T51E-01 INVITED
Elastic and Permanent Deformation of Subduction Zone Forearc Systems Driven by Great Earthquakes
One of the most remarkable characteristics of forearc systems is the structural and morphological contrast between the actively deforming outer wedge and the much less actively deforming inner wedge which is often the host of forearc basins. The dynamic Coulomb wedge model relates this contrast to the process of megathrust earthquakes. According to this model, the outer wedge overlies the shallow aseismic portion of the megathrust, and the inner wedge overlies the seismogenic zone. At the time of a great earthquake, when the seismogenic zone experiences a stress drop, the aseismic shallow segment strengthens to resist seismic rupture. After an earthquake, when the seismogenic zone is fully locked, the shallow segment tends to weaken and relax, which is most directly reflected as post-seismic afterslip. It can be shown that the inner wedge undergoes mostly elastic deformation in earthquake cycles, but the outer wedge alternates between elastic and permanent deformation. The outer wedge stays in a stable state experiencing only elastic deformation when the basal stress is moderately low, but it enters a critical state experiencing permanent deformation when the basal stress increases to a threshold level. The most important mode of the stress increase along this shallow portion of the megathrust is coseismic strengthening. Therefore, at subduction zones hosting great earthquakes, it is the peak basal stress achieved in these earthquakes that controls the taper geometry of the outer wedge. The dynamic Coulomb wedge model offers some other testable predictions regarding the relation between megathrust slip and forearc structure. For example, there should be a tendency for the outer wedge taper to be larger where moment release in great earthquakes is persistently large. The degree of coseismic strengthening of the shallow megathrust segment controls the updip propagation of coseismic slip, affecting details of the wedge geometry. In the case of complete stress relaxation of the shallow megathrust segment during an interseismic period, the outer wedge may enter an extensionally critical state developing normal faults, opposite of the compressively critical state during great earthquakes.
T51E-02
Coseismic Strengthening of the Shallow Portion of the Subduction Fault and Effects on Frontal Prism Taper
Ample evidence suggests that the shallowest segment of subduction interface, down to a few km depth and updip of the megathrust seismogenic zone, exhibits a velocity-strengthening (aseismic) behavior. According to the dynamic Coulomb wedge model, it is mainly this coseismic strengthening that is responsible for the growth and permanent deformation of the overlying frontal prism (Wang and Hu, 2006). However, the degree of the strengthening as controlled by earthquake size and the location of the seismogenic zone has not been investigated. In this work, we use numerical models to study how the stress is coseismically transferred from the velocity-weakening seismogenic zone to the velocity-strengthening shallow segment. The model is a hybrid of the frictional contact model and the classic crack model. An earthquake is simulated using a decrease in the effective friction coefficient μ' (i.e., weakening) along the seismogenic zone, producing a few MPa stress drop. The simultaneous strengthening of the aseismic updip and downdip segments is simulated using an increase in their μ'. Only the net effect of the coseismic weakening/strengthening is modeled in this work; the dynamic evolution of the friction and post-seismic stress relaxation are not included. We demonstrate how the slip distribution along the subduction fault is controlled by the degree of coseismic strengthening of the updip segment. The minimum level of strengthening required to prevent the rupture from breaking the trench depends on the force drop along the seismogenic zone, defined as the product of the average shear stress drop along the seismogenic zone and its area. For a fault geometry similar to that of Nankai Trough with about 3 MPa stress drop along a seismogenic zone of 120 km downdip width, an increase of μ' by 0.055 along the 30 km wide updip segment will render the segment on the verge of breaking the trench. This level of stress increase may readily push the outer wedge into a critical state of failure. With a much higher degree of strengthening, the rupture will propagate into the shallow segment only slightly, causing localized compression in the area of slip termination. This may explain the formation of an outer ridge between the outer and inner wedges at some subduction zones. Model results also indicate that for the same force drop, an updip segment with narrower downdip width will require a higher degree of strengthening to prevent trench-breaking rupture. We measured the taper of frontal prisms at 24 subduction zones, and found that the surface slope is generally too high to be explained using the classical Coulomb wedge model but can be explained using the dynamic Coulomb wedge model including coseismic strengthening of the shallow portion of the megathrust. Wang, K., and Y. Hu (2006), Accretionary prisms in subduction earthquake cycles: The theory of dynamic Coulomb wedge, J. Geophys. Res., 111, doi:10.1029/2005JB004094.
T51E-03
Basin migration due to subduction erosion along the Japan Trench
Basin migration in the forearc basin along the Japan Trench was imaged by high-resolution seismic profiles and the migration was attributed to the migration of the zone of intense subduction erosion at the base of the forearc wedge. The Japan Trench is a classic example of subduction erosion. Studies of drill cores of DSDP Legs 56 and 57 showed that the forearc basin has been subsiding 3 to 4 km since the latest Oligocene time. These previous studies have not assumed uneven subsidence or basin migration through the 100 km wide forearc wedge. We obtained high-resolution seismic profiles of the forearc basins along the Japan Trench using GI gun and 48 channel digital streamer. The aim of the survey is to examine the relationship between geologic structure and a source area of the subduction zone earthquakes. The profiles show that the topographically flat forearc basin has narrower depocenters which have been migrating landward. The velocity of the migration is much slower than the convergent rate between the Pacific Plate and the NE Japan arc, indicating that the migration was not caused by passage of seamount or ridges on the subducting Pacific plate. These preliminary analyses support the idea that the subduction erosion occurs in limited areas which have been migrating landward. The basin migration of the forearc basin provides important information how subduction erosion proceeds.
T51E-04
Earthquakes and crustal structure beneath the central Cascadia continental margin
In the summer of 2004, two clusters of "repeating" earthquakes occurred beneath the continental shelf of the central Cascadia subduction zone near 44.5N, 124.5W where the subduction megathrust is thought to be locked or transitional. The largest event in each cluster reached moment magnitude M=4.8-4.9. Seismicity has continued since with small (M<3) earthquakes occurring in each cluster on August 23-25, 2007. Moment tensor analysis for the main shock in each cluster indicates a 6-15 degree eastward dipping fault plane, consistent with the plate boundary dip of ~12 degrees. One cluster is serendipitously occurring on a transect along which crustal structure is well known from active source seismic experiments, and raytracing through this crustal model to match observed relative arrival times of secondary phases indicates a source depth of 16 ± 1 km, within 1 km of the plate boundary. This segment of the forearc also displays several characteristics indicative of along-strike and down-dip variations in plate coupling including: a subducted ridge on the downgoing plate; a "bright spot" on the plate boundary at a depth of ~15-20 km; an along-strike change in the gravity field and basement depth; a transition in plate coupling indicated by inversion of GPS data; geologic indications of active folding in the upper plate; and anomalous deformation in the adjacent oceanic plate. On the other hand, no obvious correlation with ETS in this region is observed. In September, 2007, we deployed an array of ocean bottom seismometers to record microseismicity and distinguish among several possible models for the physical properties of the megathrust.
T51E-05
Earthquake Cycle Deformation and Forearc Sliver Translation in the Rupture Zone of the 1960 Chile Earthquake
We present GPS velocity solutions for the entire latitudinal range of the rupture zone of the great (Mw ~ 9.5) 1960 earthquake. Previous studies presented velocity solutions only for the northern half of this zone. The new measurements indicate that the pattern of opposing, roughly arc-normal, motions previously documented in the north also occur in the south. Interseismic shortening (eastward motion) is dominant near the coast, whereas postseismic transient deformation (westward motion) is dominant near the Chile-Argentina border. The new observations also reveal for the first time than in the southern part of the rupture zone a forearc sliver is travelling northward due to dextral motion across the Liquine-Ofqui fault system. This dextral shearing has a maximum rate of about 6.5 mm/yr near the southern end of the rupture zone, and declines to the north. We also observe a southwards decrease in the margin-normal velocities of the coastal area. We suggest that this decrease reflects a change either in the width or the frictional properties of the megathrust zone. The thermal structure of the megathrust zone varies with distance to the triple junction. We present a fairly simple 3D viscoelasatic finite element model that demonstrates that while our interpretation is not unique, it is broadly consistent with the GPS observations available to date.
T51E-06
Megathrust Slip and the Care and Feeding of the Subduction Channel Through which the Seismogenic Zone Runs
HABITATS OF GREAT OFFSHORE EARTHQUAKES: High-magnitude earthquakes (Mw = or >8.5) and trans- oceanic tsunamis commonly nucleate along subduction zones (SZ) bordered by laterally continuous, sediment- flooded trenches. Examples include: south-central Chile (1960 Mw=9.5), eastern Alaska (1964 Mw=9.2), Sumatra (2004, Mw=9.1), Cascadia (historic 1700 Mw=9.0), Colombia (1906 Mw=8.8), Sumatra (historic 1883, Mw=8.8), west-central Aleutian (1965 Mw=8.7), central Aleutian (1986, Mw=8.7), Sumatra (2005 Mw=8.6), and Nankai (historic 1707, Mw=8.5). In thickness, sediment entering these SZ ranges from 2 to 3 km and the column is axially continuous for more than 800 km. The depositional pile is typically the clastic beds of a trench-axis turbidite wedge and underlying fan and abyssal plain deposits that accrued seaward of the trench axis. Great rupture events also occur at subduction zones receiving little sediment, for example the Kamchatka (1952, Mw=9.0) and the north Chile SZs (historic 1868 Mw=8.9). Both SZs are areas of rapid upper plate thinning, subsidence, and truncation effected by subduction erosion. WORKINGS OF THE SUBDUCTION CHANNEL (SC): Beneath the submerged forearc, the SC functions to transport subducted ocean floor sediment and tectonically eroded forearc debris toward and into the mantle. The SC is the lowest structural unit containing upper plate crustal material. It hosts the seismogenic zone, which probably runs along the SC's upper boundary commonly referred to as the interplate decollement. A thick, laterally continuous SC structurally smoothes or simplifies the surface of the interplate decollement and sets up conditions for lengthy, high moment-release ruptures. Maximum slip is commonly concentrated beneath the thinned crust underlying forearc basins. These structures, in positive feed-back, are likely deepened co- seismically by high-slip-rate enhanced basal subduction erosion. The detached material lowers the effective stress on the decollement and further evens this interface. The channel also works tectonically to underplate the base of the inner margin and induce uplift and co-seismic activation of high-angle reverse faults. CONSEQUENCES OF WHAT IS FED SUBDUCTION ZONES: Ridges and high relief entering the SZ can act to arrest lateral rupturing. Supplying sedimentary and erosional debris to the subduction channel appears to act differently and favors the continuation of rupture, rapid slip beneath crustally thinned areas that can be translated upward at forearc splay faults to generate trans-oceanic tsunamis, and nearshore reverse-fault can spawn near- field tsunamis. The potential for great earthquake nucleation along thickly sediment SZs must be set high. Similarly, seismogenic risk for highly erosional SZ little perturbed by subducting relief must also be set high. Margins undergoing rapid tectonic erosion produce regional tsunamis but perhaps not trans-oceanic waves of great destructiveness.
T51E-07 INVITED
Slip in Great Megathrust Earthquakes and its Relation to Crustal Structure as Revealed by Satellite Free-air Gravity
In 2003, Song and Simons and Wells et al. showed that approximately 70% of the moment released during past large, shallow subduction zone thrust earthquakes occurred beneath trench-parallel, free-air gravity lows outlining the deep-sea slope terrace and its basins. The authors suggested that the basin-centered, fore-arc gravity lows might be good predictors of high seismic slip in future earthquakes. Since 2001, ten megathrust earthquakes have occurred with magnitudes greater than Mw 7.7, including the giant, Mw 9.17 Sumatra earthquake of 2004. These earthquakes provide a robust test of the idea that seismic slip is focused beneath basin-centered gravity lows, and also the related ideas that the landward maximum gravity gradient marks the effective down-dip limit of large coseismic slip, and that intrabasin, transverse gravity highs are areas of lower slip. A compilation of seismic and geodetic slip inversions for the post-2001 earthquakes and new analyses of slip for the great Antofagasta, Jalisco, and Peru events in 1995 and 1996 indicate that more than 80% of the high-slip areas occur beneath deep-sea terrace gravity lows (DSTL), and that half of the earthquake asperities lie beneath fore-arc basins or local gravity lows. The maximum gravity gradient along the landward margin of the deep-sea terrace may mark the point where thicker overlying crust and higher temperatures on the megathrust limit the down dip extent of stick-slip behavior. Onland analogues are the mountain front of the Himalaya, which approximately marks the down-dip limit of large coseismic slip along the Main Frontal Thrust, and the front of the Taiwan Central Ranges, which coincides with the limit of slip during the 1999 Chi-Chi earthquake (Mw 7.6). In the up dip direction, coseismic slip may be partitioned onto splay faults in the wedge, as occurred in the 1964 Alaska earthquake. The observed pattern of greater slip at depth beneath fore arc basins is consistent with partitioning of slip up dip, especially if outer wedge materials deform more slowly, as suggested for parts of the 2004 Sumatra rupture. Along strike variations in fore-arc gravity also correlate with changing seismic behavior. At Cape Erimo on Hokkaido, three Mw 8+ earthquakes (1952, 1968, 2003) have occurred on either side of the gravity high that overlies the Cape, with little coseismic slip beneath the high. To the northeast, the deep-sea terrace gradually narrows, as does the rupture width of the great earthquakes, until off the central Kurile Islands, the terrace disappears and the arc gravity high occupies the fore-arc. The gravity high had been an historic seismic gap that was filled by the 2006 Kurile Island earthquake (Mw 8.3). Although the earthquake nucleated under the high, the slip occurred beneath the adjacent gravity low to the northeast. This might suggest the gravity highs are not likely sources of large seismic moment, at least in M8 earthquakes. In contrast, the main asperity associated with the 2005 Sumatra (Mw 8.7) earthquake was beneath the large gravity high of Nias Island. An alternative view is that the gravity highs are stronger asperities that only rupture in giant earthquakes. Globally, the coincidence of basin- centered coseismic slip with geologic evidence of sustained subsidence of the fore-arc suggests that subduction erosion is occurring in the seismogenic zone. Recent work off Chile, Colombia, Peru, and elsewhere shows that subduction erosion is an important process in many subduction zones.