Plate Tectonics and Crustal Evolution PDF

Chapter 1 Plate tectonics scientific disciplines tend to evolve from a stage prim- A perspective arily of data gathering, characterized by transient hypo- Plate tectonics, which has so profou...

Chapter 1 Plate tectonics scientific disciplines tend to evolve from a stage prim- A perspective arily of data gathering, characterized by transient hypo- Plate tectonics, which has so profoundly influenced geo- theses, to a stage where a new unifying theory or theories logic thinking since the early 1970s, provides valuable are proposed that explain a great deal of the accumu- insight into the mechanisms by which the Earth's crust lated data. Physics and chemistry underwent such revo- and mantle have evolved. Plate tectonics is a unifying lutions around the beginning of the twentieth century, model that attempts to explain the origin of patterns of whereas the Earth Sciences entered such a revolution in deformation in the crust, earthquake distribution, contin- the late 1960s. As with scientific revolutions in other ental drift, and mid-ocean ridges, as well as providing a fields, new ideas and interpretations do not invalidate mechanism for the Earth to cool. Two major premises of earlier observations. On the contrary, the theories of plate tectonics are: seafloor spreading and plate tectonics offer for the first time unified explanations for what, before, had seemed 1 the outermost layer of the Earth, known as the litho- unrelated observations in the fields of geology, paleon- sphere, behaves as a strong, rigid substance resting tology, geochemistry, and geophysics. on a weaker region in the mantle known as the The origin and evolution of the Earth's crust is a tan- asthenosphere talizing question that has stimulated much speculation 2 the lithosphere is broken into numerous segments or and debate dating from the early part of the nineteenth plates that are in motion with respect to one another century. Some of the first problems recognized, such and are continually changing in shape and size (Fig- as how and when did the oceanic and continental crust ure 1.1/Plate 1). form, remain a matter of considerable controversy even The parental theory of plate tectonics, seafloor spread- today. Results from the Moon and other planets indicate ing, states that new lithosphere is formed at ocean ridges that the Earth's crust may be a unique feature in the and moves away from ridge axes with a motion like that Solar System. The rapid accumulation of data in the fields of a conveyor belt as new lithosphere fills in the result- of geophysics, geochemistry, and geology since 1950 ing crack or rift. The mosaic of plates, which range from has added much to our understanding of the physical 50 to over 200 km thick, are bounded by ocean ridges, and chemical nature of the Earth's crust and of the pro- subduction zones (in part coUisional boundaries), and cesses by which it evolved. Evidence favours a source for transform faults (boundaries along which plates slide the materials composing the crust from within the Earth. past each other) (Figure 1.1/Plate 1, cross-sections). To Partial melting of the Earth's mantle produced magmas accommodate the newly-created lithosphere, oceanic that moved to the surface and formed the crust. The plates return to the mantle at subduction zones such that continental crust, being less dense than the underlying the surface area of the Earth remains constant. Harry mantle, has risen isostatically above sea level and hence Hess is credited with proposing the theory of seafloor is subjected to weathering and erosion. Eroded materials spreading in a now classic paper finally published in are partly deposited on continental margins, and partly 1962, although the name was earlier suggested by Robert returned to the mantle by subduction to be recycled and Dietz in 1961. The basic idea of plate tectonics was perhaps again become part of the crust at a later time. proposed by Jason Morgan in 1968. Specific processes by which the crust formed and evolved Many scientists consider the widespread acceptance are not well-known, but boundary conditions for crustal of the plate tectonic model as a 'revolution' in the Earth processes are constrained by an ever-increasing data base. Sciences. As pointed out by J. Tuzo Wilson in 1968, In this book, physical and chemical properties of the 2 Plate Tectonics and Crustal Evolution Figure 1.1 Map of the major lithospheric plates on Earth. Arrows are directions of plate motion. Filled barbs, convergent plate boundaries (subduction zones and coUisional orogens); single lines, divergent plate boundaries (ocean ridges) and transform faults. Cross-sections show details of typical plate boundaries. Artwork by Dennis Tasa, courtesy of Tasa Graphic Arts, Inc. Plate tectonics 3  Plate Tectonics and Crustal Evolution LITHOSPHERE Figure 1.2 Distribution of ASTHENOSPHERE average compressional (Vp) and shear wave (Vs) velocities and average calculated density (p) in the Earth. Also shown are temperature distributions for whole-mantle convection (TW) and layered-mantle convection (TL). 'O CO 0) Q. E — -H2000 1000 2000 3000 4000 5000 6000 Depth (km) Earth are described, and crustal origin and evolution are outer layer of the Earth, including the crust, that discussed in the light of mantle dynamics and plate tec- reacts to many stresses as a brittle sohd. The astheno- tonics. Included also is a discussion of the origin of the sphere, extending from the base of the Iithosphere atmosphere, oceans, and life, which are all important to the 660-km discontinuity, is by comparison a weak facets of Earth history. Finally, the uniqueness of the layer that readily deforms by creep. A region of low Earth is contrasted with the other planets. seismic-wave velocities and high attenuation of seismic-wave energy, the low-velocity zone (LVZ), occurs at the top of the asthenosphere and is from Structure of the Earth 50-100 km thick. Significant lateral variations in density and in seismic-wave velocities are common First of all we need to review what is known about the at depths of less than 400 km. structure of planet Earth. The internal structural of the The upper mantle extends from the Moho to the Earth is revealed primarily by compressional (P-wave) 660-km discontinuity, and includes the lower part of and shear (S-wave) waves that pass through the Earth in the Iithosphere and the upper part of the astheno- response to earthquakes. Seismic-wave velocities vary sphere. The region from the 4l0-km to the 660-km with pressure (depth), temperature, mineralogy, chemi- discontinuity is known as the transition zone. These cal composition, and degree of partial melting. Although two discontinuities, as further discussed in Chapter 4, the overall features of seismic-wave velocity distribu- are caused by two important solid-state transforma- tions have been known for some time, refinement of tions: from olivine to wadsleyite at 410 km and from data has been possible in the last ten years. Seismic- spinel to perovskite + magnesiowustite at 660 km. wave velocities and density increase rapidly in the re- The lower mantle extends from the 660-km discon- gion between 200 and 700 km deep. Three first-order tinuity to the 2900-km discontinuity at the core- seismic discontinuities divide the Earth into crust, mantle boundary. For the most part, it is characterized mantle and core (Figure 1.2): the Mohorovicic discon- by rather constant increases in velocity and density tinuity, or Moho, defining the base of the crust; the in response to increasing hydrostatic compression. core-mantle interface at 2900 km; and, at about 5200 Between 220-250 km above the core-mantle inter- km, the inner-core/outer-core interface. The core com- face a flattening of velocity and density gradients prises about sixteen per cent of the Earth by volume and occurs, in a region known as the D'' layer, named thirty-two per cent by mass. These discontinuities reflect after the seismic wave used to define the layer. The changes in composition or phase, or both. Smaller, but lower mantle is also referred to as the mesosphere, very important velocity changes at 50-200 km, 410 km, a region that is strong, but relatively passive in terms and 660 km provide a basis for further subdivision of of deformational processes. the mantle, as discussed in Chapter 4. The outer core will not transmit S-waves and is The major regions of the Earth can be summarized as interpreted to be liquid. It extends from the 2900-km follows with reference to Figure 1.2: to the 5200-km discontinuity. 1 The crust consists of the region above the Moho, The inner core, which extends from 5200-km dis- and ranges in thickness from about 3 km at some continuity to the centre of the Earth, transmits S- oceanic ridges to about 70 km in coUisional orogens. waves, although at very low velocities, suggesting 2 The iithosphere (50-300 km thick) is the strong that it is near the melting point.  Plate tectonics 5 There are only two layers in the Earth with anomalously low seismic velocity gradients: the LVZ at the base of the lithosphere and the D" layer just above the core (Figure 1.2). These layers coincide with very steep tem- perature gradients, and hence are thermal boundary lay- ers within the Earth. The LVZ is important in that plates are decoupled from the mantle at this layer: plate tec- tonics could not exist without an LVZ. The D" layer is important in that it may be the site at which mantle plumes are generated. Considerable uncertainty exists regarding the tempera- ture distribution in the Earth. It is dependent upon such features of the Earth's history as: 1 the initial temperature distribution 2 the amount of heat generated as a function of both depth and time 3 the nature of mantle convection 4 the process of core formation. Most estimates of the temperature distribution in the Earth are based on one of two approaches, or a combination of both: models of the Earth's thermal history involving various mechanisms for core formation, and models in- volving redistribution of radioactive heat sources in the Earth by melting and convection processes. Estimates using various models seem to converge on a temperature at the core-mantle interface of about 4500 ± 500 °C and the centre of the core 6700 to 7000 °C. Two examples of calculated temperature distributions in the Earth are shown in Figure 1.2. Both show significant gradients in temperature in the LVZ and the D" layer. The layered convection model also shows a large tem- perature change near the 660-km discontinuity, since this is the boundary between shallow and deep convec- tion systems in this model. The temperature distribution Figure 1.3 Linear magnetic anomalies and fracture zones for whole-mantle convection, which is preferred by most in the NE Pacific basin. Positive anomalies in black. After scientists, shows a rather smooth decrease from the top Raff and Mason (1961). of the D" layer to the LVZ. anomalies could be explained by seafloor spreading. No other single observation in the last 50 years has had Seafloor spreading such a profound effect on geology. With this observa- Seafloor spreading was proposed to explain linear mag- tion, we entered a new scientific era centred around a netic anomalies on the sea floor by Vine and Matthews dynamic Earth. in 1963. These magnetic anomalies (Figure 1.3), which Ocean ridges are accretionary plate boundaries where had been recognized since the 1950s but for which no new lithosphere is formed from upwelling mantle as the satisfactory origin had been proposed, have steep flanking plates on both sides of ridges grow in area and move gradients and are remarkably linear and continuous, away from the axis of the ridge (Figure 1.1/Plate I, except where broken by fracture systems (Harrison, cross sections). In some instances, such as the South 1987). Vine and Matthews (1963) proposed that these Atlantic, new ocean ridges formed beneath superconti- anomalies result from a combination of seafloor spread- nents, and thus as new oceanic lithosphere is produced ing and reversals in the Earth's magnetic field, the record at a ridge the supercontinent splits and moves apart on of reversals being preserved in the magnetization in the each of the ridge flanks. The average rate of oceanic upper oceanic crust. The model predicts that lines of lithosphere production over the past few million years is alternate normally and reversely magnetized crust should about 3.5 km^/y and, if this rate is extrapolated into the parallel ocean ridge crests, with the pronounced mag- geologic past, the area covered by the present ocean netic contrasts between them causing the observed steep basins (sixty-five per cent of tne Earth's surface) would linear gradients. With the Geomagnetic Time Scale deter- be generated in less than 100 My. In fact, the oldest mined from paleontologically-dated deep sea sediments, ocean floor dates only to about 160 Ma, because older Vine (1966) showed that the linear oceanic magnetic oceanic plates have been subducted into the mantle. 6 Plate Tectonics and Crustal Evolution I I -I \ 1 1 1 1 r—I r—n 1 1 1 1 1 1 r n I n ^p^ Figure 1.4 Distribution of world earthquakes 1961-1969. From National Earthquake Information Center Map NEIC-3005. (Figure 1.1/Plate 1, cross-sections). At collisional zones, Plate boundaries plates carrying continents may become sutured together. Introduction Plates diminish or grow in area depending on the dis- tribution of convergent and divergent boundaries. The Earthquakes occur along rather narrow belts (Figure 1.4), African and Antarctic plates, for instance, are almost and these belts mark boundaries between lithospheric entirely surrounded by active spreading centres and hence plates. There are four types of seismic boundaries, dis- are growing in area. If the surface area of the Earth is tinguished by their epicentre distributions and geologic to be conserved, other plates must be diminishing in area characteristics: ocean ridges, subduction zones, trans- as these plates grow, and this is the case for plates in the form faults, and collisional zones. Provided a sufficient Pacific area. Plate boundaries are dynamic features, not number of seismic recording stations with proper azi- only migrating about the Earth's surface, but changing muthal locations are available, it is possible to deter- from one type of boundary to another. In addition, new mine the directions of first motion at sites of earthquake plate boundaries can be created in response to changes generation, which in turn provides major constraints on in stress regimes in the lithosphere. Also, plate boundaries plate motions. disappear as two plates become part of the same plate, Modem plates range in size from < 10"^ km^ to over for instance after a continent-continent collision. Small 10^ km^ and plate margins do not usually coincide with plates (< 10^ km^) occur most frequently near continent- continental margins (Figure 1.1/Plate 1). Seven major continent or arc-continent collisional boundaries and are plates are recognized: the Eurasian, Antarctic, North characterized by rapid, complex motions. Examples are American, South American, Pacific, African and Aus- the Turkish-Aegean, Adriatic, Arabian, and Iran plates tralian plates. Intermediate-size plates (10^-10^ km^) in- located along the Eurasian-African continent-continent clude the Philippine, Arabian, Nasca, Cocos, Caribbean collision boundary, and several small plates along the and Scotia plates. In addition, there are more than twenty continent-arc collision border of the Australian-Pacific plates with areas of 10^-10^ km^. Both plate theory and plates. The motions of small plates are controlled largely first-motion studies at plate boundaries indicate that plates by the compressive forces of larger plates. are produced at ocean ridges, consumed at subduction Continental margins are of two types: active and pas- zones, and slide past each other along transform faults sive. An active continental margin is found where  Plate tectonics FAST Figure 1.5 Axial (>45cm/a) topographic profiles across (EPR 3«S) three ocean ridges. EPR, East Pacific rise; MAR, Mid-Atlantic ridge. V and F indicate widths of zones of active volcanism and faulting, respectively. After Macdonald (1982). 0 10 DISTANCE (km) either a subduction zone or a transform fault coincides gest that differences in horizontal stresses in the oceanic with continent-ocean interface. Examples are the An- lithosphere may account for the relationship between dean and Japan continental-margin arc systems and the ridge topography and spreading rate (Morgan et al., San Andreas transform fault in California. Passive con- 1987). As oceanic lithosphere thickens with distance from tinental margins occur along the edges of opening ocean a ridge axis, horizontal extensional stresses can produce basins like the Atlantic basin. These margins are char- the axial topography found on slow-spreading ridges. In acterized by minimal tectonic and igneous activity. fast-spreading ridges, however, the calculated stresses are too small to result in appreciable relief. The axis of ocean ridges is not continuous, but may be offset by Divergent boundaries (ocean ridges) several tens to hundreds of kilometres by transform faults The interconnected ocean-ridge system is the longest (Figure 1.1/Plate 1). Evidence suggests that ocean ridges topographic feature on the Earth's surface, exceeding grow and die out by lateral propagation. Offset mag- 70 000 km in length. Typical ocean ridges are 3000- netic anomalies and bathymetry consistent with propa- 4000 km wide, with up to several kilometres of relief gating rifts, with and without transform faults, have been in the axial rift zone. Ocean ridges are characterized described along the Galapagos ridge and in the Juan de by shallow earthquakes limited to axial rift zones. These Fuca plate (Hey et al., 1980). earthquakes are generally small in magnitude, commonly occur in swarms and appear to be associated with intru- sion and extrusion of basaltic magmas. First-motion stud- Transform faults and fracture zones ies indicate that rift earthquakes are produced dominantly Transform faults are plate boundaries along which plates by vertical faulting as is expected if new lithosphere is slide past each other and plate surface is conserved. They being injected upwards. Most faulting occurs in the depth uniquely define the direction of motion between two range of 2-8 km and some ruptures extend to the sea bounding plates. Ocean-floor transform faults differ from floor. transcurrent faults in that the sense of motion relative The median valley of ocean ridges varies in geologi- to offset along an ocean ridge axis is opposite to that cal character due to the changing importance of tectonic predicted by transcurrent motion (Wilson, 1965) (Figure extension and volcanism. In the northern part of the 1.6). These offsets may have developed at the time Mid-Atlantic ridge, stretching and thinning of the crust spreading began and reflect inhomogeneous fracturing dominate in one section, while volcanism dominates in of the lithosphere. Transform faults, like ocean ridges, another. Where tectonic thinning is important, faulting are characterized by shallow earthquakes (< 50 km deep). has exposed gabbros and serpentinites from deeper crustal Both geophysical and petrological data from ophiolites levels. Volcanic features range from large ridges (> 50 cut by transforms suggest that most oceanic transforms km long) in sections of the median valley where are ieaky', in that magma is injected along fault sur- volcanism has dominated, to small volcanic cones in faces producing strips of new lithosphere (Garfunkel, sections dominated by extension. The axial topography 1986). Transforms cross continental or oceanic crust and of fast- and slow-spreading ridges varies considerably. may show apparent lateral displacements of many hun- A deep axial valley with flanking mountains character- dreds of kilometres. First-motion studies of oceanic trans- izes slow-spreading ridges, while relatively low relief, form faults indicate lateral motion in a direction away and in some instances a topographic high, characterize from ocean ridges (Figure 1.6). Also, as predicted by fast-spreading ridges (Figure 1.5). Model studies sug- seafloor spreading, earthquakes are restricted to areas 8 Plate Tectonics and Crustal Evolution pH diapirs migrate, and suggests that ocean-ridge segments - and diapirs are decoupled from underlying mantle flow. ^ *^ - - -* Triple junctions TRANSFORM TRANSCURRENT Triple junctions are points where three plates meet. H 1 ^ 1 Aseismic Such junctions are a necessary consequence of rigid plates ^ Extension on a sphere, since this is the common way a plate bound- ary can end. There are sixteen possible combinations * ^ | |-^ — of ridge, trench, and transform-fault triple junctions (McKenzie and Morgan, 1969), of which only six are common. Triple junctions are classified as stable or Figure 1.6 Motion on transform and transcurrent faults unstable, depending on whether they preserve their relative to an ocean ridge axis (double vertical lines). Note geometry as they evolve. The geometric conditions for that the amount of offset increases with transcurrent motion, stability are described with vector velocity triangles in while it remains constant with transform motion. Bold Figure 1.7 and only RRR triple junctions are stable for arrows refer to spreading directions, small arrows to plate all orientations of plate boundaries. It is important to motions. understand evolutionary changes in triple junctions, be- between offset ridge axes. Transform faults may pro- cause changes in their configuration can produce changes duce large structural discontinuities on the sea floor, and that superficially resemble changes in plate motions. in some cases structural and topographic breaks known Triple junction evolution is controlled by the lengths of as fracture zones mark the locations of former ridge- transform faults, spreading velocities, and the availabil- ridge transforms on the sea floor. There are three types ity of magma. of transform faults: ridge-ridge, ridge-trench, and trench- trench faults. Ridge-ridge transform faults are most common, and these may retain a constant length as a Convergent boundaries (subduction zones) function of time for symmetrical spreading, whereas Convergent plate boundaries are defined by earthquake ridge-trcnch and trench-trench transforms decrease or hypocentres that lie in an approximate plane and dip increase in length as they evolve. beneath arc systems. This plane, known as the seismic Studies of oceanic transform-fault topography and zone or Benioff zone, dips at moderate to steep angles structure indicate that zones of maximum displacement and extends in some instances to the 66()-km seismic are very localized (< 1 km wide) and are characterized discontinuity. The seismic zone is interpreted as a brittle by an anatomizing network of faults. Steep transform region in the upper 10-20 km of descending lithospheric valley walls are composed of inward-facing scarps asso- slabs. Modem seismic zones vary significantly in hypo- ciated with normal faulting. Large continental transform centre distribution and in dip (Figure 1.8). Some, such faults form where pieces of continental lithosphere are as the seismic zone beneath the Aleutian arc, extend to squeezed within intracontinental convergence zones such depths < 300 km while others extend to the 660-km as the Anatolian fault in Turkey. Large earthquakes discontinuity (Figure 1.8, a and b respectively). In gen- (M > 8) separated by long periods of quiescence occur eral, seismic zones are curved surfaces with radii of along 'locked' segments of continental transforms, curvature of several hundred kilometres and with irregu- whereas intermediate-magnitude earthquakes character- larities on scales of < 100 km. Approximately planar ize fault segments in which episodic slippage releases seismic zones are exceptional. Seismic gaps in some stresses. Large earthquakes along continental transforms zones (e.g., c and e) suggest, although do not prove, appear to have a period of about 150 years, as indicated fragmentation of the descending slab. Dips range from by records from the San Andreas fault in California. 30 ° to 90 °, averaging about 45 \ Considerable varia- Ridge segments between oceanic transforms behave tion may occur along strike in a given subduction zone, independently of each other. This may be caused by as exemplified by the Izu-Bonin arc system in the West- instability in the convective upcurrents that feed ocean ern Pacific. Hypocentres are linear and rather continu- ridges, causing these upcurrents to segment into regularly- ous on the northern end of this arc system, (d), becoming spaced rising diapirs, with each diapir feeding a differ- progressively more discontinuous toward the south. Near ent ridge segment. Transforms may arise at the junctions the southern end of the arc, the seismic zone exhibits a of ridge segments because magma supply between diapirs pronounced gap between 150 and 400 km depth, (c). A is inadequate for normal oceanic crustal accretion. The large gap in hypocentres in descending slabs, such as persistence" of transforms over millions of years indi- that observed in the New Hebrides arc, (e), may indicate cates that asthenospheric diapirs retain their integrity for that the tip of the slab broke off and settled into the long periods of time. It appears from the use of fixed mantle. Because some slabs appear to penetrate the 660- hotspot models of absolute plate motion that both ridge km discontinuity (Chapter 4), the lack of earthquakes axes and transforms migrate together at a rate of a few below 700 km probably reflects the depth of the brittle- centimetres per year. This, in turn, requires that mantle ductile transition in descending slabs. An excellent cor-  Plate tectonics Geometry Velocity triangle Stability Example Figure 1.7 Geometry and stability requirements of six common triple junctions. Dashed lines ab, be, and ac All orientations East Pacific Rise XV in the velocity triangles join stable and Galapagos points, the vector sum of RRR R i f t zone. which leave the geometry of AB, BC, and AC, respectively, unchanged. The junctions are stable Stable if ab,ac Central only if ab, be, and ac meet form a straight Japan. at a point. Key: track line, or if be is symbol, trench; double line, TTT parallel to the ocean ridge; single line, slip vector CA transform fault. Stable if ac,bc Intersection of form a straight the Peru-Chile line, or if C lies Trench and the on ab West Chile Rise. TTF Stable if C lies Owen fracture on ab,or if ac, zone and the be form a Carlsberg Ridge straight line West Chile Rise FFR and the East Pacific Rise. Stable if ab, be San Andreas form a straight Fault and M e n - line, or if ae,be docino Fracture FFT do so Zone. Stable if ab goes Mouth of the through C, or if Gulf of California. ac, be form a RTF straight line relation exists between the length of seismic zones and extend upwards forming thrusts that dip away from the the product of plate convergence rate and age of the trench axis. Calculations of stress distributions at < 300 downgoing slab. km depth show that compressional stresses generally First-motion studies of earthquakes in subduction zones dominate in the upper parts of descending slabs, whereas indicate variation in movement both with lateral dis- tensional stresses are more important in the central and tance along descending slabs and with slab depth. Sea- lower parts (Figure 1.9). At 300-350 km depth in many ward from the trench in the upper part of the lithosphere slabs compressional stresses are very small, whereas at where the plate begins to bend, shallow extensional depths > 400 km a region of compressional stress may mechanisms predominate. Because of their low strength, be bounded both below and above by tensional stress sediments in oceanic trenches cannot transmit stresses, regions. and hence are usually flat-lying and undeformed. Seis- The seismicity in descending slabs is strongly cor- mic reflection profiles indicate, however, that rocks on related with the degree of coupling between the slab the landward side of trenches are intensely folded and and the overriding plate (Shimamoto, 1985). The low faulted. Thrusting mechanisms dominate at shallow seismicity in descending slabs at depths < 25 km may depths in subduction zones (20-100 km). At depths < 25 reflect relatively high water contents and the low strength km, descending slabs are characterized by low seismicity. of subducted hydrous minerals, both of which lead to Large-magnitude earthquakes are generally thrust-types decoupling of the plates and largely ductile deformation. and occur at depths > 30 km (Shimamoto, 1985). Dur- At greater depths, diminishing water and hydrous min- ing large earthquakes, ruptures branch off the slab and eral contents (due to slab devolatilization) result in greater 10 Plate Tectonics and Crustal Evolution T V T V T V Figure 1.8 Vertical cross-sections of * 0 t * ! »" 1 1 1 I 1 1 hypocentre distributions beneath 100 '^W^, 1 modem arc-trench systems. Each Vl*. 1 diagram shows earthquakes for 7-10 ^ 200 year periods between 1954 and 1969. E T = trench axis; V = recently active i 300 volcanic chain. Distance is measured Aleutian Islands horizontally from each trench axis in H 400 - (I74'W ITT^W) - - Northern ** kilometres. Hypocentre data from many Mariana sources, principally from National § 500 Arc Southern :''..* * -. - I Earthquake Information Center, US 600.:... Izu-Bonin Arc J Coast and Geodetic Survey. a b c 1 7nr> 1 1 1 1 1_ L _l 1 i».J 1 1 1 1 1 1 100 200 300 0 100 200 300 0 100 200 300 400 500 600 D S T A N C E ( knn ) V T V 0 ^i.M I 100...-.y. 200 t->:- New Hebrides Arc J 300 X 400 h- Q. Northern Izu-Bonin Arc UJ 500 600 700 100 200 300 400 500 600 0 100 200 300 400 500 600 700 D I S T A N C E (km) V T tively low dip of the descending slab (Figure 1.10). In A 0 1 1 \ ^ ^ 1 other areas, such as the Kurile and Mariana arc systems (Figure 1.8), slabs are largely decoupled from overrid- ing plates and extend to great depths, and earthquakes /&jtm^^ 200 j^/S^^ 1 are smaller and less frequent. In some instances, sub- J^j^m/y /Kr^O^ ducting slabs are forced beneath the lithosphere in the E / y/W^ overriding plate, a situation known as buoyant sub- ^ ^ 400 //^f0^ duction (Figure 1.10). Buoyant subduction occurs when /^t/wW^ 1 the lithosphere is forced to sink before it becomes nega- X /j(^J^W^ h- '^^^m^ tively buoyant (i.e., in < 50 My today), and thus it tends Q. / y^^r to resist subduction into the asthenosphere. Underplated LJ / /^^^ Q 600 ^V ^ 1 buoyant slabs eventually sink into the mantle when they cool sufficiently and their density increases. n i l Tension Seismic tomographic studies of subduction zones re- O Connpression J 800 veal a detailed three-dimensional structure of descend- ing slabs. For instance, P-wave tomographic images of \ \ \ 1 the Japan subduction zone correlate well with major 300 200 100 0 100 200 surface geological features in Japan (Figure 1.11). Seis- DISTANCE (km) mic velocities are several percentage points higher in Figure 1.9 Calculated distribution of stresses in a the descending slab than in the surrounding mantle, and descending slab. Subduction rate, 8 cm/y and dip = 45 °. results indicate that the slab boundary is a sharp seismic After Goto et al. (1985). V, volcanic front; T, trench. discontinuity (Zhao et al., 1992). Moreover, low- velocity anomalies occur in the crust and mantle wedge over the descending slab, a feature also common in other coupling of overriding and descending plates, and thus subduction zones. The two low-velocity zones in the to the onset of major earthquakes. Coupling also varies crust correlate with active volcanism in the Japan arc, between descending slabs. In some continental-margin and probably reflect magma plumbing systems. Deeper subduction zones (e.g., Peru-Chile, Alaska), coupling is low-velocity zones (> 30 km) may represent partly-melted very strong, resulting in large earthquakes and a rela- ultramafic rocks, formed in response to the upward trans-  Plate tectonics 11 Figure 1.10 Buoyant subduction beneath the Peru-Chile arc in central Peru. Earthquake first-motions shown by arrows 50.>^ / are tensional. Black dots are earthquake hypocentres. Modified after Sacks (1983). I ICX) X a 1501- 200 /\ 250 200 400 600 800 DISTANCE (km) 142 are the Tehuantepec, Cocos, Carnegie, Nazca and Juan Fernandez ridges along the Middle American and P e r u - Chile subduction systems. When a plateau or ridge en- counters an arc, subduction stops and a volcanic/seismic gap forms in the arc. In most instances, plateaux/ridges accrete to arcs, and only small ridges and some volcanic islands are negatively buoyant and can actually be sub- ducted. As we shall see in Chapter 5, this may be an im- portant mechanism by which continents grow laterally. X H A commonly asked question is, just where and how Q. LU are new convergent boundaries initiated? Because of the Q very high stress levels necessary for the oceanic litho- sphere to rupture, it is likely that pre-existing zones of weakness in the lithosphere provide sites for new sub- duction zones. Of the three proposed sites for initiation of new subduction zones, i.e., passive continental mar- 200 gins, transform faults/fracture zones, and extinct ocean ridges, none can simply convert to subduction zones by the affect of gravitational forces alone (Mueller and Figure 1.11 Cross-section of the Japan subduction zone Phillips, 1991). Hence, additional forces are needed to showing perturbations of P-wave velocities from normal convert these sites into subduction zones. One possible mantle (in percentages). Crosses and circles show fast and source is the attempted subduction of buoyant material slow velocities respectively. Solid triangles are active (such as a submarine plateau) at a trench, which can volcanoes. Base of crust shown at 25-30 km. Units on result in large compressional forces in both subducting horizontal axis are degrees of E longitude. After Zhao et al. and overriding plates. This is the only recognized tectonic (1992). force sufficient to trigger nucleation of a new subduction zone. Transform faults and fracture zones are likely sites fer of volatiles from the descending plate, which lowers for subduction initiation in that they are common in the the melting points of mantle-wedge silicates. vicinity of modem subduction zones and are weaker than Another interesting question about subduction is, what normal oceanic lithosphere. happens when a submarine plateau or aseismic ridge encounters a subduction zone? Because they resist sub- duction they may produce a cusp in the arc system, as Collisional boundaries illustrated for instance by the intersection of the Caroline Deformation fronts associated with collisional bounda- ridge with the Mariana arc south of Japan (Figure 1.1/ ries are widespread, as exemplified by the India-Asia Plate 1). Paleomagnetic and structural geologic data from boundary which extends for at least 3000 km northeast the Mariana arc support this interpretation, indicating of the Himalayas. Earthquakes are chiefly < 100 km deep that the arc was rotated at its ends by collision of these and first-motion studies indicate a variety of fault types. ridges between 30 and 10 Ma (McCabe, 1984). Also, Thrust fault mechanisms generally dominate near sutures, volcanic and seismic gaps in arc systems commonly occur such as the Indus suture in the Himalayas. Transcurrent at points of collision between submarine plateaux faulting is common in the overriding plate as illustrated and ridges with arcs (McGeary et al., 1985). Examples by the large strike-slip faults produced in China and 12 Plate Tectonics and Crustal Evolution Tibet during the India collision. In addition, extensional 1987). Also, the landward slope of the trench steepens faulting may extend great distances beyond the suture in and narrows in the collision zone. At the trench-ridge the overriding plate. For instance, the Baikal rift in south- collision site, the accretionary prism is reduced in size em Siberia appears to have formed in response to the by more than seventy-five per cent and part of the base- India collision 55 Ma. ment beneath the Andean arc appears to have been eroded A plate boundary in the early stages of an arc- away and subducted. Two factors seem important in continent collision is illustrated by the Sunda arc system decreasing the volume of the accretionary prism: in eastern Indonesia (Figure 1.1/Plate 1). Australia is 1 the topographic relief on the ridge may increase the beginning to collide with this arc as the Australian plate rate of subduction erosion carrying material away is subducted beneath the arc. In fact, a large bend in the from the bottom of the accretionary prism descending slab beneath the island of Timor may be produced by subduction of continental crust. Numerous 2 subduction of oceanic ridges caused by transform hypocentres at 50-100 km depth are interpreted to re- faults may mechanically weaken the base of the flect the beginning of detachment of the descending slab accretionary prism, making it more susceptible to as continental crust resists further subduction. Farther to removal by subduction erosion. the east, Australia collided with the arc system and is Heat flow increases dramatically in the collision zone accreted and sutured to the arc on the northern side of and then decays after collision to typical arc values. New Guinea. Earthquakes and active volcanism in north- Also, an ophiolite (fragment of oceanic cmst) was em New Guinea are interpreted to reflect initiation of a tectonically emplaced in the Andean fore-arc region new subduction zone dipping to the south, in the oppo- during an earlier stage of the collision 3 Ma, supporting site direction of subduction prior to collision with the the idea, more fully discussed in Chapter 3, that ridge- Australian plate. trench collisions may be an important way in which ophiolites are emplaced. The intriguing question of what happens to the convective upcurrent beneath the Chile Trench-ridge interactions ridge as it is subducted remains poorly understood. It is interesting to consider what happens when an ocean ridge approaches and finally collides with a subduction zone, as the Chile and Juan de Fuca ridges are today The Wilson Cycle (Figure 1.1/Plate 1). If a ridge is subducted, the arc The opening and closing of an oceanic basin is known should move 'uphill' and become emergent as the ridge as a Wilson Cycle (Burke et al., 1976), named after crest approaches, and it should move 'downhill' and J. Tuzo Wilson who first described it in 1966. He pro- become submerged as the ridge passes down the posed that the opening and closing of a proto-Atlantic subduction zone (DeLong and Fox, 1977). Correspond- basin in the Paleozoic accounted for unexplained changes ing changes in sedimentation should accompany this in rock types, fossils, orogenies, and paleoclimates in emergence-submergence sequence of the arc. Ridge the Appalachian orogenic belt. A Wilson Cycle begins subduction may also lead to cessation of subduction- with the rupture of a continent along a rift system, such related magmatism as the hot ridge is subducted. This as the East African rift today, followed by the opening could be caused by reduced frictional heating in the of an ocean basin with passive continental margins on subduction zone or by progressive loss of volatiles from both sides (Figure 1.12, a and b). The oldest rocks on a descending slab as a ridge approaches. Also, the outer passive continental margins are continental rift assem- arc should undergo regional metamorphism as the hot blages. As the rift basin opens into a small ocean basin, ridge crest is subducted. All three of these phenomena as the Red Sea is today, cratonic sediments are depos- are recorded in the Aleutian arc and support the subducted ited along both of the retreating passive margins (b), and ridge model. Ridge subduction may also result in a abyssal sediments accumulate on the sea floor adjacent change in stress regime in the overriding plate from to these margins. Eventually, a large ocean basin such dominantly compressional to extensional, and in the as the Atlantic may develop from continued opening. opening of a back-arc basin (Uyeda and Miyashiro, When the new oceanic lithosphere becomes negatively 1974). The subduction of active ridges may lead to for- buoyant, subduction begins on one or both margins mation of new ridges in the descending plate at great and the ocean basin begins to close (c and d). Complete distances from the convergent margin. For instance, the closure of the basin results in a continent-continent rifting of Antarctica from Australia, which began almost collision, (e), such as occurred during the Permian when 50 Ma, coincided with the subduction of a ridge system Baltica collided with Siberia forming the Ural Moun- along the northern edge of the Australian-Antarctic plate. tains. During the collision, arc rocks and oceanic crust In the case of the Chile ridge, which is being subducted are thrust over passive-margin assemblages. The geo- today, there are few effects until the ridge arrives at the logic record indicates that the Wilson Cycle has occurred trench axis. As the ridge approaches the Chile trench, many times during the Phanerozoic. Because lithosphere the rift valley becomes filled with sediments and finally is weakened along collisional zones, rifting may open disappears beneath the toe of the trench (Cande et al., new ocean basins near older sutures, as evidenced by  Plate tectonics 13 Collisional The overall pattern of stresses in both the continental Orogen and oceanic lithosphere is consistent with the present distribution of plate motions as deduced from magnetic anomaly distributions on the sea floor. Examples of stress provinces in the United States are shown in Figure 1.13. Most of the central and eastern parts of the United States are characterized by compres- sive stresses, ranging from NW-SE along the Atlantic Coast to dominantly NE-SW in the mid-continent area. In contrast, much of the western United States is char- acterized by extensional and transcurrent stress patterns, although the Colorado Plateau, Pacific Northwest and the area around the San Andreas fault are dominated by compressive deformation. The abrupt transitions between stress provinces in the western United States imply lower \ \ crustal or uppermost mantle sources for the stresses. The Arc \ \ Passive Continental correlation of stress distribution and heat-flow distribu- C m. \ Sea Level \ Margin tion in this area indicates that widespread rifting in the Basin and Range province is linked to thermal processes in the mantle. On the other hand, the broad transitions between stress provinces in the central and eastern United States reflect deeper stresses at the base of the lithosphere, related perhaps to drag resistance of the North American plate and to compressive forces transmitted from the Mid-Atlantic ridge. / a VJ ^ ''/^'^'-'-''-'fo/ - ^— m':i&;^ Lithosptiere ^ ^ ^ ^ 5 ^ ^ o n t i n enta 1 Crust \ \^_ ^^ Plate motions Figure \.12 Idealized sequence of events in a Wilson Cycloid plate motion Cycle (beginning at the bottom). The motion of a plate on a sphere can be described in terms of a pole of rotation passing through the centre of the opening to the modem Atlantic basin approximately the sphere. Because all plates on the Earth are moving along the Ordovician lapetus suture. relative to each other, the angular difference between a given point on a plate to the pole of rotation of that plate generally changes with time. For this reason, the trajec- Stress distribution within plates tory of a point on one plate as observed from another plate cannot be described by a small circle around a Stress distributions in the lithosphere can be estimated fixed pole of rotation. Instead, the shape of a relative from geological observations, first-motion studies of motion path is that of a spherical cycloid (Cronin, 1991). earthquakes, and direct measurement of in-situ stress The trajectory of a point on a moving plate relative to (Zoback and Zoback, 1980). Stress provinces in the crust a point on another plate can be described if three vari- show similar stress orientations and earthquake magni- ables are known during the period of displacement be- tudes and have linear dimensions ranging from hundreds ing considered: to thousands of kilometres. Maximum compressive stresses for much of North America trend E-W to NE- 1 the position of the pole of rotation SW (Figure 1.13). In cratonic regions in South America 2 the direction of relative motion most stresses trend E-W to NW-SE. Western Europe 3 the magnitude of the angular velocity. is characterized by dominantly NW-SE compressive stresses, while much of Asia is more nearly N-S. Within An example of cycloid motion for a hypothetical plate the Australian plate, compressive stresses range from is given Figure 1.14. Plate 2 rotates around pole P2 with N-S in India to nearly E-W in Australia. Horizontal a given angular velocity. P2 appears to move with time stresses are variable in Africa but suggest a NW-SE relative to an observer on plate 1, tracing the path of a trend for maximum compressive stresses in West Africa small circle. When the motion of plate 2 is combined and an E-W trend for the minimum compressive stresses with the relative motion of P2, a reference point M on in East Africa. Except at plate boundaries, oceanic litho- plate 2 traces a figure of rotation around two axes known sphere is characterized by variable compressive stresses. as a spherical cycloid. 14 Plate Tectonics and Cmstal Evolution Figure 1.13 Stress provinces in the United States. Arrows are directions of least (outward-directed) or greatest (inward- directed) principal horizontal compression. Province abbreviations: SA, San Andreas; SBR, southern Basin and Range province; RGR, Rio Grande rift; CP, Colorado Plateau; NRM, northern Rocky Mountains; SOP, southern Great Plains. From Zoback and Zoback (1980). magnetic anomalies are numbered beginning with 1 at ridge axes. From the Geomagnetic Time Scale, each anomaly is assigned an age (for instance anomaly 30 corresponds to an age of 72 Ma, and anomaly 7 to 28 Ma). If the spreading rate has been constant in one ocean basin, it is possible to extend the Geomagnetic Time Scale to more than 4.5 Ma using the magnetic anomaly pat- terns. Although data indicate that a constant spreading rate is unlikely in any ocean basin, the South Atlantic most closely approaches constancy (~ 1.9 cm/y) and is commonly chosen as a reference to extrapolate the time scale. Paleontologic dates from sediment cores retrieved by the Deep Sea Drilling Project and isotopic ages of basalts dredged or drilled from the ocean floor sub- stantiate an approximately constant spreading rate in the South Atlantic and allow the extension of the magnetic Figure 1.14 Spherical cycloid motion of plate 2 as its pole time scale to about 80 Ma. Correlations of magnetic of rotation P processes along a line that represents a small anomalies with distance from ridge axes indicate that circle around PI, the pole of rotation of plate 1. M is a spreading rates in the South Indian and North Pacific point on plate 2. Time progresses from t^ to tg. Modified basins have been more variable and, on the average, after Cronin (1987). faster than the spreading rate of the South Atlantic (Fig- ure 1.15). Plate Velocities in the last 150 My Plate velocities also can be estimated from disloca- tion theory, using data derived from first-motion studies Rates of plate motion can be quantified for the last 150 of earthquakes and from observed dip-lengths of sub- My by correlating seafioor magnetic anomalies with the duction zones, if these lengths are assumed to be a mea- Geomagnetic Time Scale (Figure 1.15). For convenience, sure of the amount of underthrusting during the last 10  Plate tectonics 15 Ridge Figure 1.15 Magnetic profiles from the Axis Atlantic, Indian, and Pacific Ocean basins. South Geomagnetic Time Scale given beneath the Atlantic ^ ^-^'y\/'^/^^^/yA^ profiles with normal (black) and reversed (white) magnetized bands. Proposed correlations of anomalies are shown with dashed lines. Numbers South refer to specific anomalies. Modified after Indian Heirtzler et al. (1968). North Pacific South Pacific I l i i i i i w i i i i ' i i i i i i i i m i i i i i i 11101 80 70 60 50 40 30 20 10 AGE (Ma) My. Still another method used to estimate plate veloci- reconstruction agrees well with geometric fits across the ties is by using transform faults. Rates and directions of North Atlantic. It is clear from the lines of motion that motion can be estimated from the azimuths and amount Europe and Africa have been on two different plates for of offset along transform faults, provided the azimuth the last 160 My. Changes in spreading rates and direc- and timing of motion can be estimated accurately. tions, however, occur on both the African and Eurasian Estimates of plate velocities commonly range within plates at the same time, at about 60 and 80 Ma. The a factor of two of one another using the above methods separation of Africa and North America occurred prima- averaged over several to many millions of years. Rates rily between 80 and 180 Ma, whereas separation of range from about 1 to 20 cm/y, averaging a few centi- Eurasia from North America occurred chiefly in the last metres per year for most plates. Typical velocities of 80 My. major plates (in cm/y) are as follows: North American, 1.5-2; Eurasian, 2.4-2.7; African, 2.9-3.2; Australian, 6-7.5; and Pacific, 5-7 (DeMets et al., 1990). Plate velocities from paleomagnetism From spreading rates estimated from seafloor mag- If true polar wander has been small compared with the netic anomalies, it is possible to contour the age of the rate of plate motions in the geologic past, it is possible sea floor and several such maps have been published. to estimate minimum plate velocities even before 200 From these maps, we see that the rate of spreading has Ma from plate motion rates (Bryan and Gordon, 1986; varied between crustal segments bounded by transform Jurdy et al., 1995). Also, if at least some hotspots have faults, and has even varied on opposite flanks of the remained relatively fixed, it is possible to estimate plate same ridge. The oldest oceanic crust (Jurassic) occurs motions relative to these hotspots (Hartnady and leRoex, immediately adjacent to the Izu-Bonin subduction zone 1985). Compared with typical modem continental plate south of Japan. Since the rate at which oceanic crust has velocities of 2-3 cm/y, during the last 350 My most been produced at ridges during the past several hundred continental plates have excursions to much faster rates million years is of the order of a few centimetres per (Figure 1.17). Episodes of rapid plate motion are re- year, it is unlikely that crust much older than Jurassic corded during the Triassic-Early Jurassic (250-200 Ma) will be found on the ocean floors today. The average on most continents, and Australia and India show peak age of oceanic crust is about 60 My and the average age velocities at about 150 and 50 Ma, respectively, after it begins to subduct is about 120 My. Fragments of they fragmented from Gondwana. Results indicate that oceanic crust older than Jurassic (ophiolites. Chapter 3) in the past continental plates have moved as fast as are found in continental orogenic belts where they were modem oceanic plates for intervals of 30—70 My. It is tectonically emplaced during orogeny. noteworthy that maxima in some continental plate vel- From calculated seafloor spreading directions and rates, ocities occur just after fragmentation from a super- it is possible to reconstruct plate positions and to esti- continent. For instance, the velocity maxima in the early mate rates of plate separation for the last 200 My. One Mesozoic follow the beginning of rifting in Pangea, and way of illustrating such reconstructions is by the use of the peak velocities for Australia and India follow separ- flow or drift lines, as shown in Figure 1.16 for the open- ation of these continents from Gondwana (Figure 1.17). ing of the North Atlantic. The arrows indicate the rela- Paleomagnetic data from Archean rocks in southem tive directions of movement. Earlier positions of Africa Africa suggest that plates were moving at comparatively and Europe relative to North America are also shown, slow rates of about 2 cm/y between 3.5 and 2.4 Ga in with the corresponding ages in millions of years. The this region, near the low end of the range of speeds of 16 Plate Tectonics and Crustal Evolution -90' -75' -60* Figure 1.16 Seafloor spreading reconstruction of the opening of the North Atlantic. Black continents are present positions; dates for earlier positions are indicated in millions of years ago. Arrows are flow lines. After Pitman and Talwani (1972). Phanerozoic plates (Kroner and Layer, 1992). How rep- to be determined as a function of time. GPS geodesy resentative this rate is of the Archean is not known, but uses several high-altitude satellites with orbital periods it is surprising in that a hotter Archean mantle would of twelve hours, and each satellite broadcasts its posi- seem to be more consistent with faster plate motions. tion and time. When multiple satellites are tracked, the location of the receiver can be estimated to within a few Space geodetic measurements of plate metres on the Earth's surface. Accuracy of the results depends on many factors (Gordon and Stein, 1992), velocities including the length of time over which measurements Space geodesy is measuring the precise position of sites have been accumulated. To reach accuracies of 1-2 mm on the Earth's surface from sources in space, such as for sites that are thousands of kilometres apart requires radio-wave sources and satellite tracking. Three meth- many years of data accumulation. ods are currently used: very long baseline radio inter- Space geodetic measurement are especially important ferometry (VLBI), satellite laser ranging (SRL), and the in a better understanding of modem plate tectonics. global positioning system (GPS). VLBI depends on the Results have been used to verify that plate motions are precise timing of radio-wave energy from extragalactic steady on time scales of a few years, to estimate rates sources (chiefly quasars) observed by radio telescopes. and directions of plate motions, to estimate motions of The radio waves recorded at different sites on the Earth small regions within plate boundary zones, to better are correlated and used to determine site locations, ori- understand deformation around plate boundaries, and to entation of the Earth, and azimuths of the radio sources. estimate rotations about a vertical axis of small crustal Arrival times of radio waves are measured with extremely blocks. Results are encouraging and indicate that plate precise hydrogen laser atomic clocks (Robertson, 1991). velocities averaged over a few years are similar to vel- SRL is based on the round-trip time of laser pulses re- ocities averaged over millions of years by the methods flected from satellites that orbit the Earth. Successive previously mentioned (Gordon and Stein, 1992; Smith observations permit the position of the tracking station et al., 1994). For instance, SRL velocities for the North  Plate tectonics 17 10 1 horizontal density contrasts resulting from cooling and thickening of the oceanic lithosphere as it moves away from ridges 2 the elevation of the ocean ridge above the surround- ^ 8 ing sea floor. £ The slab-pull forces in subduction zones reflect the cool- ing and negative buoyancy of the oceanic lithosphere as it ages. The gabbro-eclogite and other high-pressure phase transitions that occur in descending slabs also o contribute to slab-pull by increasing the density of the o slab. Ld > Using an analytical torque balance method, which accounts for interactions between plates by viscous cou- < pling to a convecting mantle, Lithgow-Bertelloni and -J Richards (1995) show that the slab-pull forces amount a. to about ninety-five per cent of the net driving forces of plates. Ridge push and drag forces at the base of the plates are no more than five per cent of the total. Com- North America- puter models using other approaches and assumptions CenozoicI Cret I Jur llriasl PermI Carb also seem to agree that slab-pull forces dominate (Vigny I et al., 1991; Carlson, 1995). Although slab-pull cannot 100 200 300 initiate subduction, once a slab begins to sink the slab- AGE (Ma) pull force rapidly becomes the dominant force for con- tinued subduction. Figure 1.17 Velocities (root mean square) of continental plates calculated from apparent polar wander paths for the last 300 My. Modified after Piper (1987). Geomagnetism American plate of 1.5-2 cm/y compare favourably with magnetic anomaly results averaged over 3 My. Both Rock magnetization VLBI and GPS data suggest that the motion of the To understand the magnetic evidence for seafloor spread- Pacific plate relative to the Eurasian and North American ing, it is necessary to understand how rocks become plates is about ten per cent faster than that estimated magnetized in the Earth's magnetic field. When a rock from magnetic anomaly data, suggesting that the Pacific forms, it may acquire a magnetization parallel to the plate has speeded up over the past few millions of years ambient magnetic field referred to as primary magneti- (Argus and Heflin, 1995). zation. Information about both the direction and inten- sity of the magnetic field in which a rock formed can be obtained by studying its primary magnetization. The most Plate driving forces important minerals controlling rock magnetization are magnetite and hematite. However, it is not always easy Although the question of what drives the Earth's plates to identify primary magnetization in that rocks often has stirred a lot of controversy in the past, we now seem acquire later magnetization known as secondary mag- to be converging on an answer. Most investigators agree netization, which must be removed by demagnetization that plate motions must be related to thermal convection techniques prior to measuring primary magnetization. in the mantle, although a generally accepted model re- Magnetization measured in the laboratory is called lating the two processes remains elusive. The shapes natural remanent magnetization or NRM. Rocks may and sizes of plates and their velocities exhibit large acquire NRM in several ways, of which only three are variations and do not show simple geometric relation- important in paleomagnetic studies (Bogue and Merrill, ships to convective flow patterns. Most computer mod- 1992; Dunlop, 1995): els, however, indicate that plates move in response chiefly to slab-pull forces as plates descend into the mantle 1 Thermal remanent magnetization (TRM). TRM at subduction zones, and that ocean-ridge push forces is acquired by igneous rocks as they cool through a or stresses transmitted from the asthenosphere to the blocking temperature for magnetization of the con- lithosphere are very small (Vigny et al., 1991; Lithgow- stituent magnetic mineral(s). This temperature, known Bertelloni and Richards, 1995). In effect, stress distribu- as the Curie temperature, ranges between 500 °C tions are consistent with the idea that at least oceanic and 600 °C for iron oxides, and is the temperature at plates are decoupled from underlying asthenosphere which magnetization is locked into the rock. The (Wiens and Stein, 1985). Ridge-push forces are caused direction of TRM is almost always parallel and pro- by two factors (Spence, 1987): portional in intensity to the applied magnetic field. 18 Plate Tectonics and Crustal Evolution 2 Detrital remanent magnetization (DRM). Clastic sediments generally contain small magnetic grains, which became aligned in the ambient magnetic field t cr I- QC X Z during deposition or during compaction and dia- z < ouj ^§8 UJ 3 _| 00 > O O O IDUJ genesis of clastic sediments. Such magnetization is < CD known as DRM. a. cor: -Laschamp 3 Chemical remanent magnetization (CRM). CRM - Blake is acquired by rocks during secondary processes if BRUNHES new magnetic minerals grow. It may be produced during weathering, alteration, or metamorphism. -0.73 -0.90 NRM is described by directional and intensity param- -0.97 Jaramillo 1.0 H eters. Directional parameters include declination, or the angle with respect to true north, and the inclination, or dip from the horizontal. A paleomagnetic pole can be MATUYAMA calculated from the declination and inclination deter- Olduvai mined from a given rock. 2.0 Reunion Reversals in the Earth's magnetic field Some rocks have acquired NRM in a direction opposite 2.92 to that of the Earth's present magnetic field. Such mag- 3.01 Kaena GAUSS 3.0 netization is known as reverse magnetization in con- [—3.05 Mammoth trast to normal magnetization which parallels the Earth's *"—3.1 5 present field. Experimental studies show that simultane- -3.40 ous crystallization of some Fe-Ti oxides with different -3.80 Cochiti Curie temperatures can cause these minerals to become —3.90 magnetized with an opposite polarity to the ambient field. 4.0 H —4.05 Nunivak This self-reversal magnetization is related to ordering —4.20 GILBERT —4.32 and disording of Fe and Ti atoms in the crystal lattice. Sidufjall — 4.47 Although self-reversal has occurred in some young lava flows, it does not appear to be a major cause of reverse — 4.85 magnetization in rocks. The strongest evidence for this Thvera 5.0 — 5.00 comes from correlation of reverse magnetization between different rock types from widely separated localities. For Figure J.18 The Geomagnetic Time Scale for the last 5 instance, reversed terrestrial lava flows correlate with My. Grey pattern, normal polarity; white, reversed polarity. reversed deep-sea sediments of the same age. It is clear that most reverse magnetization is acquired during periods of reverse polarity in the Earth's magnetic field. many polarity subchrons and can be dominantly normal One of the major discoveries in paleomagnetism is (e.g., the Brunhes), dominantly reversed (e.g., the that stratigraphic successions of volcanic rocks and deep- Matuyama), or mixed (Figure 1.18). Larger intervals sea sediment cores can be divided into sections that show (10^-10*^ y) with few if any reversals are known as dominantly reverse and normal magnetizations. Polar- superchrons. Based on the distribution of oceanic mag- ity intervals are defined as segments of time in which netic anomalies, it is possible to extrapolate the Geo- the magnetic field is dominantly reversed or dominantly magnetic Time Scale to more than 100 Ma. Independent normal. Using magnetic data from volcanic rocks and testing of this extrapolation from dated basalts indicates deep-sea sediments, the Geomagnetic Time Scale was the predicted time scale is correct to within a few per- formulated (Cox, 1969), extending to about 5 Ma (Fig- centage points to at least 10 Ma. Results suggest that ure 1.18). Although polarity intervals of short duration over the last 80 My the average length of polarity (< 50 000 years) cannot be resolved with K-Ar dating subchrons has decreased with time. Reversals in the of volcanic rocks, they can be dated by other methods Earth's field are documented throughout the Phanerozoic, in deep-sea sediments, which contain a continuous (or although the Geomagnetic Time Scale cannot be continu- nearly continuous) record of the Earth's magnetic his- ously extrapolated beyond about 200 Ma, the age of the tory for the last 100-200 My. The last reversal in the oldest oceanic crust. Reversals, however, have been magnetic field occurred about 20 ka (the Laschamp). indentified in rocks as old as 3.5 Ga. Two types of polarity intervals are defined on the The percentage of normal and reverse magnetization basis of their average duration: a polarity event or for any increment of time has also varied with time. subchron (10^-10^ y) and a polarity epoch or chron The Mesozoic is characterized by dominantly normal (10^-10^ y). A polarity chron may contain several-to- polarities while the Paleozoic is chiefly reversed (Figure  Plate tectonics 19 100 Figure 1.19 Distribution of CEN I CRET I JUR I T R I I P E R I CARB |DEV|SIL| ORD I CAM magnetic reversals during the Number of Studies Phanerozoic averaged over 50 My intervals. Also shown are the Cretaceous (CN) and Permian- Carboniferous (PCR) superchrons. Modified after Piper (1987). 1.19). Periodic variations are suggested by the data at about 300, 110 and 60 Ma (Irving and Pullaiah, 1976). Statistical analysis of reversals in the magnetic field indicate a strong periodicity at about 30 My. Two major superchrons are identified in the last 350 My. These are the Cretaceous normal (CN) and Permian-Carboniferous reversed (PCR) superchrons (Figure 1.19). Statistical analysis of the youngest and best-defined part of the Geomagnetic Time Scale (< 185 Ma) shows an almost linear decrease in the frequency of reversals to the Cre- taceous, reaching zero in the CN superchron. The inver- sion frequency appears to have reached a maximun about 10 Ma and has been declining to the present. Causes of changes in reversal frequency are generally attributed to changes in the relief and/or electrical conductivity along the core-mantle boundary. Both of these parameters are 100 100 km temperature-dependent and require long-term cyclical mmm^1 1 _ J changes in the temperature at the base of the mantle. o o This, in turn, implies that heat transfer from the lower e to E o Q> o mantle is episodic. A possible source of episodic heat v_ 0) to to >> 3 £1 c > 3% to V) tl> £^ o o ^ 13 loss fr