T52A-01 INVITED
Structural Control on the Megathrust Slip: the Example of the Ecuador-Colombia Active Margin.
Along subduction zones, earthquake nucleation and megathrust slip are controlled by parameters including the rheology of the fore-arc, the magnitude of transient shear stress and the physical properties of the megathrust. Furthermore, geological structures play a major role on stress and strain distribution both across and along the megathrust, and consequently affect the earthquake cycle. A marine geophysical study of the Ecuador-SW Colombia subduction zone, where three large (7.7<Mw<8.2) thrust earthquakes occurred in 1942, 1958 and 1979 following the Mw 8.8, 1906 megathrust event, reveals a correlation between a multi-scale crustal segmentation of the margin, and the spatial distribution and extent of the earthquake rupture zones. On a 500 to >1000-km scale, the subduction trench is segmented from south to north, into NS-, NNE- and NS-trending segments, respectively associated with normal, oblique, and normal plate convergence settings. These segments relate to large-scale, structural domains of both the Nazca Plate (Carnegie Ridge, Panama Basin) and the margin of the South American Plate, which consists of a mosaic of accreted oceanic terranes. The 1906 event likely ruptured the entire NNE-trending, obliquely convergent segment. The slip was blocked southward by the buoyant subducting Carnegie Ridge, and northward by the sharp change in structural trend associated with the accreted Choco arc. On a 100-300-km scale, multichannel seismic reflection and multibeam bathymetry data show that the margin is segmented by transverse crustal faults. The faults correlate with the limits of large earthquake rupture zones, suggesting that transverse faults are weak and contribute to placing a limit on the along strike propagation of the co-seismic slip. On a 20-100-km scale, geophysical data collected on the Nazca Plate reveal oceanic asperities entering the subduction. Downdip, these features may correlate with seismological asperities, which, upon rupture, can trigger large events. On a ~50 km-scale, a crustal splay fault and the underlying updip segment of the plate interface may have respectively controlled the updip coseismic and postseismic slips of the M7.7 1958 earthquake. The splay fault is interpreted as an inverted, landward dipping normal listric fault that developed initially in the oceanic plateau that makes up the margin. Reactivation and inversion of the fault occurred when subduction erosion put the fault into contact with the megathrust. The splay fault separates inner and outer margin wedges. The wedges behave differently during the earthquake cycle, according to their differing velocity structures, and the slip weakening and slip hardening nature of their respective underlying interplate fault segments. It is suggested that during co-seismic slip along the splay fault, the elastic rebound of the inner wedge load the outer wedge with elastic stress. The resulting accumulated strain is likely to be released during the post-seismic phase, by creeping along the upper segment of the megathrust. Combination of the three scales of margin segmentation generates a complex pattern of stress distribution at the plate interface, and shows the importance of geologic structures on the megathrust slip.
T52A-02
Comparison of Holocene With Coseismic Vertical Deformation Accompanying the Great 1 April 2007 Solomon Islands Megathrust Rupture
The 1 April 2007 Mw 8.1 earthquake in the western Solomons arc is the first major seismic rupture of this segment of plate boundary in historical times. A remarkable property of this region is the existence of coral- fringed islands located in a belt from ~90 km to as close as ~5 km from the trench. This setting provides a unique opportunity in which to measure forearc vertical movements using corals and other data to reveal relationships among coseismic vertical displacement, extremely rapid uplift of the outer forearc, and slower uplift of the main volcanic arc. The location of maximum coseismic uplift along a trench-parallel belt adjacent to the trench is consistent with the trench-parallel belt of maximum Holocene uplift rates. However, islands along the main volcanic arc lie in the swath of coseismic subsidence located arcward and parallel to the uplift zone. These islands typically have mean Holocene uplift rates up to ~1mm/yr. Thus, coseismic uplift correlates with rapid outer forearc Holocene uplift, but coseismic subsidence occurred throughout the more slowly uplifting volcanic arc. Using these observations, we can deconvolve and isolate the components of co- seismic, interseismic, and net vertical deformation and seek to address the underlying mechanisms. Interpretation is complicated by the late Quaternary history that includes subsidence of both the inner and outer arc islands prior to initiation of the ongoing net uplift since ~50 ka.
T52A-03
Massive sediment underthrusting beneath the Nankai Trough splay fault inferred from 3D seismic reflection data
In April-May 2006, we acquired a 3D seismic reflection data volume across the Nankai Trough and Kumano forearc basin to image the mega-splay fault system that lies along this segment of the Nankai Trough subduction zone. These data were acquired by Petroleum GeoServices (PGS) using 4 streamers, each 4,500 m long, and a 3090 m3 source, and cover an area of 12 x 56 km2. The data were processed by Compagnie Générale de Géophysique (CGG) to suppress seafloor multiples, and to migrate and stack the data using prestack time migration. The mega-splay fault system appears as a high-amplitude seismic reflection with polarity opposite that of the seafloor that lies 1-2 km above the top of the subducting ocean crust. It is a regionally significant reflection that extends from 8 km subseafloor, 50 km landward of the trench, up dip to the seaward edge of the Kumano Basin where it branches into 3-4 individual near-surface faults. Along strike the mega-splay reflection extends across the 12 km width of the 3D volume. We have analyzed this fault-plane reflection waveform in detail across the entire 3D survey area. The reflection from the deep branch of the mega-splay fault is nearly everywhere consistent with the properties of a simple single reflection interface with lower velocity and density below the splay fault rather than a thin fault-zone layer of locally low-velocity/density material. There appears to be a direct correlation between the amplitude of the splay fault and the amplitude of the top of the subducting ocean crust. We interpret this correlation to as being due to the low-velocities and low-densities within the entire underthust interval between the splay fault and ocean crust and not just within a thin interval beneath the splay fault. These results imply the entire 1-2 km section beneath the splay fault has been thrust beneath the overriding section. This massive sediment underthrusting carries significant quantities of fluid-rich sediment into the up-dip portion of the seismogenic zone, where it can affect fault zone behavior.
T52A-04
The Distribution of Interseismic Locking on the Central Cascadia Subduction Zone Inferred From Coastal Uplift Rates in Oregon
We invert for the distribution of locking along the Oregon portion of the Cascadia subduction zone using an updated dataset of the interseismic vertical displacement rates. Uplift rates are inferred from spirit leveling that is tied into an absolute vertical reference frame using tide gauge records. With absolute uncertainties less than 1 mm/yr, the data provide good constraints on the accumulation of interseismic strain along the Oregon coast. The slip rate deficit on the subduction interface is modeled using a backslip calculation in an elastic half-space. Static Green's functions are calculated using a triangular fault model that approximates the 3D geometry of the plate interface. We assume a slip rate deficit that is equivalent to the full convergence rate in an offshore locked zone and tapers to zero at depth across a transition zone. The convergence rate is calculated using the published Euler vector for motion of the Juan de Fuca oceanic plate relative the Oregon continental forearc. To minimize the number of free parameters in the inversion, the down-dip extent of locking is defined by the lower edge of the fully locked zone and the lower edge of the transition zone. These two free parameters are optimized at various latitudes by minimizing the misfit of the east-west leveling lines through a grid search of the parameter space. The north-south leveling route that runs along the coast is then used to further optimize the model by interpolating the slip deficit distribution along-strike. The inversion prefers models where the locked zone is forced up-dip (i.e. farther offshore) at the east-west profile near Newport on the central Oregon coast. This is in contrast to inversion results near Astoria where the locked and transition zones extend farther down-dip relative to a constant depth contour on the plate interface. Our optimal model of the slip-rate deficit distribution is compared to previously published models of strain accumulation constrained by only horizontal displacement rates. We also compare the distribution of locking to the extent of Siletzia in the continental forearc, gravity lows, and forearc basins to better understand the potential location of long-lived asperities on the subduction interface.
T52A-05
Subduction Dynamics and Great Earthquakes
The majority of Earth's seismic energy has been released by a minority class of great earthquakes, with magnitudes (Mw) greater than 8.5, causing unprecedented damage in many portions of the world. Most of these events have occurred along the megathrust faults of subduction zones, with orogenic tectonics considered a prerequisite. The 2004 Sumatra earthquake (Mw =9.1) occurred in a region exhibiting an extensional back-arc basin, spurring a re-evaluation of global great earthquake risk. Here we define a new dataset of maximum earthquake magnitude for 256 global subduction zone segments, showing that the majority of great earthquakes occur in neutral tectonic regions, with lower normal stresses and fewer asperities than previously argued. In addition, all great subduction zone earthquake epicenters since 1900 have occurred within 25 percent of the total slab length closest to a terminal edge. We argue that pathways for mantle flow around a slab induce larger maximum earthquake magnitudes near the terminal edges of slabs by decreasing seismic coupling and increasing a trench's capacity to produce long rupture lengths.
T52A-06
Forearc Rigidity and Rupture During Great Earthquakes
Mapping along-strike variations in subduction zone rigidity is useful for understanding great megathrust earthquakes. This is because: 1) the integrated long-term strength of the coupled slab-forearc system, which is reflected in rigidity, should be an important control on earthquake frequency and magnitude distribution; and 2) forearc rigidity is likely to play a role in the generation of great earthquakes through its control on the degree to which the overriding plate can absorb strain accumulated during convergence. Motivated by these factors, we have calculated lithospheric rigidity along circum-Pacific subduction zones using a wavelet-based Bouguer coherence technique. Whilst our results show a general correlation between oceanic plate-age and rigidity -- as expected -- intriguing along-strike rigidity variations are evident. The most interesting along-strike rigidity variations exist in the subduction zones subjected to the four largest recorded earthquakes: the Valdivia (1960, M9.5), Sumatra--Andaman (2004, Mw 9.3), Alaska (1964, 9.2) and Kamchatka (1952, 9.0) earthquakes. For all but the 2004 Sumatra-Andaman earthquake, the extensive rupture associated with these earthquakes tends to initiate in a region of higher rigidity and propagate into regions of lower rigidity. This suggests that rupture dominantly occurs in weakened parts of the plate interface that slip when triggered by rupture in an adjacent area where substantial interseismic strain has accumulated. At least for the Chilean subduction zone, this weakness appears to be linked to the thick sediments that fill the trench. In contrast to the 1960 Valdivia event, maximum slip during the 2004 Sumatra--Andaman earthquake correlates with the most rigid parts of the forearc. In this case, slip is highest where the most inter-seismic strain has accumulated, perhaps where the plate interface is not lubricated by subducting sediments. These results highlight the potential influence of trench sediments on plate coupling in subduction zones, but the influence of other factors (e.g. wedge geometry, forearc elastic properties, thermally induced oceanic plate rigidity variations) also need to be considered.
T52A-07 INVITED
Interplay of Structure and Sediment Supply May Influence Subduction Zone Rupture Patches and Propagation
The question of whether there are universal controls on the genesis and maintenance of large slip and moment patches along strike on subduction megathrusts has proved remarkably elusive, in part due to the short temporal records we have of these great events around the globe. Many events this century are poorly constrained, and many subduction zones only have one or a few events available for comparison. Long historical records and good structural constraints have made Nankai a leading case for basin centered asperities, yet the recent Sumatra Mw 9.2 rupture models show that slip and moment for the most part avoided basins and was centered under structural highs. In Cascadia, both deformation and tsunami models clearly fit the respective subsidence and runup data better if slip in past events was centered under or did not avoid these highs as opposed to basin centered model. Onshore and offshore paleoseismic evidence from 38 Cascadia earthquakes strongly suggest that structural segmentation plays a role only along the southernmost margin. These data do not provide information on moment or slip distribution, but do effectively constrain rupture lengths. Rupture lengths constrained by the paleoseismic data show that there is no Holocene segmentation for the northern margin, and that southern segments may be controlled by some of the obvious structural boundaries such as the Blanco Fracture zone, and outer arc uplifts and forearc basins. Where resolution is adequate, these data also suggest that ruptures die out into the basins and are linked multi-segment ruptures of structural uplifts, similar to that observed in the 2004 and 2005 earthquakes from Sumatra where outer arc uplifts may mark segment boundaries, high slip patches and initiation points for great earthquakes. The difference between the rupture modes observed for Nankai and Sumatra, and suggested here for Cascadia may be linked to the sediment supply for these systems. Cascadia and Sumatra are both systems where massive submarine fans are accreting to the margin in their northern regions, with incoming sections of 3-4 km thickness that taper southward. These thick sections promote high fluid pressure, but also tend to smooth the plate interface with respect to structures in both the downgoing and upper plates. A smooth plate interface has long been thought to promote long ruptures and high moment release, and so we suspect that northern Cascadia and northern Sumatra may be prone to large ruptures due to the masking of other structures by large influxes of sediment on the subducting plate. By comparison, the relatively thin sediment supply at Nankai may allow these structural boundaries to play a greater role in rupture propagation and moment release. The smaller southern Cascadia ruptures are also consistent with this model, with structural control taking precedence as the sediment supply thins southward. http://www.activetectonics.coas.oregonstate.edu/