic features of the geomagnetic field have remained fixed relative to the mantle, "indicating at least partial control of the field by mantle irregularities." That work has been conducted by another session participant, Jeremy Bloxham of the Department of Earth Sciences at Harvard University. Rather than taking a top-down, mantle-oriented approach, Bloxham has looked at the same complicated magnetic field inferred to exist at the CMB for implications about the dynamics of the deeper core. Satellite measurements of the geomagnetic field taken in 1969 and again in 1980 showed small changes. Plugging these data into the equations that describe the Earth's magnetic field allows scientists to construct a map of the fluid flow just beneath the surface of the core. Bloxham hastened to warn, however, what a small pedestal these observations provide for scientists to try to view the Earth's history, since the velocities of the liquid iron that generates these fields are on the order of 20 kilometers per year (10-3 m/s): "If we draw an analogy with meteorology where wind speeds in the atmosphere are on the order of meters per second, then—in a very crude sense—we can say that looking at the atmosphere for a day yields the same insight as looking at the core for 60 years. In other words, we need 60 years worth of observations of the core to get the equivalent information that we could obtain observing the atmosphere for a single day." Conversely, theories based on the patterns of flow in the core derived from this 11-year data slice might be compared, Bloxham said, to "trying to understand the dynamics of the atmosphere based on just a few hours of data."
The good news is that flow patterns are observed to change only slowly as one penetrates deeper into the core. Thus the picture drawn from the satellite data may well correspond to the full profile of the outer core, said Bloxham, who has also found corroboration from a surprising source. Field researchers for the past 300 years have been making reliable measurements, and some data obtained as long ago as the late 17th century can now be reinterpreted through modern theory. Bloxham said that these data suggest several fairly constant features of the geomagnetic field. For one, the field at the CMB beneath the North Pole is consistently close to zero, a feature incon-
sistent with a simple dipole model of the magnetism in which the field should produce maximum radial flux at the poles. Also, certain concentrations of high flux at the Earth's high latitudes have remained fairly constant over time.
Theory and experiment indicate that a rapidly rotating fluid forms convection columns. These columns align parallel to the axis of rotation. Bloxham said that one interpretation of the magnetic field data suggests that such a process is occurring in the outer core, with large convection rolls aligned parallel to the Earth's rotation axis then grazing the solid inner core. The observed, high-latitude, surface concentrations of magnetic flux would, according to this interpretation, correspond to the tops and bottoms of these convection rolls, while the regions of almost zero flux at the poles also fit the pattern predicted from such an effect. Bloxham has become one of the authorities on the core, and his view is that no significant layering occurs there.
Evidence in the Fossil Record
Earth scientists face another dilemma. Despite dramatic advances in the arsenal of scientific measuring techniques, data on the Earth's dynamics can only measure what is happening today. Yet the Earth has been evolving for billions of years. As session participant Michael Gurnis put it, "Current measures of the planet, as powerful as they are, are weak in terms of understanding the true dynamics of the Earth." Geologists have known for many years that mantle changes can move continents and cause fluctuations in sea level over time. This ebb and flow leaves a pattern in the rocks examined on continents because, when the shore at a continent's margin is submerged, characteristic sediments accumulate and harden into rock. Over long periods of geologic time, such rocks record this history of ocean flooding. Gurnis pointed out that "geologists have recognized for the past century that these rock formations are measuring changes in the Earth's shape," and that the sedimentary rock patterns on the offshore continental shelves provide a four-dimensional map of the movement of the ocean's edge back through time.
Continental drift does not seem to be the sole factor defining the ocean's edge. Continental crust, composed of material like granite, has a lower density than the ocean floor crust, which is rich in basalt, and so it floats higher on the mantle than do other regions. Thus arises a different approach to the earlier question: How does convection in the mantle work together with tectonic plate movement? Theoretical and modeling work on this question suffer, said Gurnis, from our "poor ability to simulate on a computer the interaction between
tectonic plates and mantle convection." Assuming simple hydrodynamic convection through a mantle modeled with the appropriate characteristics and physical properties, he and others try to predict what would happen to a single continental plate above. Gurnis said that the model demonstrates patterns of continental flooding that leave a characteristic signature, which then correlates back very well to the internal dynamics of the model system. This effort points to the possible value of the historical record of sea level changes, a potential database on the order of 500 million years old. There may yet come a computer simulation of the planet's interior that will crack the time barrier and write a true geological history of the whole Earth, inside and out.
Geology of the Other Planets
Stevenson began his presentation with the remark that "among the solid planets, the Earth is unusually dynamic." Can this observation be used to further the inquiry into the dynamics themselves? Stevenson's speculations have been fortified by spacecraft providing often dramatic and penetrating views of other planets in our solar system. A series of probes to Mars has revealed many features. The present Magellan mission to Venus is providing a high-resolution radar map of the surface of the Earth's nearest, cloud-shrouded neighbor. And the two Voyager spacecraft have yielded a wealth of data on more distant solar system neighbors—Jupiter, Saturn, Uranus, and Neptune. Bloxham reported that "until 2 or 3 years ago, I think a discussion like this would have included the belief that all magnetic fields we are aware of align with the rotation axes of their planets. This was confirmed with the Sun, Earth, Jupiter, and Saturn." But then "along came information on Uranus and Neptune, which have fields inclined at some 50 degrees to their rotation axes. That threw a small monkey wrench into the works," he pointed out.
But information from the near and far solar system has not yet supplied inferences that solve the mysteries of convection. Scientists know, for example, that all planets, large satellites, and moons convect, but only the Earth among the solid bodies observed appears to undergo the striking characteristics of plate tectonics. Venus might possibly have some plate-tectonic-like features, and the controversy over this question awaits further data from the Magellan mission. Nevertheless, said Stevenson, sliding tectonic plates have been confirmed nowhere else and are not yet fully understood for this planet. "Mars is a one-plate planet," said Stevenson, and "actually even Mercury and the Moon are believed to have mantle convection," although, at present, "it is much less vigorous."
He conceded it is a reflection of our green understanding that despite "vast terrestrial experience" we are not far along the path of fully describing other planets. Far from ripe and seasoned, then, planetology is a new and still profoundly mysterious science. And though he and most geologists may focus more on our own planet, Stevenson believes that developing a generic awareness of the other known planets will enhance and enrich the context of observations made on the Earth.
Anderson, Don L., and Adam M. Dziewonski. 1984. Seismic tomography. Scientific American 251(October):60–68.
Bloxham, Jeremy, and David Gubbins. 1989. The evolution of the Earth's magnetic field. Scientific American 261(December):68–75.
Glatzmaier, Gary A., Gerald Schubert, and Dave Bercovici. 1990. Chaotic, subduction-like downflows in a spherical model of convection in the Earth's mantle. Nature 347(6290):274–277.
Jeanloz, Raymond. 1990. The nature of the Earth's core. Annual Review of Earth and Planetary Sciences 18:357–386.
Knittle, Elise, and Raymond Jeanloz. 1987. Synthesis and equation of state of (Mg,Fe) SiO3 perovskite to over 100 GPa. Science 235:668–670.
Knittle, Elise, and Raymond Jeanloz. 1991. Earth's core-mantle boundary: Results of experiments at high pressures and temperatures. Science 251:1438–1443.
McNutt, Marcia K., and Anne V. Judge. 1990. The superswell and mantle dynamics beneath the South Pacific. Science 248:969–975.
McNutt, M.K., E.L. Winterer, W.W. Sager, J.H. Natland, and G. Ito. 1990. The Darwin Rise: A cretaceous superswell? Geophysical Research Letters 17(8):1101–1104.
Powell, Corey S. 1991. Peering inward. Scientific American 264(June):101–111.
Press, Frank, and Raymond Siever. 1978. Earth. Second edition. Freeman, San Francisco.
Scientific American. 1983. The Dynamic Earth. Special issue. Volume 249 (September).
Stein, Ross S., and Robert S. Yeats. 1989. Hidden earthquakes. Scientific American 260(June):48–57.
Vink, Gregory E., W. Jason Morgan, and Peter R. Vogt. 1985. The Earth's hot spots. Scientific American 252(April):50–57.
Anderson, D.J. 1989. Theory of the Earth. Oxford, New York.
Fowler, C.M.R. 1990. The Solid Earth. Cambridge University Press, New York.
Hemley, R.J., and R.E. Cohen. 1992. Silicate perovskite. Annual Review of Earth and Planetary Sciences 20:553–593.
Jacobs, J.A. 1992. Deep Interior of the Earth. Chapman & Hall, New York.
Jeanloz, R. 1989. Physical chemistry at ultrahigh pressures and temperatures. Annual Review of Physical Chemistry 40:237–259.
Lay, Thorne, Thomas J. Ahrens, Peter Olson, Joseph Smythe, and David Loper. 1990. Studies of the Earth's deep interior: Goals and trends. Physics Today 63(10):44–52.