S14A-01
Lithospheric Seismic Velocity Structure of the Northern Tibetan Plateau: The ASCENT Seismic Experiment
The northeastern boundary of the Tibetan Plateau is a new focus of contemporary debate concerning continental plateau formation as an continent-continent collision. Recent geological studies and limited geophysical measurements in this region have been cited to argue that Asian continental lithosphere is being detached and 'subducted' into the Tibetan mantle. Upper mantle geophysical properties beneath the northern plateau have been attributed to delamination triggered by instability of a thickened lithosphere and mantle lid detachment and/or asthenospheric counterflow associated with subduction of Indian continental lithosphere. In order to test these various models for the evolution of the Tibetan lithosphere we have deployed a 65 station broadband seismic array throughout the northeastern portion of the Tibetan plateau. Shear wave splitting measurements are remarkably consistent with prior measurements of shear wave splitting from previous temporary broadband seismic arrays. In general we observe a clockwise rotation in the fast directions that is consistent with upper mantle flow around the eastern Syntaxis; however, we also observe evidence of more complex flow within regions of the northern Tibetan plateau. We find evidence of north south fast directions along the southeastern edge of the Qaidam basin whereas we find East-West fast directions that parallel the Kunlun fault system further to the west. We are also using Rayleigh phase velocities to map the three dimensional shear wave velocity structure of the upper mantle beneath the central and northern plateau. These measurements will also help constrain the depth distribution of the seismic anisotropy in order to determine whether we are measuring asthenospheric flow or lithospheric mantle deformation.
S14A-02
Along-Strike Variation in Subducting Plate Seismicity, Related to Fluid Release and Attenuation
We analyze seismicity in the subducted slab of the Hikurangi subduction zone, along the 600-km length of the North Island, New Zealand. The degree of plate coupling is geodetically well defined and changes sharply from weak in north to strong in south. The volcanic character also changes along strike with extremely productive rhyolitic volcanism in the central Taupo zone, moderate andesitic volcanism in the northern and south-central zones, and non-volcanic subduction in the southernmost Hikurangi zone. We have relocated slab earthquakes with 3-D velocity models. The relocated seismicity shows more detail and varied clustering compared to 1-D locations. In the region of weak coupling and little potential for large subduction thrust earthquakes, there is a moderately-low level of seismicity in the slab crust, paired with a high level of seismicity in the slab upper mantle at 30-50 km depth. In the region defined by strong geodetic locking, there is abundant crustal slab seismicity, but very little underlying upper mantle seismicity. The permeability of the overlying plate appears to exert a major influence on the plate coupling properties. Dehydration at shallower depth and lower temperature is promoted where the overlying plate is permeable with large fluid flux allowing escape of released water. Such large fluid flux appears related to weak coupling. Strong coupling is evident where the overlying plate is impermeable, which allows high fluid pressure to build-up at the plate interface and throughout the slab, restraining expulsion of any released water, and promoting crustal slab seismicity. At greater depth where the subducting slab interacts with the mantle wedge, the slab seismicity forms four bands trending along-strike and deepening towards the north. The character of these bands, from shallow to deep varies: (I) low seismicity, (II) abundant clustered seismicity with cluster length of about 150-km, (III) low seismicity, and (IV) deepest slab seismicity, 180-330 km depth, clustered in patches, about 70-km long, separated by low seismicity zones. The largest most numerous cluster in II, at 150-220 km depth, underlies the pronounced low Qp zone in the mantle wedge, which is associated with the Taupo silicic volcanic zone. Extensive slab dehydration contributes to low Qp. Similar relation between slab seismicity and low Qp is observed in Chile. There may be some feedback cycle such that the more fluid-rich slab zone enhances corner flow and brings released fluid rapidly away from the slab, and so promotes more rapid and extensive dehydration.
S14A-03
Slab imaging in continental subduction under the Northern Apennines.
The seismic structure of a lithospheric collision zone plays a key role in modeling and testing different tectonic hypotheses, such as the rate and style of subduction. The Northern Apennines orogen (NA) of central Italy is a complex example of a convergent boundary between two continental lithospheres. Deformation in this area is thought to be related to the influence of a retreating slab beneath the Apennines, but is not understood in detail. A crucial issue is what causes surface extension within the orogen, particularly whether extension occurs as a steady-state process or represents a transitional process. A steady-state process is likely if a simple subduction geometry is revealed in seismic imaging; a transient process would be more likely if the seismic structure is complex. We analyze Receiver Functions (RF) using teleseismic data recorded at 48 temporary stations deployed across the NA orogen. The large data-set (more than 12000 teleseismic records) reveals new details of the seismic structure of the NA collision zone, suggesting the presence of downgoing lithosphere in the region where sub-crustal earthquake activity is sparse. In this study, we focus attention at target depths between 20 km and 120 km (i.e. where the crust-mantle boundary and the supra-slab mantle wedge develops). We present our data-set in closely spaced profiles across and along the NA chain, to highlight the 3D structure of the subducting interfaces and the related anisotropy that develops as deformation proceeds. A shallow Moho is identified on the Tyrrhenian side of the orogen crest, beneath Tuscany. A deeper Moho is found on the Adriatic side of the orogen crest, and both seismic features associated with these "Moho"-like features overlap under the crest of the orogen. Starting from the orogen crest toward the Tyrrhenian side the deeper "Adriatic Moho" becomes a deep Ps converted phase with a moveout consistent with a dipping slab discontinuity. The Ps conversion amplitude depends strongly on the back-azimuth of the incoming P wave, suggesting anisotropy related to localized shear.
S14A-04
Toward a Spectral Source Model of the ETS Process
The physical processes involved in Episodic Tremor and Slip (ETS) are not understood. Examination of possible source spectral models for ETS and comparison with those appropriate for earthquakes could help elucidate the physics of ETS. Although individual earthquakes may deviate, the standard source spectral model for earthquakes is the omega-squared model, in which displacement spectral amplitudes at frequencies above a corner frequency fall off as the inverse of frequency squared. In contrast, tremor has been observed to fall off roughly as the inverse of frequency over 1 to 10 Hz, i.e., as ω-1. The direct observation of a truly broadband spectrum of ETS is difficult, if not impossible, due to the very long durations involved and the weak signal. However, observations of the amplitudes of tremor and a handful of very-low-frequency events (VLFs) can be compared to the long-period moment of an entire ETS inferred from GPS measurements. A simple ω-1 model for ETS has a flat portion of the spectrum and a higher-frequency portion falling off as the inverse of frequency. In the simplest version, the corner frequency between the portions would be set by the duration of the ETS process and is thus very small indeed. This model under-predicts observed spectral amplitudes of Cascadia tremor at 1-10 Hz during ETS by about an order of magnitude. Inferences of spectral amplitude at about ~100 s from VLFs observed in Japan (Ide et al., 2008) are also under-predicted, although not as badly. One possible solution involves a higher corner frequency, for example, that associated with the near-daily tidal cycles, which have been shown to affect tremor generation (e.g., Rubenstein et al., 2008). Another possibility involves a spectral bump at high frequencies due to small- scale patchiness or to a difference in the physical process generating high frequencies and that responsible for slow slip. It should also be noted that above some frequency, the source spectrum must fall off faster than ω-1.5 to avoid an energy catastrophe. As more observations of the amplitudes of ETS processes at various frequencies become available, from a variety of instrument types, we will be able to discriminate between the possibilities and obtain spectral models of the ETS source.
S14A-05
The 75th Anniversary of the Great Sanriku-oki, Japan earthquake of March 2nd, 1933: New Observations and New Insights into the Largest Recorded Outer-Rise Earthquake
The shallow M8.4 outer-rise earthquake of March 2, 1933 is the largest recorded event of its kind. It produced strong ground motions and tsunami waves along the coast of Honshu and more than 3000 fatalities. Kanamori [EPSL, 1971] showed that this event was a normal-faulting rupture with ~N-S-striking nodal planes and proposed that this event represented the tectonic effect of bending of the Pacific Plate in the Japan Trench (JT). Aftershocks occurred far west of the main shock and the JT. This suggested to Kanamori that rupture occurred on a 100-km-wide plane dipping at about 45° toward the JT, implying a depth of seismogenic slip of about 70 km. However, phase data from the larger aftershocks are inconsistent with such depths. A recent investigation of the sP phase data from recent earthquakes recorded by land stations [Gamage et al., AGU 2005 and GJI, in press] indicates that small offshore aftershocks east of the JT are still occurring on this rupture plane 75 years after the main shock, consistent with known aftershock decay properties of this earthquake. None of these aftershocks in this shallow zone occur deeper than about 25 km below the seafloor. A Mw7.1 shallow normal-faulting earthquake in 2005 occurred off the Japan Trench south of the 1933 epicenter. An OBS investigation of its aftershocks confirms that rupture occurred on conjugate normal faults to depths of no more than about 20 km below the seafloor [Hino, Fall AGU, 2007]. The regional pattern of hypocenters of recent offshore events near the JT show a double seismic zone parallel to the seafloor with a zone of shallow normal faulting events within about 20 km of the seafloor and a zone of reverse-faulting earthquakes about 25-30 km deeper, both consistent with convex-upward flexure. These findings cast doubt on the original interpretation that the main shock rupture extended to a depth of 70 km. Moreover, S-P delays of the 1933 aftershocks and their recorded waveforms at Mizuzawa Station suggest that the events east of the JT represent seismic activity on this normal-fault rupture and the aftershocks west of the JT represented normal faulting or interplate thrust faulting near the top of the subducted plate. As such, the aftershocks west of the JT probably represent stress transfer from the normal- faulting rupture of main shock, similar to the pattern observed in aftershocks of the large, shallow outer-rise event in the Kuriles in 2007. A large gravity high exists in the source region of the 1933 main shock. Multibeam swath mapping of the bathymetry there shows long, relatively straight normal fault scarps that obliquely cross cut the magnetic anomalies [Nishizawa, 2001]. The aftershock distribution east of the JT and the above findings suggest source dimensions of L = 220 km, W = 35 km, average displacement of u = 8 m, and rigidity of 0.7×1012 dyn/cm2 could satisfy the scalar moment of 4.3× 1028 dyn-cm estimated by Kanamori from long-period wave amplitudes. The large-angle, cross cutting of seafloor fabric of the fault scarps, and the large gravity anomaly in the source region are consistent with unusually large bending stresses and dynamic displacements.
S14A-06
What controls interplate coupling? Implications from abrupt change in coupling on the Pacific plate across a border between two overlying plates in the southernmost extent of the NE Japan subduction zone
In the southernmost extent of the NE Japan subduction zone, the Pacific plate (PA) is subducting beneath two different tectonic plates – the North American plate (NA) to the north and the Philippine Sea plate (PH) to the south. The change of overlying plate for the PA provides a good opportunity to test the influence of the overlying plate on interplate coupling. In the present study, detailed location of the border between the PH and NA overlying the PA is estimated from slip vectors of the interplate events. Then we compared the interplate coupling coefficients between the two regions overlain by the two plates based on the small repeating earthquake data. Analysis of slip vectors of interplate events shows that the slip vectors abruptly change their slip angles off Kanto. This suggests that the location of the border between the two overlying plates is extending northwestward from the triple junction. The distribution of interplate coupling coefficient estimated from the cumulative slip of small repeating earthquakes reveals a distinct change from south (ca. 0.3) to north (ca. 0.7) across this border. This border corresponds to the southern limit of M > 7 earthquakes and intense seismicity along the Japan Trench, again indicating the stronger coupling to the north. We also investigated the structure of the overlying plates from seismic tomography using a large number of travel-time data obtained from the nationwide seismograph network. The results reveal a distinct low-velocity zone just above the PA in the region overlain by the PH, whereas there is no low-velocity zone in the region overlain by the NA. These observations imply that the overlying plate controls large-scale coupling at the plate interface. Acknowledgement: We used waveforms from the seismic networks of University of Tokyo in addition to the data from Tohoku University. Arrival time data for seismic tomography and earthquake relocation are provided by the Japan Metrological Agency.
S14A-07
Sharpened tomographic image of the subducting slab beneath the Sumatra-Andaman region from 3-D ray tracing and earthquake relocation
The inclusion of data from the 2004-2005 great earthquake sequences into the tomographic modeling of the Sumatra region has greatly improved our resolving ability and has illuminated several important tectonic features along the Sumatra and adjacent subduction zones. Single iteration results using a nested regional- global tomography method and reprocessed global phase data have shown that the slab below northern Sumatra at depth is folded. This fold may influence seismogenesis on the megathrust. To further sharpen our image of the fold and other features, we incorporate into the inversion new data from the 2007 M 8.5 event offshore southern Sumatra and local data from a temporary network around Toba Caldera. We employ 3-D ray tracing and present an improved iterative P-wave velocity model of the region that better fits the data. The new model shows a significant increase in the amplitude and connectivity of the subducting slab within the seismogenic zone and at greater depths, allowing a more complete description of its geometry. In addition, we explore the importance of relocating the events between iterations using a double difference technique recently adapted for 3-D spherical ray tracing using the pseudo-bending method. For the events with waveform data available, we include cross-correlation data and improved arrival picks. We compare the relocated-hypocenter tomographic model to the fixed-hypocenter result and discuss the implications.
S14A-08
The mantle beneath Amazonia: Implications for Precambrian lithosphere evolution
Studying the South American crust, we find an extensive, roughly N-S oriented zone of increased (up to 50 km) Moho depth in the Amazonian craton. This contrasts with the margins between the tectonic domains that form the Amazonian craton (Guyana and Guapore Shields, and the intracratonic Amazonian basin), which lie more or less perpendicular to that zone in the E-W direction, and suggests different mechanisms are responsible for the tectonic domains and the thicker crust. In this study, we explore the underlying mantle in this region using seismic tomography. We combine existing point constraints and surface wave data for the entire South American continent with newly processed data, and jointly invert the combined data set for a 3D S-velocity model of the South American upper mantle. The new data consists of 16 point constraints in eastern Brazil from receiver function analysis, which have not yet been used for seismic tomography, Rayleigh wave group velocities from more than 900 dispersion curves and Rayleigh waveform fits. We use this new 3D S-velocity model to discuss possible processes responsible for the crustal thickening, determine to which extent there are similar patterns in the underlying mantle, and what the nature of these processes implies for the evolution of the Precambrian lithosphere beneath Amazonia.