T11D-01 INVITED
Convergent Margin Structure and a Unifying Concept
Marine observations of the past decade resolve 3 domains of different mechanics in space that probably respond differently from each other during an earthquake cycle. Accretion is common along thickly (>1 km) sedimented trenches and slowly (<50km/myr) converging margins. Erosion is common where convergence is greater which also reduces trench sediment thickness by rapid subduction. However erosion and accretion can be coeval, for instance, subducted seamounts erode the upper plate as adjacent sediment accretes. Trench sediment abundance appears to be a master control of tectonic erosion or accretion. Subducting plate relief and bending, fluid systems, input plate temperature, and material differences seem less important. From recent observations a unifying framework concept to aid interpretations of both accreting and eroding margins is proposed. Over a long term (Ma) the subduction channel accepts a finite amount of material. The excess amount will accrete and a shortage of trench sediment enhances erosion (Cloos and Shreve, 1988). If conditions remain consistent over ~1 Ma periods, the margin configuration becomes typically accretionary or erosional. In each margin segment the short term inter plate friction and material strength changes during the earthquake cycle as proposed by Wang and Hu, 2006. Mechanics probably changes locally during the cycle as well. K. Wang, Y. Hu, Accretionary prisms in subduction earthquake cycles: the theory of dynamic Coulomb wedge, J. Geophys. Res. 111 (2006) B06410, doi:10.1029/2005JB004094. Cloos, M., and R.L. Shreve, (1988), Subduction channel model of prism accretion, melange formation, sediment subduction, and subduction erosion at convergent plate margins: 2. Implications and discussion, Pageoph, v. 129, n. ¾ 501-545
T11D-02
Criteria for Seismic Splay Fault Activation During Subduction Earthquakes
As sediment is added to the accretionary prism or removed from the forearc, the material overlying the plate interface must deform to maintain a wedge structure. One of the ways this internal deformation is achieved is by slip on splay faults branching from the main detachment, which are possibly activated as part of a major seismic event. As a rupture propagates updip along the plate interface, it will reach a series of junctions between the shallowly dipping detachment and more steeply dipping splay faults. The amount and distribution of slip on these splay faults and the detachment determines the seafloor deformation and the tsunami waveform. Numerical studies by Kame et al. [JGR, 2003] of fault branching during dynamic slip-weakening rupture in 2D plane strain showed that branch activation depends on the initial stress state, rupture velocity at the branching junction, and branch angle. They found that for a constant initial stress state, with the maximum principal stress at shallow angles to the main fault, branch activation is favored on the compressional side of the fault for a range of branch angles. By extending the part of their work on modeling the branching behavior in the context of subduction zones, where critical taper wedge concepts suggest the angle that the principal stress makes with the main fault is shallow, but not horizontal, we hope to better understand the conditions for splay fault activation and the criteria for significant moment release on the splay. Our aim is to determine the range of initial stresses and relative frictional strengths of the detachment and splay fault that would result in seismic splay fault activation. In aid of that, we conduct similar dynamic rupture analyses to those of Kame et al., but use explicit finite element methods, and take fuller account of overall structure of the zone (rather than focusing just on the branching junction). Critical taper theory requires that the basal fault be weaker than the overlying material, so we build on previous work by incorporating the effect of strength contrasts between the basal and splay faults. The relative weakness of the basal fault is often attributed to high pore pressures, which lowers the effective normal stress and brings the basal fault closer to failure. We vary the initial stress state, while maintaining a constant principal stress orientation, to see how the closeness to failure affects the branching behavior for a variety of branch step-up angles.
T11D-03
Rapid Late Cenozoic Subsidence Along the Aleutian Forearc Identifies Nucleation Areas of Great Earthquakes and Transoceanic Tsunamis
The age, crustal structure, thickness of sediment, and dip of the subducting plate, and, in the upper plate, the formation of forearc basins, and splay and high-angle reverse faults, are clues to identifying the potential source regions of great subduction zone earthquakes and destructive tsunamis. The Aleutian subduction zone displays the interplay of these identifying characteristics. Two of three recorded great (i.e., Mw8 or higher) megathrust earthquakes nucleated in the area of the geomorphically prominent Atka Basin sector of the forearc's deeply submerged (3-5 km) Aleutian Terrace. During Mw8.7 and Mw8.0 earthquakes the highest slip (moment release) occurred beneath the Atka Basin area from which matching tsunamis were launched. Dredging and DSDP drilling establish that area of Atka Basin is a young, rapidly subsiding structural depression filled with 2-3 km of late Miocene and younger deposits. It is fronted by Hawley Ridge, a rapidly rising, outer forearc antiformal high. GPS studies (Cross and Freymueller, JGR, v. 113, 2008) document that the subduction zone is effectively locked beneath the deepening Atka Basin sector of the Aleutian Terrace and rising Hawley Ridge, but effectively unlocked to the east where the forearc basin of the terrace is inverted and large megathrust earthquakes have not been recorded. The deepening and inverting sectors of the Aleutian Terrace are separated by the subducting and westward migrating relief of the Amlia Fracture Zone. West of the Amlia FZ, the underthrusting plate is younger (~15 myr), dips less steeply, and is bathymetrically rougher than the Pacific plate subducting beneath the low seismicity sector of the Aleutian forearc east of the FZ. The presence of a young, actively subsiding forearc basin is an important locator of where high magnitude earthquakes nucleate and potentially launch destructive tsunamis. It is not clear why the Atka Basin area is the locus of rapid subsidence and great earthquakes. But the coincidence of low-angle underthrusting of a rough seafloor, focused interplate coupling and subduction erosion, and a structurally strong sector of arc massif, are likely involved. The rupture propagation necessary to generate high magnitude earthquakes is facilitated by the insertion of a thick (<1 km), laterally continuous section of trench sediment into the Aleutian subduction zone
T11D-04 INVITED
The Formation of Forearc Basins and Their Context in Active Margin Structure
We review the structural and stratigraphic styles of forearc basins and present a model for their formation, based on geological observations and quantified with mechanical and numerical models. We posit that forearc basins form as a predictable consequence of the deformation associated with the continental margin and the formation of an accretionary wedge. The geometry and surface form of an accretionary margin is controlled by the underlying slab and the internal wedge rheology and under specific, but common, conditions will develop closed basins. We identify three types of margins with characteristic forearc basins. In the first type, a growing accretionary wedge traps sediments between an outer-arc high and the volcanic arc. In this case, a filled basin prevents arcward growth of the accretionary wedge, limiting deformation to seaward of the outerarc high. In a second margin type, curvature of the slab with its associated decollement produce curvature of the overlying wedge surface, leading to an arcward-dipping wedge surface in the region proximal to the arc. In this case accommodation space is created between the crest of the curved wedge and the arc. A basin forms above the landward-dipping segment of the wedge, forming what we call a "negative- alpha" basin, referring to the sign of the surface slope, alpha. A negative-alpha basin may be contiguous with, or overlie an older basin of the first-style. Sedimentation in a negative-alpha basin serves to stabilize the underlying wedge, thereby preserving the basin, largely undeformed, other than at its margin with the active wedge. In the third margin style, two independent basins develop seaward of the arc, separated by an outer- arc high. The more trenchward basin is a negative-alpha basin as above; the arcward basin is similar to the first type basin. The intervening structural is a late feature, often splitting an earlier, wider negative-alpha basin. We hypothesize that this structure develops in response to deep viscous flow of the wedge and only occurs in old, large accretionary wedges. In all three margin styles, the forearc basins play an important role in margin deformation by changing the surface configuration of the wedge, and stabilizing or destabilizing segments of the margin wedge.
T11D-05
Deformations Associated With Large Interplate Earthquakes Along the Sumatra-Andaman Subduction Zone
Since the occurrence of the 2004 Sumatra-Andaman earthquake (Mw9.2), the Sumatra-Andaman Subduction zone has attracted geophysicists' attention. We have been carrying on CGPS observation in Thailand and Myanmar to detect postseismic deformation following this gigantic event. Since CGPS on land is not enough to clarify the detailed image of postseismic deformation, we also make InSAR analyses in Andaman and Phuket Islands. On September 12, 2007, another Mw8.4 event occurred SW off Sumatra. We report deformations observed with GPS and SAR including co- and postseismic deformation following this event. We have analyzed CGPS data up to the end of 2007 and detected postseismic displacements all over the Indochina peninsula. Phuket, which suffered from about 26cm coseismic displacement, has shifted by 26cm southwestward till July, 2007. Postseismic transient is clearly recognized and already exceeds coseismic movements at remote sites such as Bangkok and Chiang Mai in Thailand. We processed ALOS/PALSAR data in Andaman and Phuket islands. No remarkable deformation is found in Andaman and Phuket Islands, since the operation period of ALOS/PALSAR is not long enough and the wavelength of postseismic deformation may be much longer than the swath. We try to synthesize the postseismic displacement using a 3-D viscoelastic FEM model. Its results imply that viscoelastic relaxation in mantle with a typical mantle viscosity may play an important role for the observed postseismic transients except during the first six month. An extremely low viscosity is not required beneath the Andaman Sea, though this back arc is now actively opening. Coseismic motion following the 2007 Sumatra event is detected north of Benkgulu on the coast of southern Sumatra with InSAR. The largest LOS displacement of about 35cm is observed 100km NW of Bengkulu. Coseismic westward displacements of 3.5cm from the 2007 Sumatra event are also observed at Singapore, whose epicentral distance is about 700km, with CGPS. The observed fringe can be simulated by a plane fault model gently dipping northeastward with a 10m slip. On the other hand, the maximum of postseismic LOS displacement is shifted about 50km south from the coseismic maximum and significant fringes are more localized than the coseismic ones. The shift and localization of fringe are difficult to understand with an afterslip on deeper extension of coseismic fault plane. There is a slight disturbance in fringes along the Sumatran fault in the coseismic image, although it might be a topographic error. We will continue monitoring crustal deformations in the Sumatra-Andaman subduction zone with GPS and InSAR in order to reveal stress transfer.
T11D-06
Bivergent thrust wedges surrounding island arcs: Insights from observations and sandbox models of the northeastern Caribbean plate
Thrust belts develop on both sides of island arcs at several localities around the world, such as southern Indonesia, Vanuatu, Panama, and the northeastern Caribbean. In all cases, the overall vergence of the backarc thrust belt is opposite to that of the forearc thrust belt. For example, in the northeastern Caribbean, an inactive arc (Hispaniola and Puerto Rico) is bordered by a north-verging accretionary prism and the Puerto Rico trench on the north, and by the south-verging Muertos thrust belt and the Muertos trough on the south. There are three models to explain such bivergent thrusting: (1) Bivergent thrusting develops where a reversal of the polarity of subduction is underway and the backarc thrust system overlies an incipient subduction zone; (2) Compression of the backarc region due to trenchward traction, applied at the base of the overriding plate by the subduction process; and (3) The arc and both thrust systems constitute a bivergent thrust wedge, whose development is driven entirely by crustal-level forces applied at a single subduction zone. The third model implies that island arc bivergent thrusting is analogous to that which develops during continent-continent collisions. Observations of deformational features from the Muertos thrust belt together with inferences from regional geometry of island arcs and simple sandbox kinematic models, lead to the conclusion that such island arcs are best explained as crustal bivergent thrust wedges. Modeling suggests, in particular, that an imbricate thrust wedge in the backarc region develops only if the arc behaves as a relatively rigid block that can transmit compressive stresses to the backarc region. In such circumstances, the strike-slip component of oblique convergence is accommodated entirely in the forearc and arc, and the backarc is a frontal (dip-slip) thrust system. The rigid block behavior of the arc may be explained by its mafic composition.
T11D-07 INVITED
Earthquakes, Upper Plate Structure and Subducted Seamounts in the Central Cascadia "Locked" Zone
In 2004, two moderate-size (Mw 4.8 and 4.9) low-angle thrust earthquakes occurred in the nominally "locked" zone of the central Cascadia plate boundary. Using secondary phases identified by comparing the earthquake data to regional active source seismic data, Trehu et al. (2008) argued that these events occurred on the plate boundary. Since 2003, at least 40 smaller events have occurred in clusters associated with these two earthquakes (see poster by Williams et al. in this session). The central segment of the Cascadia subduction zone also has an unusually broad coupling transition zone (McCaffrey et al., 2007) and ETS episodes that are longer and occur less regularly (Szelinga et al., 200; Brudzinski and Allen, 2007, pers. comm.) compared to segments to the north and south. Since, September 2007, we have been operating a local network of seismic stations, COLZA (Central Oregon Locked Zone Array), which includes short period and broadband OBSs from the Ocean Bottom Seismograph Instrument Pool and broadband instruments from the USArray Flex-Array, to better understand the relationship between crustal structure, deformation and seismic activity in the region. Here we present preliminary results from COLZA, which confirm that earthquakes in these "locked zone" clusters are shallower than reported by the land-based networks. We also discuss correlations between earthquake hypocenters and upper and lower plate crustal structure. The seismic activity, as well as seafloor topography and upper- plate structure, appear to be related to seamounts on the subducting Juan de Fuca plate, as determined from active source seismic and potential field data. The seamounts, which result in a rough plate boundary, may have formed in the wake of a propagating rift. Brudzinski and Allen, Geology, 2007; McCaffrey et al., Geophys. J. Int., 2007; Szelinga et al., J. Geophys. Res., 2008; Trehu et al., Geology, 2008.
T11D-08
Lithospheric Scale Deformation in Mega-thrust Subduction Zones
Although the general plate tectonic model of subduction zone deformation and its relationship to the earthquake cycle for mega-thrust earthquakes is well known, there is neither consistency in such descriptions nor compatibility among seismological, geodetic, and geologic frameworks for such events. In particular in most seismologic studies of mega-thrust earthquakes there is an implicit assumption that the co-seismic slip is essentially symmetric across the fault surface – that is both the upper and lower plates moved equal amounts (but in opposite directions) during the rupture. Implicit in many geologic studies along convergent margins is the assumption that most permanent deformation is within the upper plate and the subducting slab basically transits the seismogenic zone with little permanent deformation. This perspective serves as the framework for many animations of subduction zone tectonics. Two subduction zone locales, the Kuriles and Solomon Islands, that have hosted recent Mw 8+ earthquakes demonstrate two end-member styles of subduction zone processes neither consistent with the conventional view. The November 2006 (thrust) and January 2007 (normal) earthquake pair in the Kuriles provide an opportunity to quantify the deformation within the subducting Pacific slab during the interseismic period. Based on the correspondence in slip during these events, we are able to both estimate the deformation (dominantly in the subducting slab and not in the overriding plate) and place a constraint on the static frictional strength of the megathrust interface of approximately 2-5 MPa. The 2007 Solomon Island Mw 8+ earthquake shows a distinctly different pattern of interseismic deformation. During this event, the propagating rupture traversed an active transform plate boundary between the separately subducting Australia and Solomon Sea plates. We interpret this to represent a situation in which interseismic deformation is primarily in the upper (Pacific) plate allowing the rupture to jump the fundamental barrier of a plate boundary. This is also compatible with limited GPS data available for the Australia plate near the trench indicating unimpeded subduction of Australia and thus little internal deformation of the subducting slab. These two subduction regimes indicate that there is likely a full continuum in how deformation is accommodated during subduction, and implies that attempts to determine the megathrust (and associated tsunami) potential of subduction zones using observations of upper-plate deformation is problematic.