T13F-01
Distributed Coseismic and Early Postseismic Dip-Slip from the 1 April 2007 Solomon Islands Earthquake: A Unique Image of Near-Trench Rupture
We estimate the spatial distribution of dip-slip in the 1 April 2007 magnitude MW=8.1 Solomon earthquake, which created a locally large tsunami with runup heights up to 12 m. The event is unique in that involved the rupture of at least two subducting plates, and that land occurs very close to the trench on the hanging wall side. The occurrence of islands extremely proximal to the trench allowed for the collection of near-shore uplift and subsidence information from costal areas (including the exposure and subsidence of corals), hence giving a unique well-resolved image of the near-trench geodetically derived slip. Two surveys, taken between 1 week and 1 month after the event primarily across the southern portion of the slip zone, comprise a dataset of approximately 100 measurements of between +3.6 and -1.5 m of vertical displacements [Fritz and Kalligeris, 2008; Taylor et al., 2008]. We use the Okada [1992] elastic dislocation model, to explore the distribution of dip-slip on discrete patches. To maintain a realistic distribution of slip we smooth the solution by attempting to minimize the second-order spatial derivative of slip, hence minimizing the stress change across the system. Because data are only vertical in nature and the expected strike-slip component of the thrust is small, only the dip-slip component of rupture was considered. Early results show highly variable dip-slip both along-strike and down- dip, with a significant focus of slip in the shallow near trench area. If real, this near-trench focusing may explain the locally high runup on portions of Simbo Island. Because it is not certain how much of the modeled slip occurred due to coseismic versus post-seismic recovery and afterslip, we explore the variability of solutions between the two surveys and compare results with the available spatial distribution of co-seismic finite-slip model of C. Ji [unpublished, 2007].
T13F-02
Rethinking the mass budgets of erosive subduction zones: Evidence for coupled outer forearc subsidence and inner forearc shortening
The Middle America Trench in Costa Rica and the Japan Trench in northeastern Honshu show coastal uplift related to a contractional fault system at the boundary between inner and outer forearc. Underthrusting of the outer forearc along such a system contributes directly to trench retreat and reduces the magnitude of subduction erosion required to produce observed outer-forearc subsidence. Moreover, the global association of outer forearc subsidence with inner forearc uplift along other non-accretionary convergent margins such as Peru and northern Chile suggests that shortening and uplift of the inner forearc may be a ubiquitous feature of such margins. In the Costa Rican example, the shortening occurs by thin-skinned thrusting, with a decollement near the basement-cover contact rooted in basement beneath the extinct volcanic arc. This thrust belt is characterized by at least five thrust slices that form a hinterland-dipping duplex centered directly inboard of the axis of the colliding Cocos Ridge. Minimum shortening estimates based on line length balancing are ~20-40 km in the culmination of the thrust belt. Pliocene marine mudstones in the rear of the thrust belt and plate tectonic reconstructions suggest thrust initiation more recent than 5 Ma and possibly as recent as 2 Ma. Combining total shortening values with estimates for initiation of thrusting yields a minimum shortening rate of 4-8 km/m.y, with rates potentially as fast as 20 km/m.y. With no appreciable accretion at the toe of the margin, such rates of shortening are large enough to account for typical estimates for long-term trench retreat rate based on subsidence of the upper slope (e.g. 1-5 km/m.y.). The inner forearc of the Japan trench is dominated by two massifs of metamorphic basement rocks intruded by Cretaceous granites- the Abakuma and Kitakami mountains. The Futaba fault zone, a fault that marks the eastern margin of the Abakuma massif, places Mesozoic basement units in contact with deformed Paleogene and Neogene strata. Folded strata at the southern end of the Futaba fault indicate a basement-involved reverse fault with an upward-widening shear zone in the cover strata. Trishear forward modeling indicates a steeply west-dipping (50-70 degrees) fault and the geometry of the Futaba structure, combined with the age of fault initiation based on the pre-growth to growth transition, leads to an estimate of shortening rate on the Futaba Fault of ~0.3-0.5 km/Ma, about 6-50 per cent of the estimated trench retreat rate from subsidence of the outer forearc. However, there are numerous other active reverse faults in the forearc and backarc that would contribute to the total shortening across the upper plate. Although the long-term subsidence history of the outer forearc along many non-accretionary convergent margins almost certainly requires a contribution from basal erosion, our results along two non-accretionary margins demonstrate that shortening of the upper plate cannot be ignored in the budget of subduction systems and must be accounted for in any mass balance of the subduction system.
T13F-03
Theory and observations of mechanically heterogeneous critical-taper wedges: the role of spatial variation in fluid pressure
It is easily shown that many well-imaged accretionary wedges are mechanically heterogeneous based on spatial variations in wedge taper. For example, the toes of active accretionary wedges such as the Nankai trough and Barbados display deceasing surface slope α away from the toe, with no associated variation in the detachment dip β. Likely sources of this heterogeneity in taper are spatial variations in fluid pressure, cohesion and fault strength. However progress in understanding the mechanics of these active accretionary wedges is challenging because of [1] the obvious in-situ observational difficulties, combined with [2] the substantial complexity of heterogeneous wedge theory, making it difficult to apply to natural wedges. Here we present progress on both fronts, emphasizing the role of heterogeneity in pore-fluid pressure. We show that the spatial gradients in pore-fluid pressure can be effectively constrained using seismic velocity data, as well as borehole data, using standard petroleum techniques. Furthermore, the gradients in fluid pressure can be approximated by a single easily observable parameter, the fluid-retention depth zFRD below which compaction is strongly diminished. The Hubbert-Rubey weakening (1- λ) is a simple function of fluid-retention depth and depth (1-λ)=(1-λh)[zFRD/z]. We find that the heterogeneous critical taper wedge theory of Dahlen (1990) can be recast in terms of this fluid-retention depth, which leads to more concise and easily applied critical-taper wedge equations. The key observable parameters in addition to wedge taper (α, β) are zFRD/H, where H is the depth to detachment, and d zFRD/d z. We illustrate the application of these results with several natural examples, including the Barbados accretionary wedge and the Brazos area of offshore Texas, which is an extensional critical-taper wedge with extensive data.
T13F-04
Upper Plate Deformation in Response to Aseismic Ridge Subduction along a Convergent Margin - the Piano Player Model: Osa Peninsula, Costa Rica
The Osa Peninsula, an outer forearc high ~20 km inboard of the Middle America Trench, is deforming in response to short wavelength variations in the bathymetry on the subducting aseismic Cocos Ridge, an elongate region of thickened crust ( up to 40% thicker) created by motion of the Cocos plate across the Galapagos Hotspot. Plate convergence is nearly orthogonal to the trench at ~90 mm/yr and the plate interface occurs at a depth ~5 km under the peninsula. Relief on the Cocos Ridge locally exceeds 1 km with the dominant topography expressed as two nearly parallel, but locally offset ridges separated by an axial graben. The strike of these features is sub-parallel to the convergence vector. Modern topography of the Osa Peninsula, elevation of the basement rocks (Early to Middle Tertiary Osa Mélange), elevations of late Quaternary marine deposits, and distribution of late Quaternary deformation rates directly mirror the bathymetry on the Cocos Ridge just outboard of the MAT. New geologic mapping, radiometric dating and fission track analysis constrain distribution and rates of deformation on the Osa Peninsula. The Osa Peninsula is fragmented into a complex set of blocks that vary in size from several kms on a side to <10 kms on a side. These blocks, which closely match the size, distribution and shape of bathymetric features on the incoming Cocos Ridge, are bounded by trench parallel and trench perpendicular, high angle, normal and reverse faults that extend to the plate interface, allowing for grossly different deformation histories over short distances. Fission track analyses of 4 sandstone samples from the Osa Mélange suggest that the basement rocks reached maximum burial temperatures of 60-80 ° C indicating burial depths of ~3-4 kms assuming a reasonable geothermal gradient of ~20 °/km. This suggests a very thin margin wedge prior to late Neogene unroofing. Rates of late Quaternary deformation are constrained by over 30 radiocarbon and 5 Optically Stimulated Luminescence (OSL) ages from sections of the shallow marine Armuelles Fm. Deposition and subsequent block deformation of the Armuelles Fm are fundamentally controlled by interaction of eustatic sea level and bathymetry on the subducted Cocos Ridge. Sea level highstands (MIS 1, 3, and 5) allow for deposition of the shallow marine to estuarine Jimenez, Marenco and Rincón members while sea level lowstands allow for subaerial erosion and the development of internal unconformities. Deformation rates for individual blocks range from > 15 mm/yr directly inboard of subducting ridges to < -6 mm/yr directly inboard of axial grabens. These new results imply a model in which the thin, mechanically weak, outer margin, characterized by pervasive, penetrative, brittle deformation of the Osa Melange basement rocks, deforms directly in response to short wavelength, high-relief bathymetric features on the down going plate. In this case bathymetry dominates over basal traction in controlling deformation of the upper plate. Surface uplift or subsidence of the Osa Peninsula, is not significantly driven by shortening within the outer margin nor underplating or subduction erosion at the plate interface, but by the variations in bathymetry, relief, of the subducting Cocos Ridge.
T13F-05
Deformation characteristics of 2-3 km buried Hota accretionary complex, central Japan
The lower to middle Miocene Hota accretionary complex is a unique example of on land accretionary complex, representing deformation and its physical/chemical properties of sediments just prior to entering the seismogenic realm. The maximum paleotemperature was estimated approximately 55-70°C (based on vitrinite reflectance) indicative of a maximum burial depth about 2-3 km assuming a paleo-geothermal gradient as 25-35°C/km. Accretionary complex in this temperature/depth range corresponds with an intermediate range between the core samples collected from the modern accretionary prism (e.g. Nankai, Barbados, and so on) and rocks in the ancient accretionary complexes on land. Deformation and physical/chemical properties of accretionary complex in this range were poorly understood because the appropriate samples have not been collected yet (scientific drilling has never reached there and most of ancient examples experienced the deeper burial depth and suffered thermal and physical overprinting). This presentation will treat the detailed structural and chemical analyses of the Hota accretionary complex to construct deformation properties of accretionary complex in its 2-3 km depth range and to discuss the interrelation between the early diagenesis (hydrocarbon/cations generation and sediment dewatering, etc.) and transition of the deformation properties. The deformation in this accretionary complex is characterized by two deformation styles: one is a few centimeter-scale phacoidal deformation representing clay minerals preferred orientation in the outer rim, whereas random fabric in the core, quite similar texture to the rocks in the present-Nankai décollement zone. The other is S-C style deformation (similar deformation to the mélanges in ancient accretionary complex on land) exhibiting block-in-matrix texture and quite intense clay minerals preferred orientation in the matrix, cutting the phacoidal deformation. Positive anomaly of illite/smectite ratio and vitrinite reflectance data (Ro) inside the latter S-C structure infers frictional heating during the deformation. Carbonate cement and calcite/dolomite-filled breccias above the S- C deformation and sandstone clasts representing hydro-fractured texture in the deformation are indicative that calcareous cement and pore-fluid pressure fluctuations seemed to be an essential control of the transition of the deformation style and position of the S-C style deformation.
T13F-06 INVITED
Regional conductivity structure of Cascadia from 3D inversion of USArray magnetotelluric data
Magnetotelluric (MT) data are being acquired in a series of temporary arrays deployed across the continental US through EMScope, a part of the USArray component of EarthScope. Initial deployments in 2006 and 2007 acquired data at 110 sites covering the US Pacific Northwest. The MT sites, distributed with the same nominal spacing as the USArray seismic transportable array (~75 km), produced data in the period range 10- 20,000s of very good to excellent quality. The most striking and robust feature revealed by 3D inversion of this dataset is an extensive lower crustal conductor covering most of the study area southeast of a line running from the California border at the coast to the Blue Mountains of Northeastern Oregon. The conductance of this layer, which is about 15 km thick with a top at roughly 20 km depth, exceeds 3000 S beneath the he Northwest Basin and Range (BR) province of southeastern Oregon. The high conductivity in this region is inferred to result from fluids – including possibly partial melt at depth – associated with magmatic underplating and BR extension. The lower crust is much more resistive beneath the Coast Range, Willamette Valley and Puget Lowlands of Western Washington and Oregon, and beneath the Columbia Plateau. This area of resistive crust, which was derived from a large fragment of thickened oceanic lithosphere that was accreted to North America at approximately 48 Ma ("Siletzia"), is revealed by geological and geodetic studies to be strong, accommodating tectonic stresses through rigid block rotations. In contrast, the area to the southeast characterized by high conductivity in the lower crust is actively deforming, consistent with an important role for fluids in weakening of continental crust. The resistive Siletzia crust is broken by an elongated N-S zone of high conductivity beneath the Cascade volcanoes. High conductivities beneath the volcanoes also most likely reflect the presence of interconnected fluids, in this case released from the subducting Juan de Fuca plate. Significant variations in upper mantle conductivity are also revealed by the inversions, with the most conductive mantle beneath the northern part of the array in the backarc and the most resistive corresponding to subducting oceanic mantle.
T13F-07 INVITED
Volatile Outputs From Subduction-Related Magmatism in the Oregon Cascades Estimated From Melt Inclusions, Spring Discharges, Heat Flow Data and Geochronology
Estimates of volatile fluxes provide a primary test for models of magmatism and volatile cycling during subduction in the endmember "hot and dry" Cascadia subduction zone, which is caused by slow convergence (4 cm/a) of the young (~10-12 Ma) Juan de Fuca plate with Western North America. Intra- arc rifting in the Central Oregon segment of the Cascade arc during the past 2 Ma has caused this region to have the highest mafic output along the arc. However, estimates of major volatile (H2O, CO2, S, Cl) fluxes and comparisons with other arcs (e.g. Central America) are not straightforward because there are no passively degassing volcanoes in the area. We estimate volatile outputs for the Central Oregon Cascades by combining data for olivine-hosted melt inclusions with regional heat flow (e.g. Ingebritsen, 1989; Blackwell,1990) and geochronological (Sherrod and Smith, 1990) studies. These flux estimates can be compared with those obtained from spring water studies (e.g. James, 1999; Hurwitz, 2005). This multidisciplinary approach allows us to more accurately constrain volatile fluxes, given that uncertainties in all methods are large and difficult to evaluate. Reported fluxes for Central Oregon springs are 3.4E5 CO2 and 1.5E4 Cl kg/yr/km of arc (James, 1999; Hurwitz, 2005). Melt inclusion data indicate primitive basaltic magmas in the Central Oregon Cascades have 1.0-3.5 wt% H2O, 800-1900 ppm S, and 300-1100 ppm Cl. Assuming global arc magma CO2 contents of ~1 wt% (Wallace, 2005), we estimate H2O/CO2 (1.0-3.5), S/CO2 (0.08-0.19), and Cl/CO2 (0.03-0.11) in magmas, which when combined with spring CO2 estimates, yield an H2O flux of 0.34-1.2E6, a S flux of 2.6-6.5E4, and a Cl flux of 1.0-3.7E4 kg/yr/km of arc. Alternatively, by combining melt inclusion data with magma flux estimates (14-38 km3/Myr/km of arc; Ingebritsen et al. 1989; Sherrod and Smith 1990) we estimate volatile fluxes for H2O: 0.39-5.4E6; S: 0.39-3.9E5; and Cl: 0.16- 2.3E5 kg/yr/km of arc. Given the uncertainties involved, these are highly consistent with the estimates based on spring data. For comparison, Central Oregon S and CO2 fluxes are 6-31% and 13-80%, respectively, of the fluxes estimated for the Central American arc on a kg/yr/km of arc basis (Sadofsky, 2008; Hilton, 1992). Comparison of Central Oregon volatile outputs with slab inputs (Ito, 1983; Hilton, 1992; Jarrard, 2003) suggests low recycling efficiencies via magmatism for H2O (3-26%) and S (1-7%) and more variable recycling efficiencies for CO2 (9-55%) and Cl (9-87%). Low volatile recycling efficiencies via magmatism are consistent with both the high temperatures estimated for the subducted slab beneath the Cascades and the presence of a shallow reservoir for early devolatilized material in the serpentinized forearc mantle wedge. Low but non-zero recycling efficiencies could indicate that 1) slab devolatilization beneath the forearc is incomplete and/or 2) downdragging of the serpentinized forearc mantle by corner flow in the mantle wedge is significant in this hot arc setting.
T13F-08
Seismicity of the Oaxaca Segment of the Middle American Subduction Zone
Convergent plate boundaries generate potentially devastating great earthquakes when tectonic stresses accumulate on the plate interface. The Oaxaca segment of the Middle-America subduction zone offers an ideal opportunity for detailed studies of the plate interface due to its relatively rapid convergent rates, unusual shallow subduction angle, and ~50 km trench-to-coast distances that brings the seismogenic and transitional zones of the plate interface ~250 km inland. The short recurrence interval (decades) of megathrust events also allows us to compare current seismicity to past events in detail to examine the asperity and gap hypotheses to better characterize the seismic hazard. A network of seven broadband three- component seismometers was deployed in summer 2006 over an area of ~300 km west-east and 200 km north-south with nominal 80 km station spacing, providing the means to examine seismicity in detail for the first time in this region. We use the Antelope Software package to organize the first nine months of recorded waveforms, perform analyst event detections, generate source locations, and compute local magnitudes. We detected and located over 3000 earthquakes with this method. The bulk of the earthquakes follow the coastline with these hypocenters clustering near the plate interface, but we also detected a number of deeper (>40 km) intraslab earthquakes further inland. The microseismicity we detected outlines the down- dip end of the seismogenic zone near 25 km depth, consistent with the depth and inland extent of previous megathrust events. When the seismicity is compared with other recent studies, we find a clear spatial relationship that suggests a downward progression of deformation with the subducting plate from interplate seismicity to slow-slip events to non-volcanic tremor to intraslab earthquakes. We are also beginning to see temporal relationships between seismicity and episodic tremor and slip, including a swarm of 50 earthquakes at the down-dip end of the seismogenic zone with bursts of activity followed by pulses of non-volcanic tremor activity.