S51B-0145 0800h
Rheology of Talc: Consequences for Subduction Processes and the Localization of Deformation
Subduction of altered oceanic lithosphere results in the dehydration of serpentinite and formation of talc. Talc also forms on oceanic shear zones that have a long history of localized deformation and fluid flow (i.e., oceanic detachments and transform faults). Being an extremely weak material, the presence of even small amounts of talc can substantially weaken the subducted lithosphere, and affect its behavior and evolution. Despite its potential importance for ocean tectonics, many aspcts of the mechanical behavior of this mineral are poorly characterized. Consequently, we are conducting conventional triaxial mechanical tests on intact cores of talc at P<400 MPa and T<600\degC, and a strain rate of $\sim$10$^{-5}$ s$^{-1}$. At that strain rate, with confining pressure of 100 MPa and at room temperature, the peak strength is only $\sim$50 MPa. Unlike crystalline silicate rocks that fail by Mohr-Coulomb criterion under similar conditions, the strength of the talc rocks is modestly dependant on pressure. Even so, deformation localizes on sets of parallel and cross cutting shear zones oriented ~45\deg° from the shortening direction. Within the errors of the strain gauge systems used to measure volumetric strain, little or no dilation occurs during deformation. Suprisingly, this style of localized deformation was also observed for an experiment conducted at T$\sim$600\degC and P=300 MPa. The extremely weak nature of talc (<10% of the maximum strength of pure peridotite, and $\sim$20% of that of lizardite serpentinite) may play a major role on the dynamics of subduction zones and on the rheology of altered mantle. For example, hydration of the overlying mantle wedge by fluids arising from the dehydratation of the subducting slab could result in the formation of talc along grain boundaries, enhancing plastic deformation at very low degrees of alteration.
S51B-0146 0800h
Modeling Seismically Induced Deformation And Fluid Flow In The Nankai Subduction Zone
Fluid pressure changes induced by seismic strains in the Nankai subduction zone were investigated through numerical modeling. Seismic strains resulting from dislocations along fault surfaces were coupled to pore pressure generation, and subsequent transient fluid flow was simulated. This study is distinct from previous efforts that modeled homogeneous systems. Effects of variable mechanical and hydrologic properties on the coupled hydromechanical system were investigated by assigning different mechanical and hydrological properties to marine sediments, the decollement zone, and the upper oceanic crust. Seismicities of varying magnitude were simulated by discrete dislocation events. Model sensitivity studies suggest that for a reasonable range of parameter scenarios, transient pressure head signals caused by discrete dislocations of a few meters in the updip region of the seismogenic zone can be observed over large areas of the margin from near the seafloor to deep in the crust. Compressibility, a parameter responsible for the co-seismic response of pore pressures to strain, was varied from 10E-11 1/Pa for the crust below the decollement to 10E-9 1/Pa for the sediments above the decollement. Pressure head changes on the order of tens of meters were observed, while larger head anomalies exist near the tips of the dislocated segments. The duration for the seismically induced pressure head to dissipate depends on the hydraulic diffusivity of the system, which is a function of permeability as well as storage. Permeability in sediments wedge was varied with depth from 10E-15 m2 near the seafloor to approximately 10E-21 m2 at 10 km below seafloor. Upper oceanic crust below the decollement was assumed to have permeability on the order of 10E-12 m2. A scheme was adapted to allow permeability of the decollement to increase after dislocation. Preliminary modeling results suggest that post-seismic permeability changes involving a small region of the dislocation zone would have a limited effect on the overall picture of the hydrologic regime. Permeability increases in larger areas of the decollement, however, would shorten the pressure dissipation periods. Most pressure head dissipation takes place within the first tens to thousands of years following the seismic events. As more monitoring data become available, the coupled mechanical and hydrologic model can be further utilized to assess hydrologic responses of the seismogenic zone processes and provide an additional means to infer large-scale mechanical and hydrogeological parameters of the subduction zone.
S51B-0147 0800h
The Influence of Supercritical H$_{2}$O on Elastic Properties
Water seems to be an important controlling parameter in active tectonic settings. Especially the concept of earthquake-triggering by "dehydration embrittlement" due to fluid release during dehydration and an increasing pore fluid pressure is discussed intensively. Nonetheless, little is known about their petrophysical signature. To study the influence of fluids on the elastic properties of rocks we performed ultrasonic experiments on encapsulated and non-encapsulated cores of amphibolite and serpentinite. P and S wave velocities are deduced in an internally heated gas-pressure vessel at 900MPa and up to 700$ \deg $C. Without capsule a slight temperature derivative for P-waves (dVp/dT of -0.43 m/(s K) and -0.79 m/( s K) for amphibolite and Serpentinite, respectively) fit quite well with literature data determined and are attributed to the intrinsic elastic properties. Velocities measured on capsulated serpentinite shows a similar behaviour. In contrast, on encapsulated amphibolite we deduce far below dehydration conditions a significant reduction of P-wave derivative of -1.31 m/(s K) with increasing crack density. This can be traced back to water at grain boundaries and an enhanced pore fluid pressure, which affect the sample microstructure exceedingly. Between 400-550$ \deg $C we observe a time-dependent re-increase of velocities, which seems related to a reduction of the pore space as the result of precipitation of minerals along grain boundaries and/ or the expulsion of the fluid phase. Our results show, that already small amounts of water can effect the sample microstructure enormously and thus its elastic properties.
S51B-0148 0800h
Permeability Measurements of Ocean Drilling Program Samples Collected From the Costa Rican Subduction Zone
Permeability is a key component in evaluating pore pressures and consolidation history in sediments. Core samples collected from the Costa Rican subduction zone were tested to determine their vertical permeabilities. The samples are from Ocean Drilling Program (ODP) Leg 205 Sites 1253, 1254, and 1255 and ODP Leg 170 Site 1040. The Costa Rican subduction zone has been shown to currently be a nonaccreting margin, and all samples are from within the underthrust sediments. The measurements were made using a constant head test by inducing a pressure gradient across the sample. Measurements were taken for three different consolidation pressures with five different pressure gradients induced across the sample in each consolidation. Two types of samples were tested; light colored, fine grained pelagic carbonates and dark colored, fine grained hemipelagic clays. Our results show two groupings of permeabilities, with the pelagic carbonates approximately two orders of magnitude higher than the hemipelagic clays. These results are consistent with permeabilities found by previous researchers.
S51B-0149 0800h
Thrusting-related Fluid Flow in Subduction Zone: Geologic and Isotopic Evidence From Ancient Subduction Complex
Role of fluid in a seismogenic plate boundary of subduction zone remains poorly understood. To reveal the origin and behavior of syn-tectonic fluid in the ancient subduction complex, we examined occurrences of veins and carbon and oxygen isotope composition of vein calcite of the Mugi Melange in the Shimanto Belt. The melange preserves a relationship between rock deformation and fluid flow near the up-dip limit of seismogenic zone. The Mugi Melange is a late Cretaceous to Paleocene tectonic melange, and is characterized by blocks of sandstone and basalt in shale matrix. The Mugi Melange is divided into five units based on the repetition of basaltic layer of N-MORB, representing duplex-underplated assemblage. Paleo-temperature determined by vitrinite reflectance is 120-150 \deg C in the studied section, and fluid inclusion thermo-barometry indicates 125-245 \deg C and 92-149 MPa (Matsumura et al., 2003). This P-T condition is in the vicinity of up-dip limit of the seismogenic zone. Major deformational stage of the Mugi Melange is underthrusting (D1), underplating (D2), and uplifting (D3). There are various occurrences of quartz and calcite veins in melange formed during various stages. We classified three occurrences of veins named intra-boudin veins (IBV), network veins (NWV), and fault-parallel veins (FPV). IBV are observed only around the neck of boudined sandstone block in the black shale matrix, so they were precipitated when block-in-matrix structure was formed due to layer-parallel extension during D1. NWV are distributed in the damage zone of duplex ramp thrust, and are mainly perpendicular to shear surface of the thrust. Such occurrences suggest that NWV formed inter-faulting period of D2. FPV show implosion breccia-like texture (Sibson, 1986), and they are observed just below the basalt / shale boundary. Vein minerals are dominantly calcite. These features suggest that FPV would be precipitated as a result of co-slip fluid migration during oceanic crust underplating. We analyzed carbon and oxygen isotopic compositions of calcite obtained from these three occurrences of veins. Oxygen isotopic compositions are in the narrow range of +15.8 to +19.2 \permil (vs. V-SMOW) for all veins. \delta $^{18}$O values increase from IBV to FPV. Precipitation temperatures determined by fluid inclusion (Matsumura et al., 2003) are higher in NWV than in IBV. These results show that delta-18O values of fluids increase from IBV ($\sim$ +4 \permil) to FPV ($\sim$ +10 \permil). Assuming the same dehydration reaction of clay minerals of host rock and isotopic equilibrium, the dehydration temperature is lowest for IBV, moderate for NWV, and highest for FPV. In contrast to \delta $^{18}$O values, carbon isotopic compositions of vein calcite are broad, in -17.2 to -3.7 \permil (vs. V-PDB) for all veins. IBV and NWV include light carbon, while FPV take heavy one. This result suggests that organic carbon from shale and igneous carbon from basalt were mixed in the fault core. In conclusion, we suggest that the fluid dehydrated from both shale and basalt was accumulated in cracks in damage zone of plate boundary fault, and migrated to shallower portion in association with seismic event.
S51B-0150 0800h
Seismic Noise Correlation with Seismicity and Fluid Flow
During the Costa Rica Seismic Zone Experiment (CRSEIZE), we observed intriguing correlations between earthquake activity, seismic noise observed on the seafloor, and fluid flow through the seafloor (K. M. Brown and others, this meeting). These correlations are observed on instruments separated by as much as 30 kilometers. Here we describe the nature of the noise, the temporal patterns, and speculate about the transmission mechanism and the source(s). Seismic noise on the seafloor is strongly influenced by ocean wave activity, so we first compare the temporal variations in noise with the ocean wave field as computed from the global wind field by Tolman. This comparison shows that the long-term (months) pattern of wave data provides a baseline for the the seismic noise field, but does not predict the shorter (days) impulsive noise anomalies which correlate with variations in fluid flow. At some sites, the noise shows spectral peaks which have been explained as peaks in the mode Q of Scholte waves (Schreiner and Dorman) and shear resonances (Godin and Chapman). The frequencies of these peaks is controlled by the shear velocity structure and the frequencies themselves provide earth data constraining the seafloor velocity structure.
S51B-0151 0800h
Pore-Water Freshening in the Nankai Trough, Japan: Implications for Lateral Fluid Flow
Pore-water freshening has been observed in ocean sediments cored along the Muroto Transect in the Nankai Trough, offshore Japan during ODP Legs 190/196. Chloride concentrations up to $\sim$11% below normal seawater values ([Cl-] $\sim$555 mM) were documented at drill site 1173, 13 km seaward of the trench. Updip, lateral fluid flow from depth, and in situ clay dehydration have been proposed as freshening sources. Placing realistic bounds on the relative contributions of these potential sources has importance for quantifying fluid and chemical fluxes in subduction zones. Here, we evaluate the hypothesis that freshening is due solely to in situ clay dehydration. We simulate the general shape and magnitude of the down hole [Cl-] anomaly observed at site 1173 using a one-dimensional model which combines sedimentation, thermal history, clay dehydration, and Cl- diffusion. Sediment accumulation during arcward transport and high heat flow for the area ($\sim$180 mW/m2), result in temperatures sufficient to transform smectite to illite + H20. The volume of released water scales with initial smectite abundance (here we assume 50% by weight based on mineralogy at site 1177, where heat flow is low and clays are less reacted). In the model, released water is added at each time step creating unstable [Cl-] profiles. Chloride diffusion is modeled implicitly in a reference frame fixed to the interstitial water column, through which consolidating sediment nodes move. Basal boundary conditions used in the model are: constant [Cl-], no-flow, and the inclusion of oceanic crust in diffusion. Simulated reaction progress matches observed clay mineralogy for site 1173. Minimum simulated [Cl-] scales with initial smectite content, ranging from 4% freshening for 20% initial smectite, to 14% freshening for 70% initial smectite. Simulated [Cl-] profiles for all boundary conditions mimic the general [Cl-] profile for site 1173, but no single boundary matches observed data exactly. A no-flow basal boundary reaches a minimum [Cl-] of $\sim$500 mM (10% freshening) but does so at the sediment-oceanic crust interface, $\sim$100 m below the observed minimum. Constant-concentration basal conditions show a return to normal [Cl-] near the base of the section, but lack [Cl-] minima and maxima comparable to those observed. Oceanic crust is included in some models to simulate communication between basement and sediment. These models show similar behavior to no-flow and constant-concentration boundaries, differing mainly in magnitude.
S51B-0152 0800h
Vp/Vs Ratio of Hornblende Gabbro Xenolith from Ichinomegata Volcano, NE Japan
In order to interpret petrologically the inhomogeneous Vp, Vs and Vp/Vs structure beneath the NE Honshu arc (Nakajima et al., 2001), we measured P-wave velocity (Vp) and S-wave velocity (Vs) of Ichinomegata hornblende gabbro xenolith simultaneously within the solidus P-T conditions (up to $700\deg$C and 0.8 GPa) in this study. The rock sample was loaded in the talc-pyrophyllite high-pressure cell. Travel times of elastic-wave through the rock sample were determined with the pulse reflection technique. A sample length was cut into about 6mm diameter core and doubly polished to a length of about 6 mm. Before velocity measurements, the rock sample was oven-dried at $120\deg$C for 24 hours and the H$_{2}$O was not added to the rock sample. The measured rock includes the hornblende gabbro of Ichinomegata xenolith, which has ultrabasic composition (41.3 wt% SiO$_{2}$O) and is mainly composed of plagioclase (>An$_{86}$, 80.6 vol% in mode ratio), aluminous hornblende (13.9 wt.% Al$_{2}$0$_{3}$, 12.5 vol%) and magnetite (6.9 vol%). At constant pressure of 0.6 GPa, the Vp of the hornblende gabbro gradually decreased from 6.83 km/s at $25\deg$C to 6.59 km/s at $600\deg$C. Similarly the Vs values reduce from 3.87 km/s at $25\deg$C to 3.74 km/s at $600\deg$C. The Vp/Vs ratio of the hornblende gabbro is 1.76 at $600\deg$C. The Vp/Vs ratio of the hornblende gabbro is comparable to that of the lower crustal layer beneath the western side of Ou back-born range to back-arc side (Vp/Vs = 1.75-1.79), except near active volcanic areas (Vp/Vs > 1.83). Nakajima et al. (2001) indicated that the low Vp, low Vs and high Vp/Vs anomalies in the lower crust to uppermost mantle beneath the active volcanic areas were perhaps caused by inclusions containing melt. These high Vp/Vs anomalies do not consist with the Vp/Vs ratio of the hornblende gabbro in this study. It is suggested that the high Vp/Vs anomalies in the lower crust to uppermost mantle is comparable to the high temperature conditions beyond the solidus with partial melting and/or other materials.
S51B-0153 0800h
Central Mexico Intermediate-Depth Seismicity and its Possible Relationship to the Regional Geotectonic Structure
Although all convergent regions have the same basic characteristics, they are highly variable features. Each controlled by the type of crustal material involved and the tectonic setting. The development of the Plate Tectonics theory give us a general framework to understand the relationship between earthquakes and subduction process. Details, however, have to be treated case by case since in many regions of the Earth they are still unknown. Explanation to the unusual position of the Trans-Mexican Volcanic Belt with respect to the Middle America Trench, and its origin itself, is one of those open questions. The present report provides an overview of the intermediate-depth seismic activity throughout Central Mexico. The discussion is based on the characteristics of damaging earthquakes of moderate magnitude that have occurred in the past 76 years in a narrow belt transversal to the Trench axis at 18>| N, in an approximately E-W direction. Even though is not an easy task to link the origin of intermediate-depth earthquakes to a brittle fracture or frictional sliding, the lineament of foci and the similarity of the focal mechanisms suggest that intermediate-depth earthquakes in Central Mexico occur along a pre-existing deep-seated normal fault system, probably in the upper part of the subducting Cocos plate. The main faults as suggested by the data, would run parallel to the southern border of the Trans-Mexican Volcanic Belt. Moreover, I speculate that these pre-existing faults are, probably, an extension of the Rivera Fracture zone observed to the west. The existence of a deep fracture system would facilitate the uprising of Mantle material and would explain why the active volcanoes chain of Central Mexico runs in an E-W direction instead of parallel to the coast as in other subduction regions of the world. It would also help to understand the genesis of intermediate-depth earthquakes in Mexico.
S51B-0154 0800h
Intermediate depth seismicity - a reflection seismic approach
During subduction the descending oceanic lithosphere is subject to metamorphic reactions, some of them associated with the release of fluids. It is now widely accepted, that these reactions and associated dehydration processes are directly related with the generation of intermediate depth earthquakes (dehydration embrittlement). However, the structure of the layered oceanic plate at depth and the location of the earthquakes relative to structural units of the subducting plate (sources within the oceanic crust and/or in the upper oceanic mantle lithosphere?) are still not resolved yet. This is in mainly due to the fact that the observational resolution needed to address these topics (in the range of only a few kilometers) is hardly achieved in field experiments and related studies. Here we study the wavefields of intermediate depth earthquakes typically observed by temporary networks in order to assess their high-resolution potential in resolving structure of the down going slab and locus of seismicity. In particular we study whether the subducted oceanic Moho can be detected by the analysis of secondary phases of local earthquakes (near vertical reflection). Due to the irregular geometry of sources and receivers we apply an imaging technique similar to diffraction stack migration. The method is tested using synthetic data both based on 2-D finite difference simulations and 3-D kinematic ray tracing. The accuracy of the hypocenter location and onset times crucial for the successful application of stacking techniques (coherency) was achieved by the use of relatively relocated intermediate depth seismicity. Additionally, we simulate the propagation of the wavefields at larger distance (wide angle) indicating the development of guided waves traveling in the low-velocity waveguide associated with the modeled oceanic crust. We also present application on local earthquake data from the South American subduction zone.
S51B-0155 0800h
Understanding the Mechanism of Deep Earthquakes: A new System for Detecting Acoustic Emissions in Multianvil Experiments
One of the major goals in the experimental study of deep earthquakes is to identify slip instabilities at high pressure and high temperature (HPHT) that might be responsible for earthquake occurrence. Detecting acoustic emissions from a specimen during faulting provides unique constraints on the instability process. There are few experimental studies reporting acoustic emissions under HPHT conditions, due to technical challenges. And those studies only used one or at most two acoustic sensors during the experiments. This precludes accurate location of the acoustic emission source region and thus the ability to distinguish real signal from noise that may be coming from outside the sample. We have developed a new system for detecting acoustic emissions at HPHT. Here we present a 4-channel acoustic emission detecting system working in the HPHT octahedral multianvil apparatus. Each channel has high resolution (12 bit) and a sampling rate of 30 MHz. In preliminary runs at pressures of 1 - 6 GPa and temperatures 300 - 1130 K, we have observed acoustic emissions. Analyzing these signals, we are able to show that this system permits us to distinguish between signal and noise, locate the source of the acoustic emission, and obtain reliable data on the radiation pattern. This system has greatly improved our ability to study faulting instabilities under high pressure and high temperature.
S51B-0156 0800h
Using pulse shapes of intermediate depth seismicity to locate sources relative to subducting oceanic crust showing low seismic velocity
We are proposing a method to locate intermediate depth earthquakes in respect to the position of a relatively low velocity subducted crust by classifying pulse shapes recorded at specific receiver locations. P wave onsets of intermediate depth earthquakes at subduction zones indicate that low velocity layering of less than 10 km thickness is present in the vicinity of nearly all circum pacific slab surfaces. Whether low velocity is caused by hydrous metabasalt phases persisting to depth, released fluids, dehydration within the subducting slab, a serpentinized layer or even partial melts on top of subducted crust is a subject of ongoing discussion. Defining source locations relative to such a low velocity structure will help to rule out part of these interpretations of intermediate depth seismicity and shed more light on the source processes involved. We use 2D finite difference modelling to calculate full synthetic wave propagation in subduction zones. Our simulations show that a low velocity structure of km scale brings forth characteristic p onsets at near coastal receiver locations for sufficiently long propagation distances along the slab surface. Pulse shapes and frequency of these synthetic onsets are analysed regarding source location relative to the slab surface. We test Double Couple (DC) source position and orientation in an up dip slab geometry derived from local seismicity in Northern Chile. The layered slab consists of eclogized upper crust (5 km) and a layer of slow basaltic lower crust (2 km) on top of fast oceanic mantle. The tests account for velocity undulations in subducted crust, roughness of the low velocity structure and slab geometry as well as scale of layering. We find that sources of intermediate depth events at 21° S in Northern Chile are situated within or directly below the lower crust at distances of less than 2 km from the oceanic Moho.
S51B-0157 0800h
Seismogenic zone structure along the Middle America Trench, Nicoya Peninsula, Costa Rica, from 3D local earthquake tomography
The shallow seismogenic portion of the Nicoya Peninsula, Costa Rica, segment of the Middle America Trench generates damaging large (M$_{W}$$>$7.5) earthquakes over a recurrence interval of 50 years. We present hypocenters and three-dimensional seismic velocity models ({\it V$_{P}$} and {\it V$_{P}$}/{\it V$_{S}$}) calculated using simultaneous inversion of {\it P}- and {\it S}-wave arrival time data from 611 small magnitude earthquakes. We design a characteristic velocity model using the Nicoya experiment geometry to provide additional resolution estimates. Data was collected as part of the Costa Rica Seismogenic Zone Experiment (CRSEIZE), a collaborative geodetic, passive seismic, and fluid flux program. Along the southern Nicoya Peninsula plate interface microseismicity extends from 12-26 km below sea level ($\sim$73-100 km from the trench axis). To the north interplate seismicity extends from 17-28 km below sea level (75-87 km from the trench axis). Interplate seismicity begins downdip of a shallow geodetically locked patch and region of low {\it V$_{P}$} along the plate interface. The $\sim$5 km depth offset to the updip limit of microseismicity occurs across the plate suture of East Pacific Rise-derived and Cocos-Nazca Spreading Center-derived oceanic lithosphere and correlates with models of the $250\deg$C isotherm; the offset most likely reflects differing thermal regimes of the incoming oceanic crust due to along-strike variations in hydrothermal circulation. The downdip edge of plate interface seismicity occurs updip of the continental Moho/oceanic slab intersection and does not uniquely correlate with models of the $350\deg$C isotherm. Low forearc mantle wedge {\it P}-wave velocities suggest 20-30% serpentinization along the Nicoya Peninsula region; {\it V$_{P}$}/{\it V$_{S}$} ranges from 1.70-1.82 km/sec, values consistent with 0-10% serpentinization.
S51B-0158 0800h
Waveform Cross-Correlation Analysis of Seismic Data From the Costa Rica Seismogenic Zone
Earthquakes nucleating within the seismogenic zone at convergent margins generate more than 80% of the total world-wide seismic moment release and can pose significant seismic hazard to coastal areas. The mechanical behavior along the seismogenic zone is not well understood due to the limited constraint on the precise geometry (thickness of the planar interface as well as the up- and down-dip limits) and the degree of plate coupling in these regions. The 1999-2001 collaborative Costa Rica Seismogenic Zone Experiment (CRSEIZE) consisted of a joint seismic and geodetic investigation of the Costa Rica plate interface to learn about the processes occurring at subduction margins. As part of this study, two seismic transects were established across the Middle America Trench at the Nicoya and Osa Peninsulas in Costa Rica and velocities at 46 GPS sites throughout the country were determined. The seismic arrays consisted of both land and ocean bottom broadband and short-period stations situated directly above the seismogenic zone, allowing for direct recording of local seismicity. Ongoing work includes determination of earthquake locations and the three-dimensional P- and S-wave velocity structure in northern Costa Rica using local seismic tomography. Currently, waveform cross-correlation techniques to improve P- and S-wave arrival times are being incorporated. The automated correlation and clustering method employed greatly reduces the picking inconsistencies compared to human analysis alone and has improved relative relocations allowing previous determinations of the up- and down-dip limits of the seismogenic zone to be refined. More importantly, waveform cross-correlation has improved P-wave first-motion determinations for many events, resulting in more abundant and reliable focal mechanisms. This improves our ability to differentiate between underthrusting interplate events and intraplate seismicity. In addition, the degree of waveform similarity has been assessed to identify repeating earthquakes. In northeastern Japan, waveform similarity studies have revealed that numerous earthquakes generated by repeated slip at small asperities, with recurrence intervals on the order of months, occur only on the plate boundary (Igarashi et al., 2003). Our analysis in northern Costa Rica has identified many clusters of similar events, both above and below the plate interface; however, unlike NE Japan, we find no evidence for repeating events that generate identical waveforms at the same stations. These combined results allow for improved characterization of seismogenic zone processes.
S51B-0159 0800h
Amphibious Local Seismic Observations by SFB 574 in Costa Rica
The goal of the SFB 574 ''Volatiles and Fluids in Subduction Zones'' subproject A2 is to study the seismogenic zone of Costa Rica and Nicaragua as examples of an erosive margin and to better understand its local variability. In 2002/2003 we studied the seismogenic zone in two adjacent areas of Costa Rica. One of the areas is characterised by the subduction of a seamount, the other one contains a megalens structure, which had been inferred from reflection seismic data before and which is interpreted to indicate a possible mechanism for mass transfer from the upper to the lower plate. 23 ocean bottom sensors from IFM-GEOMAR and 15 landstations from the GFZ Potsdam were deployed in the coastal Pacific region of central Costa Rica near Jaco in April 2002. The network was moved south-east towards Quepos in October 2002 and operated until spring 2003. 1,968 earthquakes between April and October 2002 could be located by the Jaco network. Most of the earthquakes took place offshore beneath the continental margin close to or beneath the network. The hypocenter determination of these events using the on- and offshore parts of the network delivers very precise earthquake locations, because the network covers the source region very well. Another region of high seismic activity is located southeast of the network, where a magnitude 6.3 earthquake took place on June 16, 2002, followed by several hundred aftershocks. Ongoing studies focus especially on the updip limit of these events. From the Quepos network 1,241 earthquakes between October and December 2002 have been located, so far. In a pilot study focal mechanism of 13 earthquakes with clear polarities had been determined using the Jaco onshore network only. Except for two earthquakes these events were shallow and took place in the continental wedge. The orientations of their focal planes coincides well with the geological fault system of the Jaco area. Ongoing work focuses on slab related earthquakes.
http://www.sfb574.geomar.de
S51B-0160 0800h
Structure of the Ryukyu Subduction Zone at its Western end: Slab Buckling, Double Seismic Layer, and the Effect of Dehydration
Network data of Taiwan and Japan were integrated to illuminate the collision-oblique subduction complex in a region within 100 km of the island's coast, which has been poorly resolved by either single network. We relocated 4814 events applying a series of 1-D velocity model inversion and double difference method. These processes reduce the variance of traveltime residuals by about 70% with respect to each network value. The relocated seismicity delineates better the curving of the trench-forarc system toward the island and the 15--20 km apart double seismogenic layer (DSL) within the slab of the subducted Philippine Sea Plate in the depth range of 30-70 km. Not revealed before is the seismicity distribution that clearly defines bulging of the slab near its western end continuing from 50 km depth to 90 km. The overall geometry and the focal mechanisms suggest that the slab buckles against the Eurasian plate under lateral compression while subducting obliquely toward the continent. Tomographic inversion of the data set further reduces variance by 61%. The images reveal that much of the DSL is punctuated by low velocity anomalies (LVA) on the upper layer. We propose two hypotheses to explain the buckling-DSL-LVA combination, based on a previous notion that the DSL results from lateral compression. It could be that diabase and olivine present different creep rheology for crust and mantle causing the double layer, and that dehydration of hydrous minerals triggers melting that lowers the seismic velocity. Lateral compression could have thickened the typical oceanic crust to accommodate the at least 15 km gap for DSL. Or, a non-basaltic section of crust, thick and low in velocity, is subducted, which causes the 15--20 km separated double layer and the low anomalies in tomography. Subduction of a buoyant segment of crust has various implications for the dynamics of this subduction-collision junction.
S51B-0161 0800h
Imaging crust, mantle and slab structure around the Shikoku district, Japan, by Double-Difference Tomography and its relationship to the occurrence of nonvolcanic deep tremors
Obara [2002] detected nonvolcanic deep, low-frequency tremors occurring along the strike of the subducting Philippine Sea plate over a length of 600 km. The depth of the tremors averaged about 30 km, near the Moho, and some tremors migrate at a rate of 13km/day along the strike. Although the occurrence of tremors is hypothesized to be associated with fluids supplied by dehydration reactions of minerals in the slab, there have been no systematic studies of the heterogeneous structure around the source area of tremors. This study is the first attempt to estimate the detailed P- and S-wave velocity structures beneath the Shikoku district and the Kii Peninsula, where deep, low-frequency tremors are frequently found. Double-difference tomography [Zhang and Thurber, 2003] was used to estimate three-dimensional (3D) P- and S-wave velocity models. The inversion started from a one-dimensional velocity model. The total number of earthquakes and seismograph stations are about 4,000 and 120, respectively. The 3D model space is represented by a set of regular grid nodes with grid intervals spaced 20-30 km in the horizontal direction and 5-20 km in the vertical direction. The results show that high-velocity zones are distributed along the deep seismic zone beneath the western part of Shikoku and the Kii Channel, which may correspond to the subducting Philippine Sea plate. On the other hand, low velocity zones are present immediately above them in the western part of Shikoku and the central part of the Kii Peninsula with Vp/Vs ratios generally greater than 1.8. Interestingly, deep, low frequency microearthquakes (tremors) determined by the Japan Meteorological Agency (JMA) occur in and around such low-velocity and high-Vp/Vs zones. Acknowledgments: Arrival-time data and the location of deep, low-frequency microearthquakes are provided by JMA.
S51B-0162 0800h
3-D Prestack Depth Imaging the Nankai Accretionary Wedge off Cape Muroto, Southwest Japan
The Nankai Trough subduction zone off southwest Japan is one of the best-suited convergent plate margins for studying large interplate subduction-zone earthquakes as well as the formation of accretionary prisms. At this margin, the Philippine Sea Plate (PSP) is subducting beneath the Eurasian Plate (EP) to the NNW. The plate convergence rate is estimated to be 4 - 5 cm/yr. Large thrust earthquakes have repeatedly occurred along the Nankai subduction zone with a recurrence interval of 100-200 years. The most recent interplate earthquake was the Nankai earthquake (Mw=8.2), which occurred in 1946 off the Kii Peninsula. In order to figure out seismic structure and stratigraphy of the Nankai accretionary wedge off cape Muroto, southwest Japan, we have conducted three-dimensional (3-D) multichannel seismic (MCS) reflection survey using R/V Ewing in 1999. We acquired the MCS data on 81 separate lines with 100 m line spacing, each 80 km long, producing 8 X 80 km 3-D seismic volume. To obtain the 3-D prestack depth migration images, we constructed and updated a 3-D interval velocity model using the CDP bin gathers for which preconditioning processings including amplitude recovery, deconvolution, and multiple suppression were applied. Miocene to Pliocene Shikoku Basin sediments underthrusts the overlying accretionary prism along a decollement as the PSP subducts beneath the EP. The oceanic crust of the subducting PSP is traceable over the entire inlines. Several imbricate thrust faults are observed in the overlying accretionary wedge. The decollement steps down on the top of subducting oceanic crust around ~30 km landward from the deformation front. We recognize several sigmoid, landward dipping out-of-sequence thrust (OOST) faults in the landward thick wedge package. Most of the OOSTs are apparently developed from the subducting oceanic basement to the seafloor in the forearc region, cutting both underthrust sediments and the overriding accretionary prism. In this paper, we will show and discuss recent results of the 3-D prestack depth imaging, visualization, and seismic structural/stratigraphic interpretation.
S51B-0163 0800h
Along strike changes in basement topography and sediment thickness in the northern Shikoku Basin: Variable inputs to the Nankai Trough Seismogenic Zone
The very thick terrigenous sediments in the Nankai Trough and the northern Shikoku Basin mask significant basement irregularities that will soon be subducted into the Nankai seismogenic zone. We present a new seismic reflection data set across the northern Shikoku Basin that images remarkable variations in basement relief and corresponding variations in sediment thickness and type. The basement relief has a wavelength of ~5-15 km, with the amplitude of the peaks and troughs generally ~300-400m. Two significant troughs exist south of Shikoku that are each ~ 12 -15 km wide and 1900 m deep. These troughs strike NW-SE and are at least 40 km long. They are mostly filled with high amplitude continuous reflections that onlap the trough margins. We correlate these reflections with the mid-Miocene lower Shikoku Basin turbidite sequence drilled at ODP site 1177. Along strike approximately 125 km to the NE, off the Kii Peninsula, there is a significant basement high that is just entering the trench ~ 15 km from the base of the inner trench slope. It is roughly oval with a diameter of ~50km parallel to the trench axis and ~20 km perpendicular to the trench, with a peak that is 1400 m higher than the adjacent crust. This basement high (seamount) is covered by the hemipelagic Shikoku Basin sediments that thin from 900 m on the SW flank to 200 m at the summit. The Shikoku Basin turbidite section that fills the troughs to the SW is absent on the seamount and its flanks. Because of the variations in sediment thickness around this seamount, its bathymetric summit is only 800 m above the adjacent seafloor while the basement relief is 1400m. Between the seamount and Zenisu Ridge (~ 60 km NE), the basement deepens by 500 m and has subdued relief (compared to that SW of the seamount). Basement is overlain by the Shikoku Basin hemipelagic sequence and onlapped by young trench turbidites. These significant along strike variations in basement topography (1400m high seamounts and 1900m deep troughs), sediment thickness (200-2000m) and sediment type (hemipelagics vs turbidites) will soon be input to the Nankai seismogenic zone and are representative of the kinds of irregularities that have been input in the geologic past. Such variations must contribute to along-strike differences in seismogenic behavior.
S51B-0164 0800h
Flat-Slab to Steep Subduction Transition Zone in Central Chile-Western Argentina: Body Waves Tomography and State of Stress
Two temporary networks were deployed for about 100 days in the flat-slab subduction zone of Central Chile from Nov. 1999 to Feb. 2000 (30-32S, 70-71.5W; 38 SP stations), and in the transition zone from flat-slab to steep subduction zone of Central Chile-Western Argentina from Nov. 2002 to March 2003 (31.5-34S, 67-71.5W; 29 BB stations). Both networks recorded in continuous mode and were complemented with the permanent stations of the Univ. of Chile within the zone. Several thousands earthquakes were recorded and hypocenter locations were obtained for about 50 events/day. Accurate focal mechanisms were determined for a subset of the best recorded events, using first arrival polarities of P-waves, which permitted to estimate the stress tensor in different regions along the subducted Nazca plate. Local tomography for P- and S-waves velocities from data of each network shows an image of the subduction zone from the flat-slab to the steep subduction. A sharp transition zone at around 32.2S is well correlated with the subduction of the Juan Fernandez Ridge within the slab. Seismicity was relocated using the 3D velocity model obtained from tomography. It is distributed around the trench, along the coast with mainly compressive interplate events at depths down to 50-60 km, along the subducted slab with mainly extensional intraslab events at maximum depths of 150 km at the flat-slab zone and 200 km at the steeper dip zone, and crustal shallow events (depths < 30 km) along the overriding continental plate. This seismicity clusters around the subducted Juan Fernandez Ridge along its entire track within the downgoing slab. More events of high magnitude are observed on that area, hence more seismic moment and energy is released as compared with the surrounding regions. Crustal shallow seismicity occurs mostly at the Chilean side of the Andes Cordillera and foothills, and at the back-arc in the Argentinean Precordillera. This observation indicates a different stress regime and deformation styles at the very steep frontal Andes, relative to the back-arc with smoother topography. Besides a NS differentiation associated with the subduction style is also observed: the seismicity is higher at the front-arc in the steep subduction zone segment, and higher at the back-arc in the flat-slab subduction segment. Finally, a new model of the isodepth contours of the subducted Nazca plate (27.5-34.5S) was obtained from the best hypocenter locations of the temporary networks, including the catalog of the permanent stations of the Univ. of Chile and a previous temporary network at the northern end of the flat-slab segment. This model shows that the subducted Juan Fernandez Ridge controls the transition from flat to steep subduction, and that its buoyancy is associated with the flattening of the slab in the northern part of the studied area.
S51B-0165 0800h
Plate convergence along the northern Manila Trench, Taiwan-Luzon region
The Philippine Sea Plate overrides the Eurasian Plate along the east-dipping Manila Trench. From Luzon to Taiwan, the plate convergence evolves gradually from normal subduction to collision. Further north, the Taiwan orogen has been created. As evidenced by the earthquakes, the subduction-related earthquakes become diffusive close to Taiwan. The accretionary prism has also become wider toward Taiwan. To understand the transition of the plate convergence, we have collected 6 reflection seismic profiles across the Manila Trench between Luzon and Taiwan. The results show that the basement generally displays larger dipping angle in the south than in the north. The trench-fill sediments near the trench have larger quantity in the south than in the north. In the north, the trench-fill sediments have even been uplifted. Structural analysis shows that the crustal structures close to the trench area can be divided into two distinctive sub-areas: the normal fault zone and the proto-thrust zone. The normal fault zone is characterized by the distribution of numerous normal faults in the upper layer of the bent subducting plate. When the normal faults approache the trench, they are generally covered by trench-fill sediments. It implies that the normal faults occur at a maximum bending moment of the plate. Some normal faults resumes probably due to the strong plate convergence near the accretionary prism. The proto-thrust zone is located between the normal fault zone and the frontal thrust of the accretionary prism. Proto-thrust zone contains numerous blind-thrust beneath the trench area. The observation of the proto-thrust zone suggests two tectonic insights. Firstly, the compression of plate convergence comes from the base of the decollement and propagates upwards. Alternatively, the blind-thrusts come from the inheritance of the subducting normal faults.
S51B-0166 0800h
Investigation of Subducting Slab beneath Northeastern Taiwan from Sp Converted Phases
The converted waves from upper plate boundary of the subducting slab denoted as Sp phase recorded by the Central Weather Bureau Seismographic Network (CWBSN) for events in northern Taiwan provide us an advantageous opportunity to construct the geometry of upper boundary of descending Philippine Sea Plate (PSP). The events were chosen for the region with latitude of 24.4 degree N - 25.2 degree N, longitude of 121.5 degree E - 122.2 degree E, depth from 50 -300 km, and time period of January 1991 to Mary 2003 for magnitude greater than 4.0. We apply the polarization filter analysis, particle motion diagram and theoretical travel time to the digital 3-compoent short period seismograms to identify the conceivable Sp converted phases. The initial velocity model with subducting slab was used for theoretical travel time calculation based on the recent tomographic results. The model is constructed firstly by considering a fixed velocity contrasts among the discontinuities. The dip angle and latitude location of slab are allowed to change to obtain the best travel time to the data. Several tests had been made for the inversion. Considering variance reduction of the all observed data to the constructed models by grid searching technique, variance reduction is not in satisfactory. Further studies by grouping the events in north-south and northwestern-southeastern trend were examined. Although the geometry of subducting slab might be difficult to model due to disadvantage in tectonic setting of northeastern Taiwan, which is located to northern subducting slab, the most possible image of the slab from the clear identified converted phases is sketched.
S51B-0167 0800h
Structure of the Northern Cascadia Subduction Zone: A 3-D Tomographic P-wave Velocity Model
A large-scale 3D P-wave velocity model to ~60 km depth has been constructed for SW British Columbia and NW Washington through tomographic inversion of first-arrival times from controlled source experiments together with local and regional earthquake travel-time data recorded at permanent stations. 150000 first-arrival times recorded at 225 temporary stations from the 1998 Seismic Hazards Investigation in Puget Sound (SHIPS) experiment, and 60000 first-arrival times from 3000 earthquakes recorded at 91 permanent recording stations are inverted for a minimum structure velocity model. The RMS residuals for the initial and final models are 764 and 132 ms, respectively, which represents a 97% variance reduction. Checkerboard resolution tests indicate a horizontal resolution of 30 km down to 20 km depth, and 50 km down to 60 km depth. The velocity model images the structure of the forearc crust/upper mantle, and the subducting Juan de Fuca plate geometry beneath the region. The sedimentary basins in the Straits of Georgia and Juan de Fuca and Puget Sound are well defined by the velocity model. The mafic Eocene Crescent Terrane (Metchosin Igneous Complex in southern Vancouver Island) is shown to dip beneath the margin to at least 20 km depth. This terrane is regionally extensive beneath the Strait of Juan de Fuca and the Puget lowland, with higher than average velocities of ~7 km/s at approximately 15 km depth. Beneath the Olympic Peninsula, the Core rocks (accretionary sedimentary prism) are under-thrust beneath the Crescent Terrane to a depth of at least 30 km. At this location most seismicity lies within the overlying Crescent Terrane; the under-thrusting Core rocks are aseismic. The strong Crescent terrane seismicity may be due to deformation induced by the underthrusting. Beneath southern Vancouver Island, the subduction thrust zone above the Juan de Fuca plate is characterized by low velocities of 6.4-6.6 km/s at a depth of 25-35 km. Such low velocities may be due to trapped fluids, sheared lower crustal rocks, and possibly underthrust accretionary sedimentary or metamorphic rocks. This low velocity region coincides with the high conductivity region mapped in previous magneto-telluric studies and with a dipping band of seismic reflectors; it is devoid of seismicity. It probably represents a zone of aseismic slip. Low velocities of 7.2-7.6 km/s are observed in the forearc upper mantle beneath the Strait of Georgia and Puget Sound. Such low upper mantle velocities are interpreted to be due to regional serpentinization of cool forearc mantle peridotite by fluids rising from the dehydrating underlying Juan de Fuca crust. The Tertiary sedimentary basins in the Strait of Georgia and Puget lowland lie directly above the zone of forearc upper mantle serpentinization. In contrast, the sedimentary basins in the Strait of Juan de Fuca lie in a synclinal depression in the Crescent volcanic Terrane.
S51B-0168 0800h
High-Resolution Seismic Velocity Structure of the Alaska Subduction Zone Revealed by Double-Difference Tomography
We use catalog picks of tens of thousands of earthquakes selected from the Alaska Earthquake Information Center from 1988 to 2002 and between latitude N57$^{o}$ and N65.5$^{o}$ to conduct a high-resolution seismic tomography study of the Alaska Subduction Zone by double-difference tomography. Two active source data sets in the Prince William Sound (1989) and southern Alaska (1984-1985) are also included in the inversion to provide more constraints on the shallow velocity structure. There are ~150 stations used in the inversion that were operational during some or all of the time period from 1988 to 2002. The preliminary P- and S-wave velocity models show clearly the low velocities in both the crust and mantle beneath the active volcanoes. The velocity structures in and around the subducting slab are different for the Kodiak, Kenai and McKinley blocks of the subducting plate. For the Kodiak block, a low velocity zone is present within the slab right below the high velocity zone where most of the Wadati-Benioff Zone (WBZ) earthquakes are located. In comparison, for the Kenai block, the velocity is higher in the zone below the WBZ earthquakes. The McKinley block shows a more complex slab velocity pattern, with high and low velocity zones at different depths within the slab. The high-resolution velocity structure will provide additional constraint for models of the segmentation of the Alaska Subduction Zone, in addition to earthquake locations, volcanic arc geometry and composition, etc., and may provide a possible explanation for the segmentation.
S51B-0169 0800h
Precise Measurements of Depth Phase Relative Arrival Times to Refine Estimates of Double Benioff Zone Geometry
In subduction zones, inclined Wadati-Benioff zones of seismicity occur within the subducting lithosphere, and thus can be used to examine the structure of downgoing slabs. A particularly detailed set of information about the slab can be obtained from curious Double Benioff Zones (DBZs), where two sub-parallel planes of seismicity are separated in depth by as much as 30 km. Based on a handful of observations of DBZs, the planes have been proposed to mark the location of petrologic dehydration reactions, which could in turn trigger earthquakes via dehydration embrittlement. However, identification of DBZs has typically been limited to areas where dense local networks cover the subduction zone. We demonstrate a new technique that strictly uses globally recorded teleseismic waveforms to more precisely determine the location of subduction zone earthquakes, which in turn facilitates identification and characterization of DBZs. We obtain better precision in the relative depths of earthquake hypocenters by cross-correlating depth phase waveforms to determine their relative arrival times ($pP-P, sP-P,$ and $sS-S$) for a group of nearby events recorded at a common station. Relative depths are a critical measurement for determining the separation between planes of seismicity in DBZs. This technique is first applied to the Alaska subduction zone to facilitate comparison with hypocenters determined from local network data, and then we apply the technique to the Sunda subduction zone to illustrate a previously unreported DBZ. In the latter case, subsequent analysis of corresponding fault plane solutions indicates the typical pattern of down-dip extension in the lower layer, but the upper layer shows along-strike compression whereas other prominent DBZs show down-dip compression in the upper layer (i.e., Japan, Kamchatka, Tonga). The ability to identify DBZs away from local networks will be a key step in determining the overall prevalence and regularity of this feature on a global basis, which is necessary to understand the conditions, both seismic and petrologic, in the evolution of subducting slabs in general.