mantle convection, plumes, viscosity, and dynamics

mantle convection, plumes, viscosity, and dynamics The theory of plate tectonics drastically changed our understanding of the dynamics of the Earth's interior. Until the theory of plate tectonics was developed, the primary means of transporting heat from the Earth's interior to the surface was thought to be by conduction. Rock is an extremely poor conductor of heat, and the observed heat flow at the Earth's surface was therefore attributed either to conductive heat flow from the near-surface or to radiogenic sources. Conduction, unlike convection, requires no movement of material, and so the dynamics of the deep interior of the Earth would then have little effect on the observed surface processes and was of only academic interest.

In plate tectonics, large rigid plates float on a fluid plastic mantle. These plates move about the surface of the Earth at speeds of up to 10 cm per year. The motion of the plates has to be driven in some way. Furthermore, if mantle rocks are able to flow, then the transport of heat can be achieved more efficiently by convection. Heat can be transported to the Earth's surface from deep inside the Earth. This leads to the concept of a dynamic mantle composed of ‘solid’ rock that can behave as a viscous fluid over geological timescales. On a geological timescale, lithospheric plates behave as elastic solids and the mantle as a viscous fluid. If the motion of the plates is controlled by mantle dynamics, the Earth's deep interior holds the key to understanding the evolution of much of the Earth's surface.

The viscosity of the mantle

The mantle is known to flow like a fluid over geological timescales. We quantify the ability of a material to flow under pressure by its viscosity. The viscosity is defined as the ratio of shear stress to strain rate, and is thus a measure of the internal friction of a fluid: the more viscous a fluid, the more resistant it is to flow. Measuring mantle viscosity is not a trivial problem. Direct measurements are not possible, and the viscosity must therefore be inferred from surface processes.

Geologists in Finland and Scandinavia noticed that the sea-level appeared to be falling in Scandinavia: dating terraced beaches found in Scandinavia and Finland showed that during the past 10 000 years the sea level had dropped while the global trend was for a rise in sea level. This was due to the rapid removal of a large mass (the retreating Fennoscandian ice sheet which covered the region during the last Ice Age), resulting in the lightened lithospheric plate rising. This process, known as isostatic rebound, provides evidence for a flowing viscous mantle.

The rate of isostatic rebound enables the viscosity of the mantle to be estimated. Relatively small loads such as the Fennoscandian ice sheet yield estimates for the viscosity of the uppermost mantle. The rapid removal of much larger loads is required to estimate the viscosity of the lowermost mantle. Using these observations, the viscosity of the mantle is estimated to be reasonably constant at between 10²⁰–10²¹ Pascal seconds (Pa s) in the upper mantle and 10²¹–10²³ Pa s in the lower mantle (compared to the viscosity of water which is 10⁻³Pa s). Although the viscosity of the mantle is high, it is still low enough for the mantle to flow over geological timescales. Using these viscosities, it is estimated that the mantle can overturn completely by convection once every 100 million years.

Studying mantle dynamics

Seismologists have succeeded in describing the deep structure of the Earth in some detail. We know that the lithospheric plates are on average 20 km thick in oceanic areas and 80 km thick in continental areas. The mantle extends from the base of the lithosphere to the D layer (pronounced ‘dee double prime’), which acts as a narrow (about 200 km) buffer between the mantle and the core, at a depth of 2700 km. On a global scale the mantle can be thought of as having two distinct regions: the upper mantle contains many scatterers of seismic energy; the lower mantle is relatively homogeneous, with little scattering or anomalous refractors of seismic energy. The upper mantle and the lower mantle are separated by a weak seismic discontinuity (a jump in seismic velocity) at a depth of about 670 km. This weak seismic boundary also marks the lower boundary of observed seismic activity: no earthquakes have been recorded below this depth.

There are several weak seismic discontinuities in the mantle, and the nature of these discontinuities provides useful constraints on its dynamics. Seismic discontinuities may be a result either of compositional changes, where the main mineral constituents differ across the boundary, or of phase changes, where the composition stays the same but as a result of changes in temperature or pressure the crystallographic structure, and hence the physical properties, of the minerals change. The 670-km discontinuity is thought to be due to a phase change in the mainly iron and magnesium silicates that make up the mantle. There is a ‘gradient’ associated with a phase change that acts as a physical barrier to the flow of material across the boundary, which is similar to the contact between oil and water.

Seismic tomography provides a means of imaging the mantle using differences in seismic velocity or attenuation. These seismic properties can be related to more fundamental material properties such as temperature and density. The images produced by seismic tomography have become increasingly detailed, and it is now believed that descending slabs and even mantle plumes can be imaged with some confidence.

Understanding mantle dynamics requires detailed knowledge about the structure and composition of the Earth's deep interior. Unfortunately, seismology alone is not sufficient to understand mantle dynamics. Seismology tells us only about physical properties such as seismic velocity, attenuation, and, more indirectly, temperature. In order to understand how these properties relate to mantle dynamics we need to know something about the composition of the mantle. Mineral physics and laboratory experiments at high temperatures or pressures provide useful constraints on the probable structure of mantle minerals. The composition of these minerals is known from a handful of outcrops containing inclusions of mantle material. From geochemical studies of the radioactive isotopes that are present in volatiles released from volcanic eruptions the evolutionary history of the mantle can be pieced together.

Mantle convection

There is a strong correlation between the distribution of seismicity and mantle dynamics. Figure 1 shows a map of global seismicity. This pattern of seismicity, which can be interpreted as a map of dynamic surface features, owes a lot to mantle dynamics. Mid-ocean ridges bisect many of the world's oceans; the geography of the plates is intrinsically effected by mantle plumes; plate motions (and hence collisions) are controlled by mantle convection. All these processes depend intrinsically on mantle dynamics.

Oceanic crust is created at mid-ocean ridges: huge volumes of lava are extruded, which subsequently cool and thicken to form the oceanic crust. Remote diving vessels have observed ‘black smokers’: vents belching out the volatile components of the lava. Geochemical analysis of radioactive isotopes released by these vents and deposited at mid-ocean ridges can reveal the evolutionary history of the magma source. We have a fair idea of the relative abundance of radioactive isotopes such as helium-4 (⁴He) when the Earth was formed. We also know their half-lives and can thus determine their relative abundances at the present day, assuming that the isotopes have not been depleted in some way.

A magma source will first lose its volatile content by degassing. If the measured ³He/⁴He ratios are close to predicted present-day relative abundances, this indicates that the magma source has never before been tapped. The ³He/⁴He ratios measured at mid-ocean ridges are depleted, which is indicative of previous degassing of the magma source. On the other hand, oceanic island basalts deposited by volcanoes associated with mantle hot-spots show different ³He/⁴He ratios, which are in line with those predicted for primordial concentrations. This is indicative of an unmixed source that has not undergone degassing. The volatiles in the hot-spot source have not ‘partitioned out’ or been mixed with depleted mantle material.

This evidence suggests that mid-ocean ridge basalts and ocean island basalts are supplied from chemically distinct regions of the mantle. This observation, together with the different observed seismic characteristics of the upper and lower mantle, led many scientists to suggest that mantle convection occurred in two separate layers with little movement of material between the layers. The 670-km discontinuity was thought to provide a barrier to movement across the boundary, effectively splitting the mantle into two separate convection regimes. Other evidence includes laboratory work on convection. In the preferred convection patterns conforming to the boundary conditions present in the mantle the aspect ratio (the ratio of cell depth to cell width) of convection cells is about one. For two-layer convection, the cells would be about 700 km by 700 km in size, which fits neatly with the spatial distribution of surface features, such as mid-ocean ridges and subduction zones that are thought to be associated with downwelling or upwelling mantle material (see Fig. 1). For layered convection the subducting slabs must settle at 670 km, the accepted boundary between two layers of convection cells.

Evidence for whole-mantle convection includes calculations of heat flow. Without complete mixing of the mantle the Earth would simply overheat. Also, it is argued that plumes originating deep in the mantle would not be the long-lasting stationary features that we observe if layered convection were the primary method of heat transport through the mantle.

Evidence from seismic tomography experiments shows slabs of subducted lithospheric plates descending through the 670-km discontinuity deep into the lower mantle. In addition, the latest attempts to image plumes show plumes emanating from deep in the lower mantle, possibly from the D layer. This suggests that the mantle convects as a whole, although the geochemists now require an explanation for the existence of pockets of unmixed mantle material. Perhaps the key to a more complete description of mantle dynamics lies with understanding the mechanics of mantle plumes.

Mantle plumes

Mantle plumes are perhaps the most interesting features in the mantle. They originate deep in the Earth and yet have a dramatic influence on surface features. Elucidating the origin of plumes and understanding why they are such long-lasting stationary features in the mantle will hold the key to a more complete understanding of mantle dynamics.

Hot-spots are the result of plumes of molten rock rising through the mantle and impacting on the lithosphere, where in effect they burn a hole through the lithosphere and form volcanic islands. These islands are large topographic features: to rise above sea level in the Pacific Ocean, Hawaii must be at least 4 km high. The plumes show considerable longevity and are thought to be stationary over their life cycles. Evidence for this can be seen in the chains of volcanic islands associated with a single hot-spot. The trail of extinct volcanic islands shows the speed and direction in which the overlying lithospheric plate has travelled. The largest of these island chains is the Emperor chain culminating in the volcanic island cluster of Hawaii. This process is ongoing and today a new volcano can be seen off the main flank of Hawaii. In about 10 000 years' time this will break surface and form a new volcanic island, extending the chain.

Evidence for much larger mantle plumes can be seen in India. The Deccan traps are a series of lava beds hundreds of metres thick that were once thought to have been deposited over many millions of years. However, palaeomagnetic measurements tell a different story. The direction of the geomagnetic field is frozen into the lavas as they cool below the Curie temperature. The Earth's geomagnetic field changes with time and is known to reverse approximately once every million years. When the directions of the palaeomagnetic fields frozen into the lavas of the Deccan traps were measured, instead of a record of many geomagnetic reversals only two different palaeomagnetic directions were observed. Given the frequency of geomagnetic reversals this meant that the Deccan traps were deposited in only about a million years. This is an incredible volume of lava to be deposited in such a short space of geological time; it would be enough to cover the entire United States to a depth of almost 1 km.

The plume located underneath Iceland almost certainly resulted in the break-up of the supercontinent Pangea into what are now Western Europe and North America. The match between the coastlines of the Americas and Europe and Africa is almost perfect, except for the island of Iceland. This is because Iceland, the world's largest volcanic island, is located on top of the super-plume that resulted in the break-up of Pangea. Iceland is a segment of mid-ocean ridge domed up above sea level by the plume.

Large-scale eruptions such as those that resulted in the Deccan traps and Iceland are one end of a spectrum. The other end is represented by small island chains or isolated hot-spots, which indicate continuous sources of magma from deep in the mantle: a pulsating plume, rather than one large plume head. Super-plumes can obviously rapidly alter the Earth's surface. For example, the timing of the Deccan traps is coincident with a major mass extinction: the Cretaceous extinction in which many species disappeared, including the dinosaurs.

Mantle dynamics are certainly more complex than many current models suggest. It is now time to move away from simple stratified or whole-layer convection models to a more dynamic model of convection incorporating both models of convection at different times. We are aware of dramatic events in the Earth's history, and have already correlated some, such as the mass extinction in the Cretaceous and the break-up of Pangea, with the arrival of super-plumes at the surface. Possible scenarios for time-variant convection models include quiescent periods of two-layer convection interspersed with short periods of rapid mixing and whole-mantle convection. For example, subducted slabs may ‘pool’ at the 670-km discontinuity before eventually breaking through the 670-km barrier and surging into the lower mantle, entraining upper mantle material with them. The role of the D layer in mantle dynamics is still unclear. It has been suggested that mantle plumes are formed in the D layer and rise to the surface from the base of the lower mantle. On the other hand, some believe that the D layer is a ‘graveyard’ of subducted slabs. It is clear that there are many exciting discoveries to be made before our understanding of mantle dynamics is complete.

D. Sharrock

Bibliography

C. M. R. Fowler (1990) The solid Earth: an introduction to global geophysics. Cambridge University Press.