S43D-01 INVITED 13:40h
The Relation Between Tectonics, Fluid Flow and Seismogenesis at Convergent Erosional Margins
The integration of seismic reflection images, multibeam bathymetry, sidescan sonar images, seafloor observations, heat flux measurements and well-located microearthquakes show the relationship between tectonics, fluid flow and seismic activity at the erosional Middle America convergent margin. Seismic images show that the plate boundary is characterized by large amplitude reflections under the continental slope. The amplitude of plate-boundary reflectivity rapidly decreases inboard, beneath the continental shelf. Temperatures calculated at the plate boundary indicate that large amplitudes occur where dehydration reactions (~60 C to ~160 C) release chemically-bound water of subducting sediment. The rapid decrease in plate boundary reflectivity parallels the predicted decrease in dehydration-related fluid release with increasing temperature ($>$160 C). Thus, seismic images give the qualitative distribution of fluids along the plate boundary. Above the area of high fluid release, the overriding plate is traversed by numerous faults that link the hydrological system of the plate boundary with the seafloor. Fluids rising across the upper plate vent across the continental slope at fault scarps and mud diapirs with carbonate crust and chemosynthetic fauna indicative of active fluid flow. The association between areas of higher fluid release, strongly deformed upper plate and venting at the seafloor is observed to correspond to the continental slope for hundreds of km along the margin. Underneath the continental shelf, the upper plate is comparatively little deformed and the plate boundary has little fluid. Here, large earthquakes nucleate and microearthquakes indicate stronger mechanical coupling of the plates. We hypothesize that progressive changes in plate boundary fluid content influence faulting mechanisms, and that changes in upper plate strength control the amount of elastic energy that can be stored. Those two parameters may be key for earthquake generation.
S43D-02 13:55h
The plate interface revealed across the Hikurangi subduction zone, offshore New Zealand.
The North Island GeopHysical Transect (NIGHT) was conducted in 2001 to study the detailed shallow structure of the Hikurangi subduction zone. The study included a multichannel seismic (MCS) reflection and a wide-angle seismic survey using ocean bottom seismometers (OBS) and offshore-onshore stations. One of the most interesting features in the pre-stacked MCS data is a marked change in dip from 3 to 16 degrees of the subducting oceanic crust (Hikurangi Plateau), at 13 km depth and 120 km from the trench. Splay faults propagate from this point on the plate boundary interface up into the overlying crust, coinciding with the actively deforming Kidnapper and Lachlan Ridges' in Hawke Bay. Modelling of a pronounced inflection in the Free-Air gravity gradient also confirms the presence of higher density basement rocks thrust upward from the bend in the plate boundary interface. Passive seismic data also suggest that the margin ridges mark a change from a decoupled (seaward) to a coupled (landward) interplate boundary and lend support to the idea that the active splay faults accommodate most of the elastic strain seaward of the locked zone on the plate interface. Structures we have observed along the central part of the Hikurangi subduction zone appear similar to those described in the Nankai Trough margin, Japan.
S43D-03 14:10h
Incremental Growth of Quartz Slickenfibres on Shearing Surfaces in an Ancient Subduction Complex - Record of Paleomicroearthquakes?
An accretionary m$\'{e}$lange of Triassic age ocean floor sediments exposed at Chrystalls Beach in SE Otago, New Zealand is made up predominantly of competent sandstone and chert phacoids set in a cleaved mudstone matrix, all deformed in a subgreenschist environment (3 $<$ P $<$ 6 kbar; T $\sim$ 250$\pm$30 $\deg$C). Distributed brittle fracturing within the sandstone lenses contrasts with ductile deformation involving solution transfer accompanying cleavage development in the pelites. The more pelitic portions of the m$\'{e}$lange are also pervaded by an anastomosing cellular mesh comprising multiple shearing surfaces subparallel to cleavage coated with incrementally developed quartz (and locally calcite) slickenfibres. Sliding surfaces splay and amalgamate and are also commonly interlinked by sets of extension veins localised within dilational jogs. Microstructural observations reveal growth of the slickenfibres by `crack-seal' shearing increments of 10-100 microns, with incremental slip transfer of the same order across the dilational jogs accommodated by the opening of extension fractures that link en $\'{e}$chelon slip surfaces. Together with widespread extension fracturing in the more competent sandstones these structures suggest that the tensile overpressure condition (P$_{f}$ $>$ $\sigma$$_{3}$) was widely achieved during brittle deformation under near-lithostatic fluid overpressures. Individual slip surfaces can be traced for metres to tens of metres so that the ratio of slip, u, to potential rupture length, L, lies within the range, 10$^{-5}$ $<$ u/L $<$ 10$^{-4}$, characteristic of microearthquakes obeying `constant stress-drop' scaling with $\Delta$$\tau$ $\sim$ 3 MPa. For this scaling relationship, the slip increments are comparable to those expected for M0 to M-2 microearthquakes. The host-rock assemblage, metamorphic environment, inference of near-lithostatic fluid overpressures, and mixed continuous-discontinuous style of shearing, resemble conditions inferred for the active subduction channel shear zone associated with the Hikurangi subduction margin in the North Island, where clustered microseismicity is associated with creeping portions of the thrust interface.
S43D-04 14:25h
High-Resolution Subduction Zone Seismicity and Velocity Structure in Ibaraki, Japan
We use double-difference tomography (tomoDD) [Zhang and Thurber, 2003] and waveform-derived cross-correlation differential arrival times to invert for the earthquake locations and P- and S-wave velocity distributions in the subduction zone under Ibaraki Prefecture of north-central Honshu, Japan. The Ibaraki region is attractive for its high rate of slab seismicity and for the presence of an intermediate-depth double seismic zone. We relocate ~8000 events occurring in this region between June 2002 and June 2004. We use a combination of ~200,000 absolute travel times, ~5 million catalog-derived differential times, and ~5 million cross-correlation differential times derived from more than 150,000 waveforms, with roughly equal numbers of P- and S-wave data. Many of the waveforms are from HiNet borehole stations that provide particularly high-quality data. We also use data from JMA, the University of Tokyo, and Tohoku University. Since it is natural to expect sharp velocity contrasts in a subduction zone, we regularize the inversion using the total variation (TV) approach implemented through iteratively reweighted least squares. Because TV is an L1-norm regularization, sharp changes in velocity are penalized no more than gradual ones, but undulations in the velocity model remain damped. We will compare the TV results with those determined by standard least-squares, L2-norm regularization. Our results show increasingly organized seismicity including narrowing by up to 50% of the upper and lower limbs of the double seismic zone as viewed in cross-section. We find a zone of interplate events extending as deep as 60 km, forming a very distinct lineation in cross-section. Focal mechanisms support the interpretation that these are low angle, subduction interface events. These earthquakes are accompanied by a zone of very high Vp/Vs ratio within the downgoing plate, just beneath the seismicity, suggesting that high pore-pressures may enable seismic slip on the subduction interface at depths where aseismic slip would otherwise predominate. These events represent significantly deeper seismic coupling than the 37-43 km maximum depth observed in this area previously by Tichelaar and Ruff [1993], but are consistent with the maximum depth of 50-70 km for low-angle thrust events found by Igarashi et al. [2001] farther to the north.
S43D-05 INVITED 14:40h
Challenges in Defining Seismogenic Zone Using Geodetic and Structural Observations
The coseismic fault slip area in recent subduction earthquakes can be determined from seismological, tsumani, and geodetic observations. The coseismic rupture appears to be limited generally by a temperature around 125$\deg$C at the updip end and a temperature around 350$\deg$C or the intersection of the plate interface with the forearc Moho at the downdip end, with significant along-strike variations. Defining the seismogenic zone from interseismic deformation is much more challenging, because of the fewer available observation methods and the poorly understood earthquake cycle deformation process. It is reasonable to expect the rupture zone to be locked in the interseismic period, but the updip and downdip limits of mechanical locking do not usually have clear simple geodetic signatures. Fault motion outside the locked zone is not simply determined by frictional properties and fault stress. Surface deformation changes through the earthquake cycle due to both transient fault slip and viscoelastic stress relaxation. If an elastic dislocation model is used to explain decade- and century-scale viscoelastic interseismic deformation, a more gradual downdip tapering of locking is required to fit geodetic observations long after the most recent earthquake. Rate- and state-dependent friction and nonlinear mantle rock rheology are both important candidates in explaining transient afterslip of duration of days to a few years downdip of coseismic rupture, although their distinction is obscured by strain localization in nonlinear deformation. Newtonian rheology is arguably applicable a few fault dimensions from the rupture zone and several years after the earthquake. Cascadia and Nankai episodic "silent" slips indicate that the forearc material at 30-40 km depths is able to accumulate and release elastic strain energy. It has been proposed that such slip may be shear deformation of a band of km's thickness above the subducting slab (and that the shear band terminates seismogenic zone), but this gives a problem of mass balance. The slip below the forearc Moho is likely along a thin contact zone between the slab and hydrated cold mantle wedge.
S43D-06 14:55h
Uplift and subsidence along the Cascadia subduction zone determined from historical repeated leveling
A better understanding of the vertical crustal deformation throughout the Pacific Northwest of the United States will help model coupling along the plate boundary and eventually help us better assess the earthquake potential in the region. In this study, I examine historical leveling data from Cascadia and calculate uplift rates throughout the region. Vertical deformation rates are calculated from historical differential leveling. I calculate rates by taking the difference between elevations at common benchmarks and dividing by the time between surveys. I remove outliers based on a median absolute deviation technique. The uplift rates are adjusted using a least squares procedure based on a bivariate cosine series basis function. Uplift rates calculated from tide gauges (Weldon, personal communication) were included in the adjustment and held fixed, thereby making sea-level a common datum. Once outliers are identified and the data adjusted to a common datum, I smooth the data using an iterative, robust, weighted, moving average technique. The results indicate landward tilting in several regions along the coast alternating with regions of little or no deformation. There were no regions with seaward tilting. In some regions the coastal uplift continues inland but landward tilting appears to dominate the deformation. The rate of landward tilting varies significantly along the coast with the fastest deformation occurring to the north. At about 45\deg N, there is little vertical deformation either at the coast or inland. The lack of vertical deformation across central Oregon may represent a transition in subduction behavior between northern and southern segments.
S43D-07 15:10h
Large-scale Seismic and Aseismic Deformation Patterns Associated With Subduction: Constraints From Continuous GPS measurements in Mexico
We use coordinate time series from 27 continuous GPS stations in Mexico, including five new sites, to describe evidence for seismic and aseismic slip along the Middle America subduction interface. Spatial filtering of our GPS coordinate time series effectively removes inter-site correlated noise from external, non-tectonic causes, precluding them as an explanation for the movements we observe. For the 1995 M$_{W}$ = 8 Colima-Jalisco megathrust earthquake, we find previously undescribed displacements of up to 20 millimeters during and after the earthquake at sites up to 400 km from the epicenter. Periods of aseismic transient slip, some previously unrecognized, are apparent in the time series for GPS stations along most of the Pacific coast and up to 500 km inboard from the trench. Between early 1993 and early 2004, we recognize episodes of transient strain release late in 1995 and early in 1998, 1999, 2000, 2001 and 2002. In each event, over 10 millimeters of transient slip was recorded by at least three stations within two months of one another, and all of these sites are found to be moving toward the Middle America trench during the transients, consistent with protracted relaxation of elastic strain related to subduction. We estimate the onset, duration, magnitude, and direction of transient displacement by a least-squares inversion to find the best fit hyperbolic tangent functions describing the GPS coordinate time series. Considering that many sites moved coherently towards an area between Acapulco and Oaxaca, we also employ an inverse procedure to examine whether transients observed within the network are consistent with a source of slip that is stationary or propagating. In either case, our results underscore the importance of transient aseismic fault slip in accommodating subduction. Given our evidence for coordinated transient signals at sites far from the trench, our results demonstrate the geographically extensive effects of this phenomenon in Mexico, presumably due to the shallow subduction angle of the Cocos plate. These results imply that future geodetic tests for crustal deformation within Mexico must consider the complicating effects of transient slip on the motions of sites far from the trench.
S43D-08 15:25h
Testing Predictions of Along-strike Variations in the Shallow Seismogenic Behavior of Subduction Zones
Large scale inter-arc differences in the seismogenic behavior of subduction zones are usually attributed to variations in the age and convergence rate of the subducting plate, and therefore to variations in normal tractions on the plate interface. In contrast to this focus on differences between subduction zones, two recent studies have found a clear global correlation between along-strike variations of gravity and topography within a given subduction zone and the principal locations of seismic moment release in large plate interface earthquakes (Song and Simons, Science, 2003; Wells et al., JGR, 2003). In particular, these studies (henceforth referred to as SS03 and W03) found that most large earthquakes occurring on the shallow subduction interface occur in regions where trench-parallel gravity and topography anomalies (TPGA and TPTA) are negative, and regions of positive TPGA and TPTA appear to not experience significant seismic moment release. These observations suggest that along-strike variations in seismic behavior within a given subduction zone are as great as differences between subduction zones and that frictional properties of the plate interface can vary over short distances and may control long-term seismic moment release. These conclusions suggest that asperities of great earthquakes are persistent and not ephemeral, that there is a close relationship between seismogenic behavior and geologic evolution of the forearc, and that we may be able to predict where areas of high moment release will be in the future. We test predictions of SS03 and W03 against the distribution of co-seismic and post-seismic deformation from recent events including the 2003 Mw 8.3 Tokachi-Oki (Japan) and the Mw 7.8 Rat Island (Aleutians) earthquakes, as well as against a recent reanalysis of seismic, GPS, and InSAR data for the 1995 Mw 8.1 Antofagasta, Chile earthquake. The location of coseismic slip in all cases, and the along strike variations in inferred post-seismic after-slip for the Antofagasta and Tokachi-Oki earthquakes are consistent with the predictions of SS03 and W03. In addition to considering patterns of coseismic and postseismic fault slip to test the predictions of SS03 and W03, we can also consider measuring along-strike variation in interseismic deformation. We suggest that the Kuril Arc provides the ideal location for such a blind test. The northern and southern Kurils have a history of large shallow earthquakes, while the central Kurils are relatively quiescent. There are two obvious possibilities -- either the large earthquakes in this region have yet to occur, or they will never occur. SS03 finds a large region of positive TPGA over the plate interface in the central Kuril, and thus predicts this region to be relatively uncoupled relative to the regions north and south of it. The predicted along-strike variation in convergence velocity measured on the arc is approximately 2 cm/yr, well within detectability limit of GPS surveys.