9 various kineticprocesses duringsubductionP. van Keken, 2004
10 What do we want to understand .. What is the flow pattern in the wedge mantle?Temperature distribution (how hot is the corner?)2-D laminar flow versus 3-D flow involving trench parallel component?Do subducting slabs contain a large amount of water (serpentine)?What is the distribution of water in the wedge mantle?Is the wedge mantle “wet” throughout, or is it “wet” only in limited regions? (Comparison to the continental tectosphere.)Does basalt -> eclogite transformation occur at equilibrium condition?Do dehydration reactions cause earthquakes?could dehydration reactions at high-P (V<0) cause instability?open „todo“ list, MARGINS workshop, Ann Arbor (2002)
16 Complex plate boundary zone in South-East Asia Northward motion of India deforms all of the regionMany small plates (microplates) and blocksEruption Mt. Pinatubo, 2001Sumatra Earthquake, December 26, 2004Molnar & Tapponier, 1977
17 Plate tectonics - potential hazards (II) Tsunami waves
18 subductionthrust fault earthquake December 26, 2004subductionthrust fault earthquake
19 EARTHQUAKE (COSEISMIC): INTERSEISMIC:India subducts beneath Burma microplate at about 50 mm/yr(precise rate hard to infer given complex geometry)Fault interface is lockedEARTHQUAKE (COSEISMIC):Fault interface slips, overriding plate rebounds, releasing accumulated motionStein & Wysession, 2003Fault slipped ~ 10 m = mm~ takes mm / 50 mm/yr = 200 yrLonger if some slip is aseismicFaults aren’t exactly periodicfor reasons we don’t understandSumatra Earthquake, December 26, 2004HOW OFTEN ?
25 3 components of earthquake hazard at SZ (1) Large interplate thrust (rare, but paleoseismology & tsunami history from Japan find big one in 1700): largest earthquakes but further away(2) Intraslab (Juan de Fuca) earthquakes: smaller but closer to population(3) Overriding (North American) plate: smaller but closer to population
26 Deep Earthquakes Triggered mainshocks Triggering mainshocks Earthquakes and subducted slabs beneath the Tonga-Fiji area (yellow marker series, orange marker series)
27 Subduction one plate descends below another, oceanic crust is consumed understanding of subduction process completedformation of theory of plate tectonicsprovided mechanism for removing oceanic crustgenerated at mid-ocean ridges
28 how was subduction “discovered”? “Wadati-Benioff” zones: zones of dipping earthquakes to100’s kms depth (max: ~670 km)deepshallowintermediateseismicity
29 plate tectonics: convergent boundaries Wadati-Benioffzonenorthern Japanepicentershypocentersred dots are deepest earthquakesso they plot on map as farthestfrom trench
30 variations in dips of Wadati-Benioff zones plate tectonics: convergent boundariesvariations in dips of Wadati-Benioff zones
31 plate tectonics: convergent boundaries “imaging” the subducting plate with seismic velocities- subducting plate is cooler than surrounding mantle -slowfastfast:cooler(denser material)slow:hotter(less dense material)
32 3 types of convergence plate tectonics: convergent boundaries less buoyant plate dives below more buoyant plateoceanic lithosphere density > continental lithosphere3 types of convergence• ocean-ocean convergence• ocean-continent convergence• continent-continent convergence (collision)
33 (1) ocean-ocean convergence • one oceanic plate subducts below another• earthquakes occur along interface between two plates• trench, accretionary wedge, forearc basin, volcanic arc
34 (1) ocean-ocean convergence • trench: deep, narrow valley where oceanic plate subducts• accretionary wedge: sediments that accumulated onsubducting plate as it traveled from ridge are scrapedoff and accreted (added) to overriding plate
35 (1) ocean-ocean convergence • forearc basin: between accretionary wedge and volcanic arc• volcanic arc: mantle is perturbed by subduction process andmelts at depths of km, creating magmathat rises to the surface to form island volcanoes
37 (2) ocean-continent convergence • oceanic plate subducts below less dense continental crust• features same as with ocean-ocean convergence except thatvolcanoes are built on continental crust and in some casesa backarc thrust belt may form
38 (2) ocean-continent convergence • volcanoes (magmatic arc): more silicic from addition ofcontinental material; batholiths form at depth• backarc thrust belt: thrust faults form behind arc inresponse to convergence; “stickiness” between platesAndes;Cascades
39 arc-trench gap distance between the trench and volcanoes because the depth at whichmagmas are generatedin subduction zonesis about km,this distance dependson the dip of thesubducting plateif the dip of the subducting plateis flat enough, no volcanoes formsubducted plate doesn’t go deep…infer dip by looking at distancebetween volcanoes and trench
40 trench can migrate through time response to forcing either by overriding or subducting platesubductingplatesteepensand pullsoverridingtowardtrenchoverridingplatepushestrench
41 (3) continent-continent convergence neither plate wants to subduct(both are buoyant)result iscontinental collision• mountain belts• thrust faults• “detached” subducting plate• suture zone - plate boundary
42 (3) continent-continent convergence model for India and Asia collision
44 (3) continent-continent convergence deformation fromcollision extendsfar into Tibet/Asia
45 what causes plates to move ? ridge push: sea floor spreading and gravitysliding of plate downhill from ridge to trenchwhile being pushed by sea floor spreading
46 what causes plates to move ? slab pull: weight of subducting slabsubducting slab sinks into mantlefrom its own weight, pulling therest of the plate with itas subducting slab descendsinto mantle, the higherpressures cause minerals totransform to denser forms(crystal structures compact)
47 what causes plates to move ? slab pull is more important than ridge pushHow do we know ? - Plates that have the greatest length ofsubduction boundary have the fastest velocities
48 what causes plates to move ? slab pull is more important than ridge pushForsyth & Uyeda, 1975How do we know ? - Plates that have the greatest length ofsubduction boundary have the fastest velocities
49 what causes plates to move ? mantle convection is the likely candidate,but is it the cause or an effectof ridge push and slab pull ?
50 How Mantle Slabs Drive Plate Motions C.P. Conrad and C. Lithgow-Bertelloni"How mantle slabs drive plate tectonics"Science, 298, , 2002
51 Observed plate motions Observed plate motions. Arrow lengths and colors show velocity relative to theaverage velocity. Note that subducting plates (Pacific, Nazca, Cocos, Philippine,Indian-Australian plates in the center of this Pacific-centered view) move about 4times faster than non-subducting plates (North and South American, Eurasian,African, Antarctic plates around the periphery).
52 How Mantle Slabs Drive Plate Motions bending forces
53 Diagram showing the mantle flow associated with the "slab suction" plate-driving mechanism in which the sinking slab is detached from the subductingPlate and sinks under its own weight. This induces mantle flow that drives both theoverriding and subducting plates toward each other at approximately equal rate.
54 Predicted plate velocities for the "slab suction" plate-driving model. Note that subducting and non-subducting plates travel at approximatelythe same speed, which is not what is observed (compare to Fig. 1).
55 The "slab pull" plate-driving mechanism The "slab pull" plate-driving mechanism. Here the slab pulls directly on thesubducting plate, drawing it rapidly toward the subduction zone.The mantle flow induced by this motion tends to drive the overriding plate awayfrom the subduction zone. This results in an asymmetrical pattern of plate motions.
56 Plate motions driven by the slab pull plate-driving mechanism. In this case, plates move with about the right relative speeds, but overridingplates move away from trenches, instead of toward them as is observed.
57 Preferred model for how mantle slabs drive plate motions Preferred model for how mantle slabs drive plate motions. Slabs in the uppermantle pull directly on surface plates driving their rapid motion toward subductionzones. Slab descending in the lower mantle induce mantle flow patterns that excitethe slab suction mechanism. This flow tends to push both overriding andsubducting plates toward subduction zones.
58 Predicted plate motions from our combined model of slab suction from lower mantle slabs and slab pull from upper mantle slabs (Fig. 6).This model predicts both the relative speeds of subducting and overriding plates,as well as the approximate direction of plate motions(compare to observed plate motions, shown in Fig. 1).
59 Thermal-mechanical structure a more detailed quantitative understanding of subduction zonesThermal-mechanical structureof subduction zones
60 Wadati & Benioff zones Some earthquakes appear to result from flexural bendingof the downgoing plateas it enters the trench.Focal depth studies show a pattern of normal faulting in the upper part of the plate to a depth of 25 km, and thrusting in its lower part, between km.These constrain the neutral surfacedividing the mechanically stronglithosphere into upper extensionaland lower compressional zones.Bodine et al., JGR 86 (1981)
62 Simple thermal slab model (McKenzie, 1969) Thermal modeling predicts maximum depth of isotherms in slab varies with thermal parameter” f ”
63 Thermal modeling predicts maximum depth of isotherms in slab varies with thermal parameter Deepest earthquakes never exceed ~700 kmMaximum depth increases with Earthquakes below 300 km occur only for slabs with > 5000 kmKirby et al., 1996
64 deep earthquakes stop at 660 km, perhaps because: Transition zone between upper & lower mantles bounded by 410 km and 660 km discontinuitiescorresponding tomineral phase changesdeep earthquakes stop at 660 km, perhaps because:- slabs equilibrate thermally- slabs cannot penetrate 660 km- earthquakes are relatedto phase changesRingwood, 1979
65 Seismicity decreases to minimum ~300 km, and then increases again Deep earthquakes below ~ 300 kmtreated as distinct from intermediate earthquakes with depths kmDeep earthquakes peak at about 600 km, and then decline to an apparent limit at ~ km
66 Slabs are not thermally equilibrated with mantle Coldest portion reaches only ~ half mantle temperature in about 10 Myr, about the time required for the slab to reach 660 km.Thus restriction of seismicity to depths < 660 km does not indicate that the slab is no longer a discrete thermal and mechanical entity.From thermal standpoint, there is no reason for slabs not to penetrate into lower mantle.When a slab descends through lower mantle at the same rate (it probably slows due to the more viscous lower mantle), it retains a significant thermal anomaly at the core-mantle boundary, consistent with models of that regionStein & Stein, 1996
67 “SLAB PULL” plate driving force Thermal modeling gives a driving force for subduction due to the integrated negative buoyancy (sinking) of cold dense slab from density contrast between it and the warmer and less dense material at same depth outside. Negative buoyancy is associated with the cold downgoing limb of mantle convection pattern.Since the driving force depends on thermal density contrast, it increases for(i) Higher v, faster subducting & hence colder plate(ii) Higher L, thicker and older & hence colder plateExpression is similar to that for “ridge push” since both forces are thermal buoyancy forces
68 “SLAB PULL” plate driving force Significance for stresses in slabs and for driving plate motions depends on their magnitude relative to resisting forces at the subduction zone:As slabs sink into the viscous mantle, displacement of mantle material causes force depending on the viscosity of mantle and slab subduction rate.Slabs are also subject to drag forces on their sides and resistance at the interface between overriding and downgoing plates, which are frequently manifested as earthquakes.
69 Forces within subducting plates (I) (1) Average absolute velocity of plates increases with the fraction of their area attached to downgoing slabs, suggesting that slabs are a major determinant of plate velocities(2) Earthquakes in old oceanic lithosphere have thrust mechanisms showing deviatoric compressionForsyth and Uyeda, 1975
70 Forces within subducting plates (II) The “slab pull'' force is balanced by local resistive forces, a combination of the effects of viscous mantle and the interface between plates. This situation is like an object dropped in a viscous fluid, which is accelerated by its negative buoyancy until it reaches a terminal velocity determined by its density and shape, and the viscosity and density of the fluid.Forsyth and Uyeda, 1975, Wiens & Stein, 1984
71 Forces within subducting plates (III) Different stresses result if weight of column of material supported in different wayssimilar to what seismic focal mechanisms show !Stein & Wysession, Blackwell 2003
72 Clapeyron slope describes how mineral phase changes occur at different depths in cold slabs use thermal model to find dT, phase relations to find and thus dPconvert to depth change dz
73 Opposite deflection of mineral phase boundaries Because spinel is denser than olivine, V < 0. This reaction is exothermic (gives off heat) so H < 0 is also negative, causing a positive Clapeyron slope. The slab is colder than the ambient mantle (T<0 ), so this phase change occurs at a lower pressure (P<0), corresponding to shallower depthIn contrast, the ringwoodite ( spinel phase) to perovoskite plus magnesiowustite transition, thought to give rise to the 660 km discontinuity, is endothermic (absorbs heat) so H > 0. Because this is a transformation to denser phases (V < 0), Clapeyron slope is negative, and the 660 km discontinuity should be deeper in slabs than outsideUpward deflection of the 410 km and downward deflection of the 660 km discontinuities have been observed in travel time studies.
74 Metastable delay of mineral phase transformations Kirby et al., Rev. Geophys. 1996
75 Metastable delay of mineral phase transformations Predicted mineral phase boundaries and resulting buoyancy forces in slab with and without metastable olivine wedgeFor equilibrium mineralogy cold slab has negative thermal buoyancy, negative compositional buoyancy from elevated 410 km discontinuity, and positive compositional buoyancy from depressed 660 km discontinuityMetastable wedge gives positive compositional buoyancy and decreases force driving subductionStein & Rubie, Science 1999negative buoyancy favours subduction, whereas positive buoyancy opposes it.
76 Deep earthquakes from metastable olivine ? Kirby et al., Rev. Geophys. 1996
77 Deep earthquakes due to large viscosity contrast between transition zone and lower mantle ? Predicted stress orientations are similar to those implied by focal mechanisms. Moreover, magnitude of the stress varies with depth in a fashion similar to the depth distribution of seismicity - minimum at km and increase from km.Vassiliou & Hager, Pageoph 128 (1988)
78 Intermediate depth earthquakes (I) Oceanic crust should undergo two important mineralogic transitions as it subducts. Hydrous (water-bearing) minerals formed at fractures and faults warm up and dehydrate. Gabbro transforms to eclogite, rock of same composition composed of denser minerals.Under equilibrium conditions, eclogite should form by the time slab reaches ~70 km depth. However, travel time studies in some slabs find low-velocity waveguide interpreted as subducting crust extending to deeper depths. Hence eclogite-forming reaction may be slowed in cold downgoing slabs, allowing gabbro to persist metastably.Kirby et al., Rev. Geophys. 1996
79 Support for this model comes from the fact that the intermediate Intermediate depth earthquakes (II)In this model intermediate earthquakes occur by slip on faults, but phase changes favor faulting. The extensional focal mechanisms may also reflect the phase change, which would produce extension in the subducting crust.Support for this model comes from the fact that the intermediateearthquakes occur below the island arc volcanoes, which are thought to result when water released from the subducting slab causes partial melting in the overlying asthenosphere.Kirby et al., Rev. Geophys. 1996
80 phases are no longer thermodynamically stable. Complex thermal structure, mineralogy & geometry of subducted slabs in the mantle transition zoneDeep subduction process is a chemical reactor that brings cold shallow minerals into temperature and pressure conditions of mantle transition zone where thesephases are no longer thermodynamically stable.Because there is no direct way of studying what is happening and what comes out, one seeks to understandthe system by studying earthquakes that somehow reflect what is happening.Kirby et al., 1996
81 ZusammenfassungDie Dynamik von Subduktionszonen ist gekennzeichnet durch die komplexe Wechselwirkung tektonischer, mineralogisch-petrologischer und geophysikalischer Prozesse auf verschiedensten Raum- und Zeitskalen.Diese hochgradig nichtlinear miteinander verbundenen Prozesse haben einen entscheidenden Einfluss auf den Lebensraum des Menschen (Vulkanismus,Erdbeben, Tsunamis). Ihr quantitatives Verständnis erfordert das Zusammenwirken von mineralogisch- petrologischen Untersuchungen, geophysikalischer Beobachtung und geodynamischer Modellierung.