seismology and plate tectonics

seismology and plate tectonics Earthquake seismology may be divided into seismicity—the study of earthquake distribution and mechanisms—and seismology sensu stricto—the use of earthquakes to probe the interior structure of the Earth. Both branches have had a profound influence on the development of plate-tectonic theory.

Seismicity and plate boundaries

One of the fundamental tenets of plate tectonics is that plates are rigid and undeformable. This idea arose largely from the observed distribution of earthquakes around the world, which became readily apparent in the 1960s after the setting up of the World-Wide Standard Seismograph Network (WWSSN). Global maps of earthquakes (Fig. 1) show that most events, especially in the oceans, are concentrated in relatively narrow bands, with broad aseismic areas between them. The narrow banks of seismicity are taken to indicate active plate boundaries, while the aseismic areas are the rigid interiors of the plates. Details of the seismicity of different types of plate boundary are considered below.

Earthquake mechanisms

As well as marking the positions of plate boundaries, earthquakes can be used to determine the nature of faulting, the major stress directions, and sometimes the directions of relative plate movement. This is done by determining the ‘fault-plane solution’ or ‘focal mechanism’.

A relatively simple way of doing this is to measure the direction of first motion recorded by P-waves on a seismograph. In P-waves, the ground motion is back-and-forth along the ray path. If one considers an earthquake at a strike-slip fault, for example (Fig. 2a), the ground around the fault can be divided into four quadrants, according to whether its initial direction of motion is toward or away from the epicentre. At the lines separating the quadrants the initial motion is zero. In fact zero motion is also experienced on the vertical extensions of these lines, which are called ‘nodal planes’. Similar divisions into quadrants can be made for faults in any orientation. For simple fault models the nodal planes are always perpendicular to each other.

A seismograph in a quadrant with first motion away from the epicentre will itself experience a first motion away from the epicentre (i.e. towards the seismograph station). Such motion tends to compress the ground it moves into, and this quadrant is thus referred to as a compressional quadrant. In contrast, quadrants in which the first motion is toward the epicentre are dilatational, and first motions on seismographs in these quadrants will be away from the station.

First motions are conventionally plotted on a hemisphere just below the hypocentre (or focus), at the position where the seismic ray crosses the hemisphere. The two best-fitting orthogonal nodal planes are then fitted to the plotted data and the compressive quadrants are shaded, producing a ‘beachball’ diagram. The axes of maximum compressive and tensional stress run through the centres of the dilatational and compressive quadrants, respectively. One cannot tell from first motions alone which is the actual fault plane. Other information, such as the strike of a mapped fault, observed breaks in the ground, or alignments of aftershocks has to be used.

More recently, new methods have been developed for determining focal mechanisms, including modelling the complete radiation patterns of both P- and S-waves.

As shown in Fig. 2, different patterns of compressions and dilatations are characteristics of strike-slip (horizontal), normal (extensional), and thrust (compressional) faulting. These vary at different plate boundaries.

Divergent boundaries

The narrowest seismic zones are associated with the crests of the mid-ocean ridges, which contain divergent plate boundaries. This association was one of the prime pieces of evidence showing that ridges are tectonically active, which led to the theory of sea-floor spreading, itself an important component of plate tectonics. Global seismic networks such as WWSSN resolve the epicentres of these mid-ocean ridge earthquakes to about 20 km, and show that the bands are normally little wider than this. More detailed local studies, for example using local networks on ocean islands or ocean-bottom seismographs, show that mid-ocean ridge seismicity commonly occurs in clusters only a few kilometres across. These can in some instances be seen to be associated with individual faults or inferred dyke-intrusion zones and fissure eruptions. Such activity rarely occurs over a zone more than a few kilometres wide, although occasional earthquakes may occur tens of kilometres from ridge axes.

Earthquakes at divergent boundaries are virtually entirely confined to depths of 8 km or less which may be taken to define the depth of the brittle–ductile transition or, effectively, the base of the zero-age lithosphere.

Fault-plane solutions (Fig. 3) show that divergent boundaries are mainly characterized, as expected, by extensional, normal fault mechanisms. Almost all earthquakes here are of moderate magnitude, usually less than magnitude 6, because the young, weak lithosphere and predominantly extensional faults do not support very large accumulations of stress before breaking or slipping.

Transform boundaries

Along many mid-ocean ridges, especially those opening at higher spreading rates, such as the East Pacific Rise, most seismicity is concentrated not on the divergent boundary itself but on the transform faults (conservative, strike-slip boundaries) which offset the divergent boundaries at intervals of tens to hundreds of kilometres. Here, detailed studies have shown that most seismicity is centred on a single strike-slip fault representing the very narrow plate boundary.

The concept of transform faults is another critical aspect of plate-tectonic theory. The idea was introduced by the Canadian geophysicist J. Tuzo Wilson in 1965. He defined transform faults as a class of strike-slip fault that connects two belts of tectonic activity; in plate tectonics, transform faults are thus themselves one type of plate boundary. Their importance is that they show how major transcurrent faults can terminate by transforming into different types of plate boundary.

The transform faults that offset mid-ocean ridges connect two similar types of tectonic activity (two spreading centres or divergent boundaries). If they are true Wilsonian transform faults, they should be active only between the offset spreading centres, and not on the prolongation of their trend beyond these centres (Fig. 3). The distribution of earthquakes on ridge–ridge transform faults shows that this is the case. Note also that the sense of motion on a transform fault is the opposite of that required on a simple transcurrent fault offsetting a ridge in the same direction. This is also confirmed by observed focal mechanisms.

Transform faults are not confined to the oceans, but occur on land in areas such as California (the San Andreas fault complex), Turkey (North and South Anatolian faults) and Asia (Altyn Tagh fault), and elsewhere. Many of these continental fault systems are quite complex, with numerous secondary anastomosing and splaying faults, many of which may be active simultaneously. Many of these complexities are reflected in the distribution of seismic activity, as shown for example by the very detailed seismic maps produced in California.

Continental transform faults can generate larger earthquakes than those on mid-ocean ridges, magnitudes of about 7 or more being not uncommon (e.g. the North Ridge earthquake, California 1994). These larger-magnitude events normally occur on faults with a component of thrusting, where larger stresses can build up.

Seismicity on continental transform faults extends to depths of about 15 km, where much of the deformation appears to change from brittle to ductile.

Subduction zones

A third vital concept in plate tectonics is that of the subduction zones. If, as palaeomagnetic and other studies indicate, the Earth's radius is not increasing rapidly, plate created at divergent boundaries must be destroyed elsewhere. Subduction zones are where this happens. There, plates descend at an angle into the mantle. The traces of these dipping, subducted plates are marked by zones of earthquakes (Fig. 4) called Benioff or Benioff–Wadati zones after the Californian and Japanese seismologists who first recognized them.

Benioff zones normally dip at about 45°, although the angle ranges from about 20° to almost vertical. The earthquakes can extend down to the base of the upper mantle at 670 km, and the epicentres of the deepest can thus be several hundred kilometres horizontally away from the plate boundary (which is usually marked by an oceanic trench) at the surface. This is what accounts for the broadening of epicentral zones around the Pacific rim (Fig. 1).

Where the subducting plate bends down into the subduction zone, most earthquakes are tensional (with normal fault mechanisms), and are thought to result from the tensional stresses developed in the upper part of the plate as it bends. These earthquakes extend to about 25 km depth. At the plate boundary, and for a few tens of kilometres beyond it on the side of the overriding plate, earthquakes are mainly compressional. They have thrusting mechanisms reflecting the contact between the two plates and the compressional deformation of the overriding plate. Deeper earthquakes make up the Benioff zone itself.

Detailed studies using local seismograph networks have resolved two parallel bands of dipping earthquakes in the Benioff zone (Fig. 4). The uppermost corresponds to the upper surface of the subducting slab, while the lower lies within the slab. Those earthquakes in the upper band have compressional mechanisms and are thought to be related to the unbending of the slab, combined with the effects of the frictional resistance of the surrounding asthenosphere. Those in the lower band reflect internal deformation of the descending plate. Events in the shallower part of the lower band are tensional, but at greater depths these intraplate events become compressional. Their precise cause is still unknown, but suggested mechanisms include extension of the slab under its own weight (in the upper part), compression due to increasing subduction resistance (in the lower part), and the volumetric effects of mineral phase changes.

The largest-magnitude earthquakes found on Earth occur within subduction zones, mainly in the shallow region of thrusting, where very large stresses can build up. The largest recorded earthquake (Good Friday 1964 in Alaska, magnitude 9.2) occurred on the Aleutian subduction zone.

Continental collision zones

The broadest zones of seismicity occur within continental collision zones such as the Himalayas, Tibet, and the surrounding region (Fig. 1). This reflects the complex nature of tectonic boundaries in such areas, where plate tectonics is least applicable. Here, the lithosphere, which is thicker but weaker than oceanic lithosphere, is broken into large numbers of tectonic blocks delimited by active faults in zones many hundreds or even thousands of kilometres wide. Although individual earthquakes still lie on discrete faults, the plethora of such features produces a complex pattern. Since these small tectonic blocks may all be in motion relative to each other, the full range of thrust, strike-slip, and tensional mechanisms can occur in different parts of the zone.

Most earthquakes in continental collision zones are relatively shallow, being confined to the thickness of the rigid lithosphere: 100 km or so. However, sinking lithospheric slabs may exist below this depth, either as the remnants of the subduction zones that led to the collision, or as recently delaminated blocks of lower lithosphere. These may be marked by deeper earthquakes extending hundreds of kilometres below the surface.

As with subduction zones, continental collision zones can support large stresses, and so large-magnitude earthquakes are also found here.

Seismology and the lithosphere

As well as defining the positions of plate boundaries and the nature of the motion and deformation across them, seismology has provided important information on the nature of lithospheric plates and their surrounds.

The lithosphere is defined as a high-strength, rigid layer, and seismology has been able to estimate its thickness. Study of P-wave and S-wave travel times worldwide has shown that the velocities of these waves exhibit distinct minima at depths of around 100–200 km. This feature is almost ubiquitous, and is referred to as the low-velocity zone (LVZ). All else being equal, a lower seismic velocity indicates that a material is nearer to its melting point. The LVZ is thought to correspond to mantle material that is close to its melting point and is thus significantly weakened. In consequence, the LVZ is often approximately equated with the asthenosphere (the weak layer below the lithospheric plates), the strong plates corresponding to the slightly higher velocities above.

The thickness of the plates themselves can be estimated by using seismic surface waves. Like water waves on the ocean, their particle motion penetrates a fraction of a wavelength into the Earth; deeper for longer wavelengths. Surface waves have a range of wavelengths from tens to hundreds of kilometres. Their velocities reflect seismic velocity in the Earth at the depths to which they penetrate. The longer surface waves are thus sensitive to the Earth's properties near the base of the lithosphere. To fit the observed surface-wave velocities requires models with a high-velocity ‘lid’ (the plate) overlying a lower-velocity channel (the asthenosphere). It is found that the lid thickens with increasing age of the plate in just the way that is predicted by thermal models of plates.

Roger Searle

Bibliography

Isacks, B. L.,, Oliver, J.,, and and Sykes, L. R. (1968) Seismology and the new global tectonics. Journal of Geophysical Research, 73, 5855–900.
Kearey, P. and and Vine, F. J. (1996) Global tectonics. Blackwell Science, Oxford.