Subduction Zones - Structure

Australia: The Land Where Time Began

Subduction Zones - Structure

Intense seismic activity is associated with subduction zones, many of these events occurring on a plane dipping on an average angle of about 45^o away from the oceanic plate that is underthrusting. This plane is called a Benioff (Benioff-Wadati) zone, earthquakes occurring on it extending from near the surface beneath the region of the forearc to a maximum depth of about 670 km. When earthquake foci are projected onto a vertical plane parallel to the underthrusting direction of the Tonga-Kermadec island arc system it can be seen that the foci are of progressively greater depth away from the underthrusting site in the Tonga Trench. The body wave amplitude of deep earthquakes has provided information on the nature of the Benioff zone. Seismic arrivals were found to be of much greater amplitude at islands such as Tonga, and other volcanic islands of the arc, than those from Rarotonga or Fiji, to the front and rear of the arc.

The term Q-factor, the inverse of the specific attenuation factor, is usually used to quantitatively describe the magnitude differences. Generally the higher the Q-factor the stronger the rock, and vice versa. The attenuation is lower for high Q-factor travel paths, the highest Q travel paths being associated with the strongest rock. It has been found that seismic waves travelling along the length of the seismic zone appear to pass through a high Q region of about 1000, but those travelling to the lateral locations pass through rocks with low Q of about 150. According to Kearey et al. (Source 1), the evidence suggests that the top of a high Q zone, about 100 km thick, is defined by the Benioff zone.

It was initially believed the Benioff zone was a large thrust fault between different crustal provinces, but seismic evidence has led to a new interpretation in which a belt of Pacific lithosphere of high Q had underthrust into the mantle. In the region of Tonga a local seismometer network has been used to refine this interpretation (Barazangi & Isacks, 1971)(see source 1, fig. 9.7).

A zone of very low Q, about 50, (low attenuation zone) was indicated in the uppermost mantle above a subsiding slab, in a region about 300 km wide that extended from Tonga, the active island arc, to the Lau Ridge, a backarc ridge. The implication of this evidence is that either the mantle is much weaker beneath the Lau Basin (backarc basin) than elsewhere, or alternatively the lithosphere is much thinner. This evidence has important ramifications for backarc basin origins (Kearey et al. (Source 1, Section 9.10).

A section through the Tonga Arc, where a region of relatively high P-wave velocity clearly defines a subducting slab, has been found by the use of seismic tomography (Source 1, Section 2.1.8, Plate 9). Corresponding to the region of extremely low Q (Source 1, Fig. 9.7), a region of low velocities beneath the Lau Basin (Source 1, Section 9.10) is present above the subducting slab, the lowest velocities being found beneath the volcanoes of the Tonga Arc.

There are 4 distinct processes involved with the earthquake activity associated with the subducting slab. As the slab begins its decent the bending , downward flexure of the lithosphere generates earthquakes. The upper surface of the plate is put into tension as a result, this stress regime leads to faulting that generate observed earthquakes that occur up to a 25 km deep (Christensen & Ruff, 1988).

The topographic bulge in the subducting plate, on the oceanward side of the island arc, results from flexural bending of the lithosphere, producing a raised seabed topography, with an amplitude of several hundred metres, that occurs 100-200 km from the axis of the trench. It can be predicted from simple beam theory that this bulge results from the deflection downward of the subducting plate. It has been found that this flexure is not completely elastic, also involving a large component of permanent, plastic deformation (Turcotte et al., 1978). It has been deduced that most topographic profiles fit with bending of a plate that is 2-layer elastic-perfectly plastic, 50 km thick, with the upper 20 km under tension and the lower 30 km under compression, and that variations of regional stress fields probably lead to variations in these profiles (Chapple & Forsyth, 1979).

Earthquakes are also generated by thrust faulting along the contact zone between underthrusting and overriding plates. The earthquake belt to the south of the islands of the Aleutian Island arc results from normal faulting associated with flexure of the top part of the Pacific Plate that underthrusts the Bering Sea to the northwest. Thrust faulting is indicated by the grouping of the epicentres of earthquakes beneath or a bit to the south of the island chain. The nodal planes dipping steeply to the south and gently to the north. It has been suggested the fault planes are represented by the latter, the earthquakes being generated by the relative motion between the Pacific and Bering Sea lithosphere. Strike-slip movement is indicated by the single focal mechanism solution, that is either on a sinistral strike-slip fault perpendicular to the island chain or on a dextral strike-slip fault parallel to the island chain. As the underthrusting in this region is in an oblique direction it has been suggested that the latter interpretation is probably more likely to be the correct one (Kearey et al., source 1).

Earthquakes occurring deeper than the thickness of the lithosphere at the surface, in the Benioff zone, result from the internal deformation of the relatively cold, therefore strong descending lithosphere slab, not being generated by thrusting at the top of the descending slab, as the aesthenosphere that is in contact with the plate is not strong enough to support extreme faulting. Beneath the Japan Arc 2 Benioff zones were identified, by the use of a local seismograph array, that appear to merge down-dip (Hasegawa et al., 1978). The upper of these zones corresponds to the crustal part of the descending slab and the lower to the mantle, as indicated by the arrival times of different seismic phases (Hasegawa et al., 1994).

Many subduction zones have been found to have double seismic zones, between 70 and 200 km deep (Peacock, 2001), leading to the conclusion that such double seismic features are common in subduction zones (Kearey et al., Source 1). Down-dip compression is sometimes implied by focal mechanism solutions for earthquakes in the upper zone, while in the lower zone earthquakes down-dip tension is implied. It has been suggested that unbending may be an important feature of descending plates, as it has already undergone some degree of plastic deformation as it bent to be subducted (Isacks & Barazangi, 1977). In subducting plates the double seismic zones continues down much further than the region in which the plate is unbending, suggesting that this is not the main cause of earthquakes. It has been proposed that most earthquakes are the result of metamorphic reactions. These are thought to involve eclogite formation in the upper zone (Kirby et al., 1996) and dehydration of serpentinite in the lower zone (Meade & Jeanloz, 1991). The suggestion is that high pore pressures are generated along fault planes that are pre-existing in the oceanic lithosphere that is subducting, the earthquakes being produced by brittle failure.

In the subduction zone beneath northeast Japan it has been found, based on a detailed thermal model, that as the focal depth increases from 70-180 km, the lower seismic zone migrates across the isotherms, about 800^o C to 400^o C. A P-T diagram of these temperatures and implied pressures shows that the pressure/temperature values are very analogous to those for the dehydration of serpentine to forsterite + enstatite + water, strongly suggesting that earthquakes result from the dehydration of the serpentinised mantle in the subducting plate of oceanic lithosphere. The assumption in this suggestion is that oceanic mantle is serpentinised to depths of several 10s of kilometres. At mid-ocean ridges it is believed that hydrothermal circulation and alteration occurs only in the crust. Oceanward of the trench normal faulting associated with the outer rise, as well as the oceanward bending of the oceanic lithosphere, could possibly allow the penetration of seawater to hydrate the lithosphere to depths of 10s of kilometres (Peacock, 2001).

Earthquakes are believed to result from a sudden phase change from olivine to spinel structure, to produce transformational or anticrack faulting, for depths greater than 300 km. Along planes where minute crystals of spinel have grown, the crystal lattice rapidly shears (Green, 1994). This phase change occurs at depths of about 400 km at normal mantle temperatures. In the core of subducting plates the temperature is anomalously low, allowing olivine to exist metastably to greater depths, possibly until its temperature reaches about 700^o C (Wiens et al., 1993). Occasionally it may be at depths of about 670 km in old slabs that are being subducted rapidly, which would explain the lowest depth of seismicity in subduction zones. It has also been suggested that in this depth range in subduction zones, similar transformation, from enstatite to ilmenite, may contribute to seismicity (Hogrefe et al., 1994). At depths of about 700 km, fine-grained material is believed to be produced by the phase changes occurring in the slab. It is thought this material cannot produce earthquakes as it behaves in a superplastic manner (Ito & Sato, 1991).

Principal stress directions that are either parallel or orthogonal to the dip of the descending plate characterise deep events at 'c' and 'd' in Fig. p.8 in Source 1, at about 200 km and about 500 km (Isacks et al., 1969). The result is that the nodal planes determined by focal mechanism solutions correspond to neither the dip of the Benioff zone nor the a plane that is perpendicular to it. The descending plate is thrown into either down-slope compression or tension, as indicated by the principal stress directions. It has been suggested that the distribution in the seismic zone of stress type may result from the degree of resistance on the plate as it descends (Isacks & Molnar, 1969), this resistance being described in terms of the net effect of ridge push and slab pull forces (Source 1, section 12.6), Spence, 1987). According to Kearey et al., (Source 1, 9.14a), negative buoyancy of the plate causes it to sink through the aesthenosphere, and as its descent is resisted, it is thrown into down-dip tension. According to Fig. 9.14b (Source 1) the leading edge of the plate is thrown into compression as the bottom of the plate approaches the mesosphere, that resists the descent of the plate. Fig. 9.14c (Source 1) illustrates the the mesosphere preventing further sinking of the plate, supporting the lower margin, leading to compression being experienced by the majority of the seismic zone. In Fig. 9.14d (/source 1) decoupling of a section of the subsiding plate occurs, throwing the upper portion into tension and the lower portion into compression. Focal mechanism solutions have been used to determine stress directions, and a summary has been produced (Isacks & Molnar, 1971).

At some trenches, such as the Peru-Chile Trench, seismic gaps have been observed along the middle sections of the Benioff zone, in places where the slab is known to be continuous. This is believed to be explained by the stress distribution along the plate, as seen in Fig. 9.14b,d, Source 1), where there is down-dip tensional stress along the top of the slab, a gap in the middle section and down-dip compressional stress at the tip section of the slab (James & Snoke, 1990). At some island arcs it is believed a further type of seismic gap is present at shallow depths. There is a prominent seismic gap at the Aleutian-Alaska Arc (Jacobs et al., 1977) between the trench and a point about half way towards the volcanic arc, that is progressively greater from west to east. In this region there is a very shallow angle of underthrusting. It is believed the cause of this seismic gap is the large quantity of terrigenous sediment, especially in the section of the trench nearest Alaska. It has been suggested the build-up of strain energy necessary for the production of catastrophic earthquakes is probably prevented by the presence of this unconsolidated sediment, with its high positive buoyancy, the plate subducting at an anomalously shallow angle as a result.

In has been noted in data from many subduction zones that within or just beneath the transition zone the subducting slabs are either deflected horizontally, or alternatively, descend into the lower mantle after penetrating the 660 km discontinuity. The slabs appear to flatten out within the transition zone beneath places such as the Chile, the Aleutians, southern Kurile, and Izu-Bonin, while in other places, such as beneath the Aegean, central Japan, Central America and Indonesia, they continue deep into the lower mantle. Beneath Tonga the slab does both, flattening out within the transition zone as well as extending into the lower mantle (van der Hilst, 1995). There doesn't appear to be a relationship between the age of a slab and whether or not it penetrates into the lower mantle (Kearey et al., Source 1). According to some researchers there is some evidence that suggests the slabs are descending through the lower mantle into the core-mantle boundary, though others believe there is little evidence for slabs going deeper than 1700 km (Karason & van der Hilst, 2000) (Kearey et al., Source 1).

See Source 1 for more detailed information on all aspects of plate tectonics

Mantle Plume Head - Influence on Dynamics of Retreating Subduction Zone

Sources & Further reading