G33D-01 INVITED
MORVEL: A new estimate for geologically recent plate motions
MORVEL is a new global set of closure-enforced relative angular velocities for the geologically recent motions of 20 tectonic plates. It is determined from a greatly expanded data set consisting of 1696 seafloor spreading rates, 163 transform fault azimuths, and 56 earthquake slip directions from 29 plate boundaries, and 381 GPS station velocities from the Caribbean, Scotia, and North America plates. The former Africa plate now comprises separate Nubia, Lwandle, and Somalia plates, and the former Australia plate now comprises separate Capricorn, Australia, and Macquarie plates. Seafloor spreading rates, which are estimated over the past 3.16 Ma for slow spreading centers and 0.78 Ma for intermediate and fast spreading centers, are adjusted downward to compensate for the several kilometer outward displacement of magnetic reversal zones, which otherwise biases seafloor spreading rates upward by 0.6-2.6 mm yr-1. MORVEL gives a significantly improved description of recent global plate motions. The largest differences in the motions estimated from MORVEL and NUVEL-1A are along the boundaries of the Nazca plate. The motion of the Caribbean plate along its boundaries with the North and South American plates is nearly twice as fast as predicted by NUVEL-1A and is estimated mainly from GPS measurements at 16 locations in the Caribbean plate interior. The differences between plate velocities that we estimate along 19 plate boundaries using MORVEL and GPS are smaller on average than are the differences between plate motions estimated from NUVEL-1A and GPS, which indicates that MORVEL more accurately describes recent plate motions than does NUVEL-1A. These differences are reduced the most for the boundaries of the Nazca plate and the Arabia-Eurasia and India-Eurasia plate boundaries. MORVEL also reduces by more than half the difference between Pacific-North America motion estimates derived independently from GPS and the NUVEL-1A data.
G33D-02
Geodetic Observations of an Earthquake Cycle at the Sumatra Subduction Zone: The Role of Interseismic Strain Segmentation
The Sumatra subduction zone is a highly seismogenic portion of the plate boundary between the India/Australia and Sunda Shelf plates, and was the site of the 26 December 2004, Mw 9.1 Sumatra- Andaman, the 28 March 2005, Mw 8.7 Nias-Simeulue, and the 12 September 2007, Mw 8.4 Mentawai earthquakes. We examine the role of fault segmentation in the earthquake cycle at the Sumatra megathrust using survey-mode and continuous GPS data from 1991-2007, including survey GPS data from 2006 and 2007 that have not been previously published. Our study makes use of the recently completed reanalysis of the entire GPS data archive at the Scripps Orbit and Permanent Array Center (SOPAC). The reanalysis includes data from 1991 up to the present, and updates all our solutions to be consistent with current models (including the adoption of absolute phase center antenna models, latest Differential Code Biases tables, GMF global troposphere mapping function, FES2004 ocean tide loading model, with center of mass correction) and the ITRF2005 reference frame. We use the newly reprocessed GPS data from global stations to provide an updated (ITRF2005) reference for our regional Sumatra solutions. The reprocessed global GPS data are also used to compute updated angular velocities for the major tectonic plates which provide the backdrop of our regional study. The regional data from Sumatra, comprising both survey-mode and continuous GPS, were also reprocessed with the updated software and reference frame. These data allow us to observe the final few years of one earthquake cycle and the beginning of the next. Analysis of the interseismic, coseismic and postseismic phases of recent great earthquakes and comparing them to what is known about historical earthquakes reveals that the Sumatra megathrust is segmented, a characteristic that may persist through multiple earthquake cycles. The Nias-Simeulue earthquake ruptured approximately the same region that broke in 1861, a 300-km-long segment directly SE of and abutting the Sumatra-Andaman rupture zone. Farther southeast, the Mentawai segment of the megathrust (0.5°S-5°S), that produced M >8 earthquakes in 1797 and 1833, appears to be fully locked in the interseismic period but is flanked by two freely slipping regions, the Batu Islands in the NW and Enggano Island in the SE. The 12 September 2007 Mentawai earthquake sequence ruptured only the southern one-third of the 1833 rupture zone. The subduction zone underneath the Batu Islands and Enggano Island which, prior to 2001 was partially to fully coupled, appears now to be slipping freely. We thus find that while the segmentation of the subduction zone is preserved, the interseismic coupling on the subduction fault may vary with time.
G33D-03 INVITED
Upper Plate Deformation is Dominated by Varying Interface Coupling in the Chilean Subduction Context
It has been almost two decades now that GPS has been used to measure plate tectonics and quantify plate
deformation. In South America, the debate rapidly focused on the motion of the Nazca plate relative to the
South America plate. Space geodesy allows to compare plate motions averaged over a few years to plate
motion averaged over several million of years. Since the initial work of (Larson et al., 1997) which found
similar rates, it is well known now (eg: Norabuena et al, 1998; Norabuena et al, 1999; Angermann et al.,
1999; Altamimi et al., 2002; Kendrick et al, 2003; Vigny et al, 2008) that in fact the present day motion of the
Nazca plate is around 15% slower than its Nuvel-1A estimate. This finding has the important consequence
that along the South American margin, instead of nearing 8 cm/yr, today's subduction rate ranges from 5.5
cm/yr in Equator to 7 cm/yr in central Chile, before it decreases again to 6.5 cm/yr in southern Chile. Part of
this convergence rate is taken up by permanent strain contributing to the building of the Andes, but most of it
generates elastic deformation recovered during the seismic cycle with an average of one M=8 event every
ten years and at least one M>8.7 per century in what corresponds to the Chilean portion of the Nazca
subduction. Surface deformation is representative of these processes and GPS measurements made in the
area aim at quantifying the different contributions and defining the style of deformation. Up to now, two
different families of models have been presented: 2-plates model, involving homogeneous medium where the
slab geometry varies with latitude and depth, (Klotz et al., 2001; Khazaradze et al., 2003); 3-plates model
involving a rigid sliver between the 2 main plates (Kendrick et al., 2003; Brooks et al, 2003). Steady state
velocities predicted by these models differ at the cm/yr level in places and GPS measurements should allow
to discriminate easily between them. However, recent measurements we carried out on small scale dense
networks in Chile in the vicinity of the trench (Concepion - 36° S, Coquimbo - 30° S, Antofagasta –
22° S) show that the deformation exhibits very different patterns in distinct areas and abrupt changes with
latitude. We demonstrate that to model these patterns with a full coupling on the trench is not possible
everywhere, and following others, we conclude that coupling must be varying on the subduction interface,
both with depth and along strike, and can reach value as low as 40% regionally and even less locally.
Moreover, these low-coupling areas could correspond to transient deformation associated to the seismic
cycle. This signal completely dominates the deformation patterns we measured at the surface over the last
decade, and renders extremely difficult to detect the permanent deformation not taken up by the recoverable
elastic deformation or even to simply quantify the style of deformation.
http://www.geologie.ens.fr/~vigny
G33D-04
Accommodation of convergence in North Chile seismic gap: questions raised by 2007 Mw 7.7 Tocopilla earthquake
Nazca plate subducts under the South American continent at a rate of ~7 cm/yr oriented ~80° at the latitude of North Chile (17°S to 23°S). This area was recognized as a major seismic gap of about 1000 km length since no earthquake occurred during the 120 years following the South Peru (Mw= 9.1, 16 August 1868) and the Iquique (Mw=9.0, 10 May 1877) megathrust events. This gap was reduced to 500 km length after the Arequipa (Mw = 8.3, 23 June 2001) and the Antofagasta (Mw = 8.1, 30 July 1995) earthquakes, and might have accumulated up to 9 m of slip deficit by interseismic loading on the subduction interface. According to usual scaling laws for subduction earthquakes this corresponds to an earthquake of magnitude up to 8.5. On November 14, 2007, a Mw 7.7 subduction earthquake occurred in the southern part of this seismic gap. It was recorded by a network of ~20 cGPS instruments, and ASAR coseismic interferograms were calculated. Analysis of geodetic data show that the earthquake initiated in the vicinity of Tocopilla city and was arrested ~150km south below Mejillones peninsula, area already identified as a potential seismic barrier. The earthquake activated the deep part of the seismogenic zone down to the transition zone (35-50 km depth) and did not reach the surface. It slipped parallel to the convergence direction requiring no slip partitioning and released a very small portion (<2m) of the slip deficit accumulated in the seismic gap. These specificities raise a series of questions relative to the way convergence is accommodated in North Chile seismic gap. Is Tocopilla earthquake a precursor to a future megathrust event that would release the whole elastic strain accumulated since 130 years? Shall it be regarded as part of a series of Mw 7.5-8 earthquakes occurring every 5 years that break progressively the gap? What part of the convergence is accommodated by aseismic slip events? How is it related to earthquakes? We address these questions by analyzing GPS and InSAR data acquired before and after 2007 Tocopilla earthquake.
G33D-05
Kinematic modeling of crustal deformation in the Central Japan
Central Japan is situated on a complex plate boundary region where 3 or 4 tectonic plates or microplates converging one another. Crustal deformation along the southern coast is dominated by locking effects on the subducting plate boundary, while East-West compression caused by collision between northeast and southwest Japan is prominent further inland. In order to understand plate interaction and seismotectonics in this region, we conduct an analysis of crustal deformation using a geodetic inversion method by McCaffrey (2002). In this model, crustal movement is interpreted as a superposition of block rotations, locking effects of block-boundary faults, and uniform strain within each block. Continuous GPS observation data are available since April 1996. We devide the whole data set from 1996 to 2008 into 3 periods, 1996-2000, 2001-2004, and 2006-2008, and estimate displacement rate at each GPS station from daily coordinate data. The first and third periods represent the normal, interplate locking status. On the other hand, there occurred the Tokai slow slip event (TSSE) in the second period. We construct a block division model referring active fault traces, microseismicity, and GPS displacement rate data. In addition, configuration of the subducted Philippine Sea plate is modeled based on microseismicity and seismic exploration results. We build up a number of block models by combining these blocks and estimate the optimal block model based on the Akaike information criterion (AIC). During 1996-2000, plate boundary shallower than 30 km depth is fully locked. On the other hand, a negative slip deficit appeares under the north of Lake Hamana during 2001-2004, which corresponds to TSSE. In our analysis, slip distribution of TSSE is estimated without removing the steady component of crustal deformation as was done in previous studies. In addition, the direction as well as the magnitude of the slip deficit has changed from those of previous analysis by taking inland deformation into account. On the other hand, we found that relative motions between neighboring blocks are consistent with the geologic slip rate of the block boundary faults in most cases. Also, it is shown that the strain rate of a block corresponding to the strain concentration zone in central Japan is balanced by the average seismic moment release rate for about 400 years.
G33D-06
Subduction Interface Coupling, Strike-Slip Faulting, and Block Rotation in Southwest Japan From Interpretation of GPS Data
We invert GPS and earthquake slip vector data from southwest Japan for poles of rotation of tectonic blocks, and the degree of interseismic coupling on faults in the region, including the Nankai Trough and Ryukyu Trench. Our estimate of interseismic coupling on the Nankai Trough and Ryukyu Trench subduction interfaces highlights a major along-strike change in depth of interseismic coupling on the subduction interface between southwest Honshu and Kyushu. In some cases, our estimates for long-term slip rates on the major active faults (such as the Median Tectonic Line) agree well with geological estimates. However, interpretation of GPS velocities and historic earthquakes suggest that a highly active left-lateral shear zone (10-15 mm/yr slip rate) cuts across the island of Kyushu. Surprisingly, no active faults have been identified within this zone of rapid contemporary deformation. We propose that the shear zone can be explained as a response to the subduction of an aseismic ridge (the Kyushu-Palau Ridge) at the southwest end of the Nankai Trough. Because of rapid (~40 mm/yr) along-strike migration of the ridge, we suggest that the ridge subduction point (and resulting left-lateral shear zone) is never in one place long enough to enable the development of a through-going fault zone that can be identified at the ground surface, reconciling the mismatch between the GPS, seismological and geological data in this region. Our conceptual model is supported by numerical modeling results, showing that when the locus of deformation migrates with time at 40 cm/yr, the cumulative strain in any given location is small, even for large rates of instantaneous deformation. Our interpretation of GPS velocities also requires back-arc rifting (up to several mm/yr) in the Kagoshima Graben in southern Kyushu. Due to the small dimensions of southern Kyushu, it is difficult to uniquely determine if anti-clockwise rotation of the southern Kyushu forearc (determined from paleomagnetic studies of Miocene-Pliocene aged rocks) is ongoing today.
G33D-07
Plate motion, slip rates, and partitioning of deformation in Japan
The Japanese Islands mark the leading edge of the Pacific-greater Eurasia plate boundary zone, where deformation is partitioned between upper plate faults and subduction zones. We use a three-dimensional spherical block model, constrained by the spatially dense Geonet GPS network, to simultaneously estimate plate motions, fault slip rates, and spatially variable coupling on the Japan, Sagami, and Nankai subduction zones. We find high slip rates (up 15 mm/yr) on the Niigata-Kobe Tectonic Zone, which has previously been proposed as a major tectonic boundary. Oblique convergence at the Nankai Trough is partitioned between dextral slip on the Median Tectonic Line and a component of margin-parallel slip on the subduction interface. In northeastern Japan, we find that the lateral component of relative plate motion across the Japan-Kuril trench is accommodated dominantly by up to 40 mm/yr of strike-slip on the plate boundary and/or distributed deformation throughout the forearc. GPS observations from 1996--2000 allow for high-resolution imaging of spatially variable coupling on the Japan, Sagami, and Nankai subduction zones and reveal concentrations of elastic strain accumulation that coincide with the rupture areas of the 2003 Tokachi-oki earthquake and the 2003--2005 sequence of seismic events offshore Sendai. This suggests that similar analyses with contemporary GPS data will aid in constraining rupture areas of impending earthquakes.
G33D-08
Reconciling Geodesy and Geology at Subduction Plate Boundaries: Forearc Translation and Orogeny
It has long been recognized that elastic processes associated with the seismic cycle dominate the surface velocity or strain field at subduction zones. However, it may also be true that some fraction of geodetically measured displacement involves permanent deformation of the upper plate, for example, translation of forearc blocks at high angles to plate convergence and crustal shortening and mountain building. We explore the potential for reconciling geodetic versus geologic estimates of deformation at subduction zones, utilizing new results from the Central America convergent margin, but drawing on data from convergent margins worldwide. The interseismic, GPS derived velocity field in Central America has been interpreted in terms of two major processes: elastic strain accumulation due to locking on the shallow (less than 50 km down-dip depth) part of the dipping plate boundary and trench-parallel translation of the forearc due to oblique subduction. Our new velocity field also indicates significant strain partitioning in southern Costa Rica where tectonic shortening and uplift of a young mountain belt takes place. To a first approximation, the former may be regarded as a purely elastic process, generating no net deformation of overriding plate lithosphere, while the latter two processes are capable of generating considerable long-term lateral displacement of forearc terrains and upper plate shortening. We present a model of these processes assuming elastic half space rheology, and draw inferences from the model results regarding the driving forces for both trench-parallel forearc motion and shortening. We then compare our results with estimates of upper plate deformation based on geologic mapping and structural analysis and discuss the issues associated with correlating these disparate data sets.