G21A-0653
Real Plates and Dubious Microplates
From the onset of plate tectonics, the existence of most of the plates was never put in doubt, although the boundaries of some plates, like Africa, were later revised. There are however, two microplates in northeast Asia, the Amurian and Okhotsk, whose existence and the sense of rotation was revised several times. The rms value of plate-residual GPS velocities is 0.5-0.9 mm/a for sets of stations representing the motion of the following plates: Antarctic, Australian, Eurasian, North American, Nubian, Pacific, and South American. This value can be regarded as an upper bound on deviation of real plates from infinite stiffness. The rms value of plate-residual GPS velocities is 1.2-1.8 mm/a for the Indian, Nazca, and Somalian plates. Higher rms values for India and Nazca are attributed to the noisier data. The higher rms value for Somalia appears to arise from the distributed deformation to the east of the East African Rift; whether this statement is true can only be decided from observations of denser network in the future. From the analysis of plate-residual GPS velocities, the Canadian Arctic and northeastern Siberia belong to the North American plate. The detailed GPS survey on Sakhalin Island shows that the Sea of Okhotsk region also belongs to the North American plate while the region to the west of it belongs to the Eurasian plate. These results provide a constraint on the geometry of the North American plate and put in doubt the existence of smaller plates in northeast Asia.
G21A-0654
Elastic Strain Released in Earthquakes at Convergent Plate Boundaries
At convergent plate boundaries, the accumulated elastic strain is released mostly in great shallow thrust earthquakes. Understanding which fraction of the elastic strain due to plate motion is released in earthquakes is important in assessing the seismogenic potential at convergent boundaries where most of great earthquakes occur. Conventionally, an estimate of the released fraction of the total elastic strain relies on the Kostrov summation of GCMT moment tensors. However, the GCMT moments of great earthquakes can be erroneous by a factor of 2 because of parameterization as a simple point source with triangular moment-rate function and because of trade-off between estimates of the moment and dip angle. Therefore, assessment of the current seismic potential caused by plate motion can be wrong. For the great 2006-2007 Kuril earthquakes, coseismic moments from inversion of GPS offsets are larger than GCMT moments by at least a factor of 1.5. We attribute this disagreement to the simplified source model used in GCMT inversion. From kinematic GPS solution, a contribution of slow coseismic slip (captured by GPS but not by seismometers) was insignificant. Earlier, a similar disagreement between geodetic and GCMT moments was demonstrated for the great 2004 Sumatra earthquake.
G21A-0655
Geologic Slip Rates and Interseismic Deformation in the Ventura Region, Southern California
The Ventura region lies in the Transverse Ranges of southern California and contains a complex network of
non-planar reverse, oblique slip, and reactivated normal faults. Existing two-dimensional models of the region
match much of the recent geodetic velocity field, but utilize highly simplified fault geometries. To determine if
a model incorporating geologically-reasonable three-dimensional fault surfaces can match geodetic velocities
in the region, we create a set of numerical models based on the Southern California Earthquake Center
Community Fault Model and compare results to an updated GPS velocity field. Because significant non-
tectonic ground motions occur in southern California, we use Principal Component Analysis to remove non-
tectonic transient signals from a network of 15 permanent GPS stations in the Ventura region. Instead of
prescribing slip rates to the modeled faults a priori, we use GPS velocities to determine a regional shortening
rate. We apply this shortening rate to the model to drive slip along frictionally-weak fault surfaces. To
simulate interseismic deformation, we subtract the portion of fault slip within the seismogenic crust and
compare the model-predicted surface velocities to geodetic velocities. We find that the three-dimensional
model agrees well with geologic slip rate estimates and matches much of the regional GPS velocities. Model-
GPS residuals suggest that zones of localized contraction at the edges of the Ventura sedimentary basin are
more pronounced in the GPS velocities than the model results, suggesting that a homogeneous crustal
structure may be insufficient to match high rates of contraction observed at the margins of the Ventura
sedimentary basin.
http://www.people.umass.edu/stmarsha/ventura.html
G21A-0656
Fault slip rate estimates for southern California using viscoelastic plate-block models
To estimate fault slip rates of faults in southern California is a main task for earth science community. However, estimates of fault slip rates using geodetic data are model dependent. The most comprehensive fault slip rate estimates using GPS data for southern California are derived from elastic block models. In these models, tectonic blocks bounded by faults are assumed to move as undeformed bodies over the long- term. During the interseismic period, elastic strain accumulation due to locking of faults is modeled as a perturbation to this long-term block motion and steady fault slip with back-slip on locked portions of faults using dislocations in an elastic half-space. However, using this model, slip rate estimates for the Garlock Fault, Mojave segment of the San Andreas Fault, and the San Bernadino bend in the San Andreas Fault are all significantly lower than estimates from geologic data. We are examining the possibility that the discrepancy between slip rates estimated using geologic and geodetic data can be attributed to assumptions in the forward model and inversion scheme. We examine a suite of models with either spatially uniform or variable interseismic creep on faults and either low or high asthenosphere viscosities to examine the effect of model assumptions on slip rate estimates. In particular, we are developing a new block models for southern California in which faults are modeled in an elastic lithosphere overlying a viscoelastic asthenosphere. The 3D viscoelastic block model is an extension of 3D elastic block models previously developed. Our 3D viscoelastic cycle model also uses the concept of a ¡§steady-state¡¨ deformation field in the absence of fault locking, similar to elastic block models. The interseismic velocity field is obtained by modifying the steady-state velocity field with contributions from interseismic fault locking and periodic earthquakes on faults. Preliminary results suggest that models with low asthenosphere viscosity infer low slip rates on the above-mentioned faults, consistent with the results of elastic block models. Decreasing the asthenosphere viscosity raises the estimated fault slip rates.
G21A-0657
Global Optimization for the Determination of Lithosphere Rigidity Variations Using Geodetic Strain with Application to Southern California
Interseismic velocity field provided by geodetic methods in tectonically active areas is often interpreted using the concept of thick elastic blocks cut by slipping faults at depth. This approach is purely kinematical and the best adjusment between observed and modeled velocity fields is obtained by searching the fault slip rate distribution along the block boundaries below the seismogenic depth. Here, we adopt a different viewpoint in which interseismic strain occurs in response to remote forces or velocities imposed by the plates. In this case, lithosphere and faults control interseismic strain rate thanks to their rheological properties. If one assumes that faults are fully locked within the seismogenic zone, strain rate variations must occur in response to lateral variations in the elastic rigidity of the lithosphere. In line with this view, low rigidity zones should lead to high interseismic strain while high rigidity areas remain weakly deformed. We present a mathematical method to approximate the observed GPS velocity field by a mechanical model displaying a horizontally variable plate rigidity. We adopt an optimisation scheme that aims to minimize a cost function representing the offset between a forward elastic model and the data (interseismic velocity field). The plate rigidity is described by a small number of parameters defined on control points. An optimization algorithm then allows to minimize the cost function by computing successive direct problems and the sensitivity of the cost function with respect to control parameters. We first apply this method to search the rigidity distribution in the case of a well-posed problem for which 1) the rigidity map is a priori known 2) the data velocity field is the FEM solution of the direct problem. This approach allows us to define a parametrization suitable to describe respectively sharp and smooth rigidity variation and also to implement a strategy in which the parameter space is progressively enriched during the optimization process. We then use the southern California zone in the vicinity of the San Andreas fault to test the ability of the method to solve the rigidity distribution over an interplate area. This zone is of a special interest because it is crossed by numerous active faults. Also, it includes different tectonical areas such as the border of the Pacific plate, the San Andreas fault zone and the Mojave region. Therefore, one can expect that the analysis of the continuous GPS velocity field (CMM3) leads to contrasted rigidities across the studied domain. We discuss the rigidity distribution obtained by our method in the light of heat flow, seismology and tectonic knowledge of southern California.
G21A-0658
GPS Velocity and Strain Rate Fields in the Great Basin and California
New data from EarthScope's Plate Boundary Observatory (PBO) allow for an improved spatial resolution of
crustal motion in the Pacific-North America plate boundary zone. In addition to PBO and other GPS networks
(BARGEN, BARD, SCIGN, EBRY, CORS) we further increased densification between the northern Walker
Lane and the Garlock fault by establishing the 300+ station NEARNET network. The first NEARNET sites
were established in 2004, with the core part of the network (known as MAGNET) being measured semi-
continuously. Sites on the fringes of the network are measured every 1-2 years, with each occupation being
at least 3 weeks. We routinely analyze NEARNET data together with those from all publicly available
continuous sites in the western United States (and most other sites in the rest of the world) with the GIPSY
precise point positioning software, and routinely produce solutions that have phase ambiguities resolved
using Ambizap. Position time-series are relative to stable North America and have common-mode errors
removed on both the continental scale as well as on the scale of the Great Basin, western United States. We
present velocities for all sites with time-series longer than 2 years, together with velocity uncertainties that
account for the presence of colored-noise. We aim to remove all significant co- and post-seismic deformation
from recent earthquakes that significantly affect the estimate of the secular motion. The dense GPS velocity
field in California and the western Great Basin provides the critical input to refine the crustal deformation
field, and hence improve understanding of western U.S. lithospheric dynamics and seismic hazard. Here we
present a high-resolution strain rate map for all of California and the Great Basin using a spline interpolation
and least-squares fit to the GPS velocities. For this model we also include published GPS velocities obtained
from campaign-style measurements. The resolution of our model allows, for the first time, to identify strain
localization across individual strands of fast-slipping faults. We also include geologically estimated fault slip
rate information to better relate strain to fault structures which is especially important to relate geodetic strain
to faults in the eastern part of the plate boundary zone. To illustrate how the strain rate map can provide a
base map of interseismic strain accumulation for InSAR studies, we present an InSAR equivalent fringe-map
of expected line-of-sight rates from our model. An additional analysis of the statistics of seismic versus
tectonic moment release based on our strain rate model is presented elsewhere (Torres et al., Statistics and
Correlations of Seismic and Tectonic Moment Rate in California and the Great Basin, session NG04).
http://geodesy.unr.edu/networks/
G21A-0659
Transpression-Induced Structural Deformation in the Southern Mecca Hills, California
Off-fault deformation, evidenced by faulting and folding, plays a role in the strain accommodation along transpressive plate margins. The Mecca Hills, located adjacent to the southern San Andreas Fault in the Salton Trough, display folded and faulted Pliocene-Pleistocene sediments. The region is undergoing simple- shear-dominated transpression due to different azimuths of the Mecca Hills segment of the San Andreas Fault and the Pacific-North American plate motion vector. To investigate the role of off-fault deformation, we mapped faults and folds exposed in the southern Mecca Hills near Box Canyon. Two distinct deformation domains are apparent, defined by the predominant orientation of folds and located in elongate zones subparallel to the fault-zone boundaries: a wrench- dominated homogeneous strain domain with folds oriented obliquely to the San Andreas Fault, and a partitioned strain domain with fault-parallel folds. The two deformation domains identified in this study are consistent with, and appear to be an extension of, those found by previous researchers in the northern and central Mecca Hills. The region displays oblique en echelon folds which indicate an area of homogeneous strain transpression and fault-parallel folds and faults which indicate an area of partitioned strain transpression. The two deformation domains across the region indicate that bimodal transpression is consistent throughout the Mecca Hills. A mechanically weak layer is found throughout the partitioned strain transpression domains, suggesting a detachment surface is associated with partitioned strain transpression. We determine the amount of strain partitioning expected for this region by incorporating regional values for instantaneous strain and plate convergence angle [1] as well as regional simple shear strain and fault- parallel slip [2]. Results from these calculations suggest that the southern Mecca Hills shear zone may be accommodating up to 64% of the total strain along the southern Mecca Hills segment of the San Andreas Fault. The fault-normal shortening for the region suggested by this method (65%) agrees favorably with that determined from our field observations (66%). [1] Teyssier, C., Tikoff, B., and Markley, M., 1995, Geology, 23, 447–450. [2] Gray, M., 2000, M.S. Thesis, Calif. State Univ., Long Beach.
G21A-0660
Neotectonic and Geodetic Observations Bearing on the Structural Development of the Walker Lane, Western Nevada, USA.
Geodetic observations indicate the Walker Lane accounts for upwards of 20-25 percent of relative Pacific- North America plate motion across the San Andreas plate boundary. Cumulative right-lateral offset across the Lane ranges from about 30 to greater than 60 km. Yet, at latitudes between the northern end of Walker Lake and Reno, there exist no well-developed strike-slip faults oriented to accommodate the motion. Rather, deformation currently and in the past appears to be expressed in the development of a set of major left- stepping normal faults and basins extending from Walker Lake in the south to Lake Tahoe in the north. In this sense, the basins may be viewed as an en-echelon arrangement of crustal scale tension gashes. We report on our ongoing effort to resolve the relationship of geodetic patterns of strain accumulation to the pattern of strain release as expressed in the development of the basins and the implications of this relationship to the tectonic development of the Walker Lane.
G21A-0661
Accommodation of Right-lateral Shear Along the Northwest Boundary of the Snake River Plain, Idaho
The northwest boundary of the Snake River Plain (SRP) is a transition from range-bounding normal faults in the Centennial tectonic belt (CTB) to the topographically low and volcanic-dominated province of the SRP. Within the CTB, the northern and central segments of three prominent NW-trending normal faults are seismically active, but their activity decreases southward toward the SRP. Deformation in the SRP is associated with infrequent small magnitude (M<2) microearthquakes and NW-trending volcanic rift zones that result from basalt dike intrusion since 6 to 4 Ma. GPS results from 1994-2008 indicate that the CTB extends via normal faulting at a rate that is an order of magnitude greater than the extension rate in SRP. Right-lateral shear in a NE-trending transition region, what we call the Centennial Shear Zone (CSZ), is a geometric consequence of the different strain rates. The inferred slip rate across the CSZ increases from ~1 mm/yr in the southwest to ~2 mm/yr in the northeast. The shear may extend as far as the Yellowstone Plateau at an even faster rate. Estimated slip rates within the CSZ suggest up to 2 km of lateral offset in a million years. In the CSZ, investigators have identified only NE-trending normal faults but without significant strike-slip offset on them. The most recent and larger fault offsets (15-110 ka) are associated with the NW-trending, range-bounding normal faults. To assess how right-lateral shear is accommodated, we use block models to invert GPS horizontal velocities with earthquake slip vectors and fault slip rates for relative block motions. The predicted slip along the block faults is evaluated with geologic observations such as fault linkages within segment boundaries. We explore two possible alternatives for how right-lateral shear may be accommodated; 1) strike-slip earthquakes on NE-trending faults, or 2) oblique slip along the NW-trending faults, which is indicative of a regional transtensional environment.
G21A-0662
Constraints from GPS on Block Kinematics of the Transition between the Southern Walker Lane and the Basin and Range Province
The southern Walker Lane (SWL) is a part of the Eastern California Shear Zone that lies north of the Mojave region, bounded by the Garlock Fault to the south, the Sierra Nevada to the west, the Basin and Range to the east and by Mono Lake to the north. The region includes many northwest striking right-lateral strike slip and sub-parallel normal faults (e.g. Death Valley/Furnace Creek, Fish Lake Valley, Owens Valley), which together accommodate ~25% of the Pacific/North American relative motion. For many of these faults, and the system as a whole, there appears to be a discrepancy between geodetically and geologically inferred fault slip rates. Since the installation of the EarthScope Plate Boundary Observatory (PBO), and the Nevada Earthquake Response Network (NEARNET) of the University of Nevada, Reno, many recently obtained high- precision GPS data are now available to place improved constraints on the pattern and rates of crustal deformation of this region. In this study we use a block modeling methodology to estimate block motions and fault slip rates from GPS velocities of PBO, NEARNET and BARGEN continuous sites. Time series were obtained from raw RINEX data that we processed using the GIPSY-OASIS II software from the Jet Propulsion Laboratory together with the Ambizap software for ambiguity resolution. We have additionally included earlier published campaign-style velocities, in those areas where we do not have better coverage from other continuous/semi-continuous networks. Geologic slip rates have been obtained from the published literature. We solve for the motion of blocks using the GPS velocities that have been adjusted based on the viscoelastic modeling to estimate long term motion. To evaluate the consistency between the geologic and geodetic data, we compare long-term fault slip to slip rates inferred from geodetic results obtained over <10 years. We account for transient earthquake cycle effects by modeling the viscoelastic postseismic relaxation following major historic earthquakes in the region. In particular we model the 1999 Hector Mine, 1992 Landers, 1952 Kern County, and 1872 Owens Valley earthquakes. GPS velocities adjusted for transient effects indicate that there is a distinct NW trend in the motions of the blocks with rates decreasing to the east. However, deformation rates are greater than zero east of the SWL in the Basin and Range. The preliminary results obtained from the block model indicate significant slip at the easternmost edge of the model, in the vicinity of Yucca Mountain and the Stateline fault.
G21A-0663
The accommodation of distributed crustal strain across the northern Basin and Range on active faults: A paleoseismic study
The Great Basin encompasses over two thirds of the Pacific/North American plate boundary and accommodates up to 25 percent of the relative plate motion by active faulting. Although the majority of the deformation within the Great Basin is concentrated along its margins, a small amount of geodetically measured strain is observed between the Central Nevada Seismic Belt and the Wasatch Front. In this area, comparison between short and long-term patterns of strain accumulation and release has been difficult due to slow deformation rates and insufficient paleoseismic data. Thus, we performed geologic studies along 10 rangefront faults distributed across US HWY 50 to document the amount and timing of late Pleistocene displacements. The data include geologic maps of Quaternary deposits and fault traces, paleoseismic trenches, scarp diffusion analyses, and soil characteristics in displaced alluvial surfaces. The observations are combined with paleoseismic information from previous studies to estimate a net long-term extension rate and compared to geodetic measurements. The results indicate that the rate of strain release recorded by faulting (about 1 mm/yr) over the late Pleistocene is, to first order, similar to the geodetic rate of strain accumulation. The agreement in extension rate estimates over different time scales implies that relatively slow extensional deformation has been operative in the Great Basin east of the Central Nevada Seismic Belt through the late Pleistocene. The geomorphic pattern of late Pleistocene ruptures characterized by left stepping enechelon strands with north easterly trends and a divergence of rupture orientations of about 20- 40 degrees from their respective range orientations may be the geomorphic expression of a component of right lateral shear within the interior of the Great Basin.
G21A-0664
Geodetic Observations From the Region Surrounding the M 5.2 Mt. Carmel, Illinois Earthquake
We present new results from a GPS geodetic network in the southern Illinois Basin, including a post-seismic survey in the aftermath of the M5.2 Mt. Carmel Illinois earthquake. A 56-station regional network at ~30 km spacing in SW Indiana, S Illinois and W Kentucky is augmented by continuous data from IGS, GAMA, and CORS GPS stations. We also present results from a densified, 35-station GPS network (~10 km spacing) in the Fluorspar district of southernmost Illinois, located near near a complex transitional zone between the Wabash Valley and New Madrid seismic zones. The region is traversed by steep-angled, basement-penetrating faults and characterized by moderate (M ~3 to 5.5) earthquakes. The most recent of these events was the April 18, 2008 M 5.2 event, located close to the New Harmony Fault at ~14 km deep near Mt. Carmel, Illinois. We combine data from the regional network with GPS observations from five stations near the epicenter of the April 2008 earthquake. Predictions based on the depth and mechanism of the earthquake suggest horizontal coseismic motions of < 1mm at most network stations. Observed data suggest small, but marginally significant displacements as compared to block motions in the area. Results from the regional network show highly improved position and velocity estimates of these campaign sites relative to previous campaign measurements, with station velocities suggesting systematic northwestward motion of about 0.5-0.7 mm/yr with respect to the Stable North American Reference Frame. We then investigate strain patterns using models to explain tectonic deformation within the Wabash Valley. We use an elastic block modeling approach, supplemented by continuum-based methods to explain variable strain between GPS stations. Block models which assume boundaries along the Cottonwood Grove-Rough Creek Graben (CGRCG) and the WVFS indicate marginal block velocities with possible strike- slip motion along the WVFS, and E-W motions along the CGRCG. We also present results from eight years of GPS observations (2000-2008) from the dense Shawnee network, which appear to be consistent with the regional strain models from the regional network. We also explore geodynamic models that integrate seismicity and crustal deformation in the region.
G21A-0665
Accomodation of the deformation in the Cocos-Caribbean-North America triple junction area, from GPS meaurements
Northern Central America is located in a complex zone of interaction between three major tectonic plates: the North American (NA), the Caribbean (CA) and the Cocos (CO) plates. While the CO plate is subducting under the NA and CA plates along the Mid-American trench, the on-land relative motion between the NA and CA plates is mainly accommodated by the east-west trending left-lateral Polochic-Motagua fault system. We use GPS data to characterize the deformation in this CO-CA-NA triple junction area in terms of interactions and relative contributions of the different structures (Polochic and Motagua faults, a series of north-south grabens south of them and the Mid-American Volcanic Arc, MAVA), and coupling at the subduction interface. We analyse GPS-campaign data from different networks in Guatemala (24 sites measured in 1999-2003- 2006), Salvador (3 sites measured in 2003 and 2006) and Chiapas, southern Mexico (8 sites measured each year from 2002 to 2004). Data are processed and combined using the GAMIT/GLOBK softwares. The obtained velocity field is first fitted using simple elastic models. The Motagua fault seems to accomodate most of the present left-lateral motion between CA and NA, with less than 30 procent taken by the Polochic fault despite its comparable surface morphology, historical seismicity and microseismic activity (that we recorded during a 6 months experiment in 2005). This may suggest complex mechanical interactions between the Motagua and Polochic faults at the scale of several seismic cycles. A single fault model, centered on the Motagua fault, locked at a depth of 15 km, indicates a slip rate decrease from eastern (20 mm/yr) to central Guatemala (10 mm/yr) towards the CA-CO-NA triple junction (0 mm/yr?). This decrease seems to be consistent with east-west extension rates estimated across the Ipala and Guatemala city grabens south of the Motagua fault. We also observe a right-lateral movement across the MAVA: 15mm/yr in western Salvador and 10mm/yr in eastern Guatemala. This movement is consistent with that observed along the nicaraguan and costa-rican volcanic arcs. To take into account the rotation of blocks and the elastic deformation localized on faults at their boundaries in this area, we finally use the 3D-elastic inverse model DEFNODE (McCaffrey 2002) to fit the GPS-derived deformation field. Our 4 blocks (CO,CA, NA and a microplate in between the MAVA and the CO subduction interface) seems to suggest a coupling decrease at the subduction interface from Chiapas to Guatemala, although not fully resolved with the present data set.
G21A-0666
Deformation of Mexico from continuous GPS: 1993 to 2008
We report and interpret Global Positioning System measurements at 15 stations in Mexico from early 1993 to mid-2008 in the context of of North American and Pacific plate motions and earthquake cycle effects from the Mexican subduction zone. New velocities that we derive from high-quality, code-phase GPS data from early 2003 to mid-2008 differ significantly from velocities that we previously estimated from lower-quality, phase- only GPS data for the period 1993 to mid 2001. On mainland Mexico, the updated GPS coordinate time series define three geographic station clusters with different temporal behaviors. Stations far from the Pacific coast exhibit linear behavior consistent with steady-state tectonic processes such as plate motion and interseismic strain accumulation. The motions of stations between the volcanic belt and Pacific coast oscillate between 1-yr to 2-yr-long periods of landward motion and several-month-long periods of reverse-sense trenchward motion, consistent with repeated, deep transient slip events along the Mexican subduction interface. Finally, the motions of two stations within 300 km of the Rivera plate subduction zone are strongly influenced by the coseismic and post-seismic effects of M=8.1, Oct. 9, 1995 Colima-Jalisco earthquake and M=7.9, Jan. 22, 2003 Tecoman earthquake off the coast of Jalisco. Tests of two simple models for the present deformation of Mexico yield a root-mean-square misfit of 2.3 mm yr-1 to the velocities of sites on the mainland when they are fit by a North American plate angular velocity vector, larger than the estimated velocity uncertainties. The RMS misfit for a model that also includes the elastic effects of assumed frictional locking of the Mexican subduction interface and faults in the Gulf of California is reduced by 60% to only 0.9 mm yr-1, comparable to the velocity uncertainties. We conclude that the present deformation of mainland Mexico excluding deformation in Chiapas that is related to Caribbean-North America plate motion is well described by a combination of North America plate motion and the elastic effects of plate boundary faults.
G21A-0667
ANALYSIS OF 2007 DOME EXTRUSION AT THE VOLCAN DE COLIMA, MEXICO USING TILT METER SURVEYS
The Volcan de Colima, Mexico is located on the wetern front of the Trans-Mexican Volcanic Belt and it is considered one of the most activ volcano in Mexico. The last dome extrusion at the Volcan de Colima, Mexico occurred in 2007 without manifestation of seismicity. This extrusion is carried out using the tilt meter netz installed around the volcano edifice appearing this lava extrusion with the manifestacion of etapes of inflation tilt change accompanied by the deflation changes. The recent unrest at andesitic Volcán de Colima, México began on 28 November 1997 with a sharp increase in seismic activity and a significant shortening of geodetic lines around the volcano edifice. During this period of activity 4 lava extrusions are occured: November 1998, May 2001, September 2004 and the actual February 2007. Destruction of the lava dome occurred with a serie of explosions and emission of ash and pyroclastic flows during these activity phases. Associated with the growth of the dome variations of deformation is observed. Here we show the analysis of the last dome extrusion 2007 in comparision with the previous extrusion during the recent activity phase.
G21A-0668
Investigation of Deformation Sources for the Miyakejima, Japan Intrusive Event
On June 26, 2000, a large seismic swarm began beneath the west coast of Miyakejima, Japan, an active volcanic island in the Izu arc. After a week-long propagation of seismicity to the northwest, seismic activity continued for approximately two months in the region. The swarm was accompanied by eruptions and caldera collapse on Miyakejima, as well as continuing crustal deformation measured by GEONET, the continuous GPS network of the Geographical Survey Institute of Japan. Based on observed deformation and locations of hypocenters, it has generally been assumed that dike propagation from a magma chamber beneath Miyakejima was responsible for the swarm. Although several models of varying complexity have been proposed to explain the deformation, there is little consensus on a geologically motivated model that explains both the long-lived deformation and seismicity. In addition, this event provides an interesting opportunity to study the relationship between dike propagation and earthquake nucleation, which, though long recognized, is not well understood. We tested several models of deformation based on observed displacements, hypocenter locations, overall tectonic setting, and pre-event data. Sources of deformation considered include various modes of dike propagation, magma chamber deflation, and faulting to the northwest of Miyakejima in a zone that was seismically active before the 2000 event. In order to test the viability of these mechanisms, we utilized both inverse and forward modeling of displacements during various stages of the event. Initial results show that a model consisting of only dike opening and Mogi deflation beneath Miyakejima can fit the total deformation well, but the best-fitting model does not coincide with the seismic swarm. In addition, there are substantial differences between the orientations and magnitudes of displacements observed during the first week (maximum 4.8 cm) and those for the entire event (maximum 48.4 cm). These preliminary results indicate that a simple model of dike propagation during the first week followed by steady opening does not adequately explain the observed deformation, and more complex source geometries and evolutions are considered.
G21A-0669
Yes, There is a Northern Lesser Antilles Forearc Sliver: Results From a Decade of GPS Observations
Investigations of the discrepancy between thrust fault earthquake slip vectors and the direction of North American plate convergence and arc-normal trending normal fault systems have led previous workers to suggest that the northern Lesser Antilles arc/forearc (NLAAF) region is subject to strain partitioning and sliver motion with some component of internal deformation. Results from a decade of GPS observations in the northern Lesser Antilles and the British Virgin Islands (BVI) are used to directly constrain estimates of microplate motion. We use the observed GPS velocity field to quantitatively test between three kinematic models for the region: 1) the NLAAF moves with the Caribbean plate and does not show a significant amount of sliver motion, 2) the NLAAF shows rigid sliver motion that differs statistically from Caribbean plate motion, or 3) the NLAAF shows motion consistent with an internally deforming forearc sliver. Our analysis of the GPS data indicates that statistically significant northwest directed motion of a forearc sliver is occurring in the region. The observed velocity field also supports internal deformation in the southern part of the sliver where the majority of the extensional fault systems occur. Boundary parallel motion within the NLAAF ranges from ~1-2 mm/yr in the southern region to ~5 mm/yr in the northern region. These rates are in agreement with past studies of plate convergence and earthquake slip vectors for the central part of the NLAAF, but are slower than past estimates of possible sliver motion in the northern region (BVI) indicating that oblique convergence in this area is only partially partitioned. Our velocity field shows predominantly boundary parallel motion along the arc, while islands in the southern forearc region show predominantly arc- normal strain accumulation of ~3 mm/yr.
G21A-0670
Global Positioning System (GPS) Determination of Motions, Neotectonics, and Seismic Risk in Trinidad and Tobago
The twin island nation of Trinidad and Tobago is located in the actively deforming Caribbean-South American (Ca-SA) plate boundary zone. Recent GPS (Global Positioning System) studies have accurately determined the relative plate motion between the Caribbean plate and South American plates: ~east-west, dextral motion at 20 mm/yr. Earthquakes do not clearly mark many of the active faults in Trinidad. A low-precision triangulation-to-GPS comparison at 23 sites and paleoseismology showed that significant strike-slip faulting is probably occurring on the Central Range Fault (CRF). The lack of recent seismic activity on the CRF may indicate that it is elastically locked and storing motion. We study Trinidad's neotectonics using new GPS data from 19 high-stability campaign sites that were built and measured in 2005, then measured again in 2007; for a few additional sites we have data that go back from 2007 to 1994. These new data will allow us to test and refine the previous lower-precision geodetic results to better quantify the rate of slip across the CRF and its mechanical behavior. We compiled, and then processed the data using GIPSY/OASIS II (Release 5.0) software at the University of Miami RSMAS Geodesy Lab. We find that, in a South American reference frame, sites north of the CRF move at about 20 (±1-5) mm/yr; sites south of the CRF are stationary (±1-8 mm/yr). Tobago site velocities are slightly oblique to overall Caribbean plate motion due to a major (Magnitude 6.7) earthquake that occurred off Tobago's south coast in 1997. Our new results support the hypotheses that the CRF is the principal active strike-slip fault in Trinidad (i.e., is the current Ca-SA plate boundary). We are fitting locked fault (elastic dislocation half-space) models to the data which will allow us to look more closely at the mechanical behavior of the CRF to further test whether it is locked or creeping to help evaluate its seismic risk.
G21A-0671
Geologic Estimates of Northeastward Andean "Escape" for the Last 0.5 Ma
CASA GPS measurements (Trenkamp et al., 2002) suggest that a large part of the northern Andes is "escaping" to the northeast relative to stable South America at a rate of 6±2 mm/a. The displacement rates of seven sites in Venezuela, Colombia, and Ecuador are statistically identical at the 2 sigma confidence level. This study compares this GPS rate to 19 published field geologic estimates of displacement rates, such as displaced glacial moraines and offset pyroclastic flow. Dated displacements compiled in this study were obtained from the Gulf of Guayaquil, Pallatanga, Chingual-La Sofia, and Cayambe-Afiladores-Sibundoy fault systems in Ecuador and southern Colombia and the Bocono fault system in Venezuela. Right-lateral slip estimates on the individual fault segments range from 1.4 mm/a to 9.7 mm/a. The mean estimated geologic slip rate for the last 86,000 BP is 7.6 mm/a with an R2 value of 0.97. This estimate is very similar to the GPS measurements of present day motion at the 2 sigma level. Three estimates indicate that slip rates of 4 to 6 mm/a continued back to 500,000 BP. No geologic slip estimates have been reported for Ecuador prior to that time period. The "escape" of the Northern Andes is believed to be a result of increased coupling between the obliquely subducting Nazca plate and the overriding South American plate due to the subduction of the Carnegie ridge in the Ecuador-Colombian trench. If this is correct, the slip estimates for the northern Andes suggest that the Carnegie Ridge arrived at the trench prior to 500,000 BP.
G21A-0672
GPS Versus Seismological Observations in two Seismogenic Zones in the Adria-Alps- Pannon System; Block Motion vs. Diffuse Deformation, Increased Earthquake Potential vs. Aseismic Slip
The tectonic activity, seismicity and the associated seismic hazard is highly variable in the Adria-Alps-Pannon region. The engine of the system is the Adria microplate that compresses a puzzle of crustal blocks towards the European Platform. Based on seismicity and data of continuous and campaign style GPS measurements between 1991 and 2007 we investigated the existence of different blocks and their present kinematics. At the resolution and signal level we have, deformation seems to be more diffuse and block motion is no longer recognizable over the Pannonian basin towards the Carpathains. Although towards the basin seismicity decreases to moderate, the vulnerability is still high, as three capital cities are located near to the two most active seismic zones in this subregion. Each cities and their suburbs produce about 30- 40 % of the GDP of the respective countries. In the second par of our analysis these two seismically active areas, the Mur-Murz and Central Pannonian zones, are investigated. Uniform strain rates and relative displacements were calculated for these regions. The GPS data confirm the mostly left lateral strike slip character of the Mur-Murz fault zone and suggest a contraction between the eastward moving Alpine-North Pannonian unit and the Carpathians. The computation of the seismic strain rate was based on the Kostrov summation. The averaged unit norm seismic moment tensor, which describes the characteristic style of deformation, has been obtained by using the available focal mechanism solutions, whereas the annual seismic moment release showing the rate of the deformation was estimated using the catalogs of historical and recent earthquakes. Our analysis reveals that in both zones the geodetic strain rate is significantly larger than the seismic deformation. Based on the weakness of the lithosphere, the stress magnitudes and the regional features of seismicity, we suggest that the low value of the seismic/geodetic strain rate ratio in the Central Pannonian and probably also in the Mur-Murz zone can be attributed to the aseismic release of the prevailing compressive stress and not to an overdue major earthquake.
G21A-0673
Present-Day Kinematics of the Central Mediterranean Plate Boundary Region from Large GPS Network Analysis Using the Ambizap Algorithm
The large, recent increase of continuous GPS (CGPS) stations in the Central Mediterranean plate boundary
zone offers the opportunity to study in detail the present-day kinematics of this actively deforming region.
CGPS data from scientific and commercial networks in the Italian region is now available from more than 350
stations, including more than 130 from the RING network
G21A-0674
Active Tectonics and Deformation Pattern in the Central Adriatic Sea: Evidence by Integrating GPS Data, Earthquake Focal Mechanisms and Seismic Profiles
The central Adriatic represents the foreland of the Dinarides and the Apennines, and is characterized by the occurrence of a poorly defined belt of moderate seismicity that connects the two fold-and-thrust belts, just north of the Gargano promontory (Italy). The pattern and nature of the tectonic deformation that is responsible for the observed seismicity are the subject of an ongoing debate. This contribution aims at integrating different sets of data, namely medium-high resolution multichannel seismic profiles, GPS velocities and earthquake focal mechanisms, in order to define the deformation pattern of the region. GPS data indicate a limited NE shortening across the central Adriatic region, in agreement with the limited seismic activity and with compressional focal mechanisms. The hypocentral depth of the larger events is typically around 15-20 km, suggesting that the basement is likely involved in the deformation. Contractional structures, trending NW-SE to NNW-SSE and NE-SW to ENE-WSW, represent the most common type of deformation on seismic profiles. The Tremiti belt, a narrow NE-SW belt located north of the Gargano, shows Plio-Quaternary growth strata all the way to the sea floor, in agreement with the NW-SE shortening from GPS and focal mechanisms, and merges to the N to a broad NW-trending belt of diffuse but limited deformation. In some instances the WNW-ESE folds seem to be the lateral prolongation of the thrust-related folds occurring in the southern part of the Dinarides front, where contractional deformation is still active and the direction of the frontal folds swings from NW-SE to almost E-W. The observed pattern of deformation can be interpreted as an example of active foreland tectonics. We use GPS velocities and elastic block modeling to explore the relationships between the observed deformation zones and the kinematics of micro-plates and crustal blocks.
G21A-0675
Constancy of Strain Release Rates Along the North Anatolian fault
The behavior of major active faults at various temporal and spatial scales is one of the most fundamental, unresolved problems in modern tectonics. Determining the degree to which fault loading and strain release rates are constant (or non-constant) and documenting past earthquake occurrences are key approaches for understanding this phenomenon. We have employed geomorphic mapping, Quaternary dating methods, and paleoseismic trenching to generate fault rate and earthquake age data that help provide a better understanding of the North Anatolian fault's behavior in various temporal and spatial scales. Specifically, we mapped offset geomorphic markers along the North Anatolian fault and used cosmogenic nuclide (10Be and 36Cl) and radiocarbon (14C) dating methods to constrain the ages of these features. Using these and other published data, we also constructed one of the first compilations of strain-release rates for the North Anatolian fault. Our compilation of these rate data reveals a constant slip rate of ~15-20 mm yr-1 over time scales of 103-105 years. This result, however, is slower than the geodetically constrained slip rate of 25 ± 2 mm yr-1 (Reilinger et al., 2006), possibly indicating a strain transient. In addition to our slip rate studies, we performed paleoseismologic trenching on the eastern part of the North Anatolian fault at the village of Lorut. Our results from this site demonstrate a relatively regular occurrence of large earthquakes along this stretch of the North Anatolian fault. We attribute the relatively constant strain release rates and regular earthquake recurrences to the mechanical simplicity of the Anatolian-Eurasian plate boundary in northern Turkey, which is dominated by the slip on the structurally mature North Anatolian fault.
G21A-0676
New geodynamic model for the South Caspian surroundings based on numerical modeling constrained by GPS and geological data
Over the last decade, GPS measurements in Iran, Azerbaijan, and adjacent regions have provided a direct quantification of the displacements of the earth surface in that region of the Arabia-Eurasia collision zone. We present an up to date velocity field in the southern Caspian region of the collision zone. A striking feature of the velocity field is the change in the azimuth of the velocity vectors across the Tabriz fault. Roughly northward motion south of the Tabriz fault changes to NNE motion north of the fault, with respect to Eurasia. We interpret the velocity field using an elastic block model to obtain an accurate kinematic model of this region of the collision zone. We use this kinematic model to constrain the boundary conditions for dynamic modelling. Our modelling studies indicate that indenter models can not account for the change in orientation of the velocity field of the NW Iran, the Lesser Caucasus and the Kura Basin, and that additional forcing is needed. We suggest that subducted oceanic crust beneath the Caucasus and the onset of subduction beneath the Apsheron-Balkan region could be contributing to the dynamics of this part of the collision zone.
G21A-0677
Utilizing Information Technology Infrastructure for Tectonic Studies in Turkey
There is an overflow in all of the branches of science today, especially in Earth sciences. Geoscientists have performed geodetic observations for monitoring both local and regional crustal movements by establishing microgeodetic networks along plate boundaries on the western part of NAFZ in Turkey over three decades. Geodesy Department of Kandilli Observatory and Earthquake Research Institute of Bogazici University has started the studies of monitoring horizontal crustal movements on the western part of NAFZ in 1990. Three geodetic control networks were established in the regions which have different seismotectonic features to monitor crustal displacements. There are also several ongoing and completed projects which are being conducted by scientists from the universities and the other research institutes (e.g. General Command of Mapping, ITU, TUBITAK, MIT) for the region of interest. There are an increasing number of data inferred from these tectonic studies and also geophysical data such as focal mechanims and other. For rapid analyze of these huge data and obtaining results, we need to use Information Technologies (IT). Use of IT in tectonic studies is necessary for efficiency, and to avoid the repeated efforts. This study covers the concepts of the information technology tools for tectonic studies and identifies the benefits and challenges associated with their implementation. Besides, it presents an application for speeding up the assessment of results from tectonic studies.
G21A-0678
Strain distribution in the East African Rift from GPS measurements
Rifting of continental lithosphere is a fundamental process that controls the growth and evolution of continents and the birth of ocean basins. Most rifting models assume that stretching results from far-field lithospheric stresses from plate motions, but there is evidence that asthenospheric processes play an active role in rifting, possibly through viscous coupling and/or the added buoyancy and thermal weakening from melt intrusions. The distribution of strain during rifting is a key observable to constrain such models but is however poorly known. The East African Rift (EAR) offers a unique opportunity to quantify strain distribution along and across an active continental rift and to compare a volcanic (Eastern branch) and a non-volcanic (Western branch) segment. In 2006, we established and first surveyed a network of 35 points across Tanzania and installed one continuous station in Dar Es Salaam (TANZ), followed in 2008 by a second occupation campaign. We present a preliminary velocity field for the central part of the EAR, spanning both the Western and Eastern rift branches. We compare our results with a recent kinematic model of the EAR (Stamps et al., GRL, 2008) and discuss its significance for understanding rifting processes.
G21A-0679
Geodetic Transect in Central Nepal
Presented here are the preliminary results from the installation of six permanent, high precision GPS receivers in central Nepal in June 2008. The purpose of this installation is to monitor ground movement along major faults in the Himalaya. The GPS stations define a roughly south-to-north transect across three major Himalayan structures, the Main Frontal Thrust, the Main Boundary Thrust and the Main Central Thrust. The Himalaya is the largest and most active continental convergent plate boundary in the world. It is the archetype for understanding the development of continental collision mountain belts, yet the processes that are occurring within the Himalaya are debated within the scientific community. In particular, researchers debate whether all the strain is accommodated along the southern topographic front (Main Frontal Thrust) or whether out-of-sequence faulting and reactivation of the Main Central Thrust Zone is occurring further to the north. Intensified monsoon precipitation is concentrated along the Main Central Thrust Zone and may result in a zone of rapid denudation and rapid rock uplift between the Main Central Thrust and the South Tibetan Detachment. Conversely the steeper topography of the High Himalaya could be due to steepening of the crustal ramp, rather than out-of-sequence thrusting. A better knowledge of Himalayan structure will also allow more detailed assessment of earthquake hazard. Elastic dislocation routine models of Himalayan tectonics have been made to simulate surface movement under several scenarios for active geologic structures. Comparison of the actual velocities with modeled velocities offers a first-cut assessment of which structures seem to be accommodating the most strain.
G21A-0680
Geodetic Tying of Antarctica and India With 10 Years of Continuous GPS Measurements for Geodynamical and Strain Accumulation Studies in the South of Indian Peninsula
To holistically understand the geodynamical and crustal deformation processes between India and Antarctica, two global networks (IND and ANT) have been chosen. The objective is to geodetically connect the two continents. The IGS Station at Diego Garcia (DGAR) is the common station between the two networks. 10 years of data from 1997 to 2007 were used. By these global networks' analyses, the stations HYDE in India and MAIT at Antarctica are geodetically tied through the station DGAR. Very long baselines have been estimated from HYDE and also from Kerguelen (KERG) to other chosen IGS stations in and around India and Antarctica. Our analysis and results using ANT network show an increase in the baseline lengths between Kerguelen in Antarctic plate and other stations such as SEY1, DGAR and COCO and shortening of baseline lengths between HYDE in Indian plate and all these above stations using IND network. The analysis using ANT network also shows lengthening of baselines from Kerguelen to the sites Yaragadee (YAR1) and Tidbinbilla (TID2) in Australian plate; and Seychelles (SEY1) in Male plate, COCO in the diffuse plate boundary between India and Australia and DGAR in Capricorn plate at the rates of 5.3cm/yr, 3.8cm/yr, 5.6mm/yr, 3.03 cm/yr and 5.5 cm/yr respectively. The high rate of movement of COCO Island in comparison to Seychelles could be the result of excessive strain accumulation due to the Indo-Australia diffuse plate boundary forces acting upon this region. The estimated elastic strain accumulation shows an increasing trend of 1.27x 10-8 yr-1 in the south of Indian peninsula. Our results show the precision of approximately 3-4mm (North), 5-6 mm (East), and 10-12mm (vertical) for the estimation of site coordinates. These results provide new information on the direction and rate of Indian plate motion, the driving mechanisms of Indian plate and intraplate seismicity of the Indian Ocean on the whole.
G21A-0681
Episodic GPS Campaigns at Lakshadweep Islands Along the Chagos-Laccadive Ridge to Investigate the Inferred Continental Flexure in the West of India and the Non-Rigidity of the Oceanic Part of the Indian Plate
The Episodic GPS campaigns were initiated at Lakshadweep islands for the first time in India by National Geophysical Research Institute (NGRI) with the objectives of refining the already estimated strain accumulation in the south of Indian peninsula, reaffirming the inferred continental flexure in the south-west of India, and investigating the rigidity of larger oceanic part of the Indian plate. To start with two sites Kavaratti and the northern most island Chetlat were chosen. With the new state-of-the-art GNSS receivers, which could track 30 GPS and 11 GLONASS Satellites with 5° elevation mask, GPS measurements were carried out simultaneously at both Kavaratti and Chetlat for two weeks during March 2007 and repeat measurements were carried out recently in these islands. In February 2008 the southern most island Minicoy also was included in the experiment design and simultaneous GPS measurements were carried out in both Minicoy and Kavaratti. The acquired data was processed in the latest ITRF 2005 reference Frame. The site coordinates of Kavaratti, Chetlat and Minicoy and also the baseline lengths between Hyderabad and these three sites were estimated in the Global Network Solution. The methodology involved, the results of estimated site coordinates and the baseline lengths between Hyderabad and these islands are discussed in this paper. The estimated baseline length between Hyderabad and Kavaratti is 991,303.3067± 0.0082m. The 8mm accuracy in the estimation of baseline length shows the quality of data processing. The estimated baseline length between Hyderabad and Chetlat is 892,216.5594± 0.0040m. The estimated baseline length between Hyderabad and Minicoy is 1171,071.8777± 0.0065m. The estimated accuracy of the baseline length is in the range of 4to8mm. These studies across a 1,200-km-long "strain gauge" that is optimally oriented almost parallel to the compression seen on the land would enable the understanding whether this is due to the Himalayan collision, or the extension of the Capricorn-India diffuse boundary that could have extended this far north.
G21A-0682
Geodetic Constraints for Back-Arc Spreading Across the Mariana Trough
The Mariana Islands Arc (MA) in the eastern margin of the Philippine Sea plate (PH) provides the best example for the studies of a formation of island arc and a mechanism of back-arc spreading. In the east of the arc, the Pacific plate is subducting at the Mariana Trench with a steeply dipping slab. In the west, the back-arc spreading is in progress at the Mariana Trough. Kato et al. (2003) determined stationary horizontal displacements of six islands located at 13.6-18.7°N from several GPS campaign measurements since 1992 to 1999. By comparing GPS velocities with the predictions from a global plate model, they showed that MA is moving apart from PH at a rate ranging from 15 mm/yr at 18.7°N to 45 mm/yr at 13.6°N, which are consistent with geological and geomagnetic estimates for the spreading rates. Also they found a north-south, along-arc extension. However, the Euler pole of the MA-PH relative motion, which is estimated at around 21.5°N, is significantly south of the geographical pole where the width of the back-arc basin converges. We conducted additional GPS campaign measurements at three northern islands aligned from 19.7 to 20.5°N in 2003 and 2004 to re-determine the MA-PH relative motion. As far as we assume MA to be a single block, the estimated location of the MA-PH Euler pole and the spreading rates across the Mariana Trough do not differ significantly from those of Kato et al. (2003). However, residuals between the observations and model predictions become systematically larger in the northern islands. It is not clear currently whether the larger residuals in the northern islands have resulted from inaccurate velocity estimates or from a wrong assumption that MA consists of a single block. We will revisit the MA-PH relative motion using GPS data from the latest campaign measurements conducted at nearly all islands in summer 2008.
G21A-0683
Recent Observational Results of Seafloor Crustal Deformation Along the Suruga-Nankai Trough, Japan
The Suruga-Nankai Trough is one of the active plate boundaries in the world. The Philippine Sea plate is subducting beneath the Amurian (Eurasian) plate along the tough, and major subduction earthquakes, Nankai and Tonankai earthquakes, have repeatedly occurred with intervals of about 100-150 years. The 1944 Tonankai and 1946 Nankai earthquakes are the most recent significant earthquakes along the trough. Therefore, the 50-years probabilities of the next major earthquakes are estimated at 80-90% by Headquarters for Earthquake Research Promotion, Japanese Government. It is, therefore, necessary to start monitoring crustal deformation above the source regions of the major earthquakes where in the ocean area. We developed a new system composed of the precise acoustic ranging and kinematic GPS positioning techniques for monitoring of seafloor crustal deformation [Tadokoro et al., 2006, GRL; Ikuta et al., 2008, JGR]. We had installed seven seafloor benchmarks for acoustic ranging at the Suruga-Nankai Trough region between 2002 and 2004. The water depths at the benchmarks are about 800 to 2000 m. We installed a new seafloor benchmark at the eastern margin of the Kumano Basin on June 23, 2008. Three seafloor benchmarks had been aligned perpendicular to the trough axis. In contrast, the new benchmark was installed eastward relative to the pre-installed benchmarks, and we can monitor lateral variations in crustal deformation at the region. We started the repeated measurements at four benchmarkes (two at the Kumano Basin named KMN and KMS, and the other two at the Suruga Bay named SNW and SNE) in 2005. The number of times we have measured are seven, eleven, three and nine times at KMN, KMS, SNW and SNE, respectively. Recent results of the repeated measurements show the following horizontal velocities with relative to the Amurian Plate: 6.4 cm/yr, N86W at KMN; 5.3 cm/yr, N71W at KMS; 3.3 cm/yr, N57W at SNE. The errors of the horizontal velocities are 1-3 cm/yr. Unfortunately, we have not detected any velocity at SNW benchmark because of the insufficient number of measurements as of August 2008. The orientations of the horizontal velocities is almost consistent to those derived from the on-land dense GPS observation network, GEONET of Geographical Survey Institute, Japan. Temporal and spatial variations in sound speed are major source of error in the benchmark positioning. Now we are developing other system with plural sea-surface transducers for positioning sea-bottom benchmarkes with acoustic tomography method to reduce the error caused by the sound speed variations. Acknowledgments: We are grateful to the captain and crews of R/Vs "Hokuto," Tokai University and "Asama," Mie Prefecture Fisheries Research Institute, Japan. This study has been promoted by "Research Revolution 2002" and "DONET project" of Ministry of Education, Culture, Sports, Science and Technology, Japan.
G21A-0684
Inversion method counting the spatial variation of sound speed structure in the measurement of Ocean Base Crustal Deformation
We developed a new geodetic method of monitoring crustal deformation on the ocean floor. The measurements were conducted at two sites beneath the Pacific Ocean, near the Nankai trough, where the Philippine Sea plate subducts into the Pacific plate. We have conducted repetitive measurement for about 5 years at the two sites, one is in Suruga Bay and the other at the Kumano Basin. At each survey site, three seafloor transponders, whose positions were repetitively measured, were deployed to define a benchmark unit. We combined GPS and acoustic technique to determine the positions of the transponders. A surface vessel, whose position was precisely determined by kinematic-GPS technique, drifted over the benchmark unit transmitting an ultrasonic signal to the seafloor transponders. The signal was reproduced by the transponders replying to the vessel. Positions of the transponders and the sound speed structure of the sea water were determined simultaneously from the round-trip travel time of the signal using a tomographic technique. We repeatedly carried out measurements over the two sites. For the Kumano Basin, the horizontal precision of the benchmark location was ƒãh=¡Ó5 cm and its vertical precision was ƒãv=¡Ó10 cm through the repetitive measurements. At the Kumano Basin, a 21.5-cm southward displacement of the benchmark unit was detected just below the site before and after a large earthquake (Mw 7.5). Our observation system therefore proved itself capable of detecting seafloor crustal deformation associated with crustal activities in offshore areas. The benchmark location was determined by inversion method using the precisely determined vessel position and the travel time of the ultrasonic signal. The lateral variation of the sound speed structure is very large reaching to several tens cm as a sound path so that we need to estimate the sound speed structure as well as benchmark locations. In the present analysis, we assume that the sound speed does not change horizontally but only in time. This is based on the nature that the lateral variation of the sound speed is limited only in the shallower part than 500m deep. We can neglect the effect of the variation on the sound path because the sound path generated by the vessel is very close to each other in the shallower part. Therefore, we can solve the inverse problem, whose equation system has 3N (N shots for each benchmark) equations and N (sound speed structure for each moment) plus 9 (positions of three benchmarks) unknown parameters, as an over determined system. However, the actual sound speed structure has lateral variations. Larger variance of the benchmark positions corresponding to larger variance of the sea surface temperature suggests that the spatial variation in the sound speed affects the solution. We must estimate the lateral variation of sound speed structure to improve the benchmark positioning. We conducted several numerical experiments to find the suitable design of the measurement to solve the lateral sound structure. We present the result of the numerical experiment and the benchmark positions solved with the new approach.
G21A-0685
Improvement on GPS/Acoustic seafloor geodetic observation: constraint on acoustic velocity estimation and evaluation
Observations of seafloor crustal deformation is very important to understand the dynamics of plate boundary that include strain accumulation processes, great interplate earthquakes mechanisms, and submarine volcanoes activities. For these purposes, we installed acoustic transponders as geodetic reference sites on the Suruga trough, central Japan, where a huge earthquake is expected to occur in the near future. Using kinematic GPS and acoustic ranging technique, we have established two geodetic reference sites on the trough at about 800 m depth, approximately less than 30 km from the on-land costal GPS station. In our campaign observation, we measured ranges to the reference sites from on-board acoustic transducer whose position was determined by kinematic GPS. Repeated our campaign observations can reveal directly the seafloor crustal deformation in focal area of subduction zone. Our analysis method can simultaneously estimate the temporal variation of acoustic velocity and the weighted center of three seafloor transponders based on horizontally stratified structure of acoustic velocity (we defined this method as simultaneous method hereafter). Applying the simultaneous method, our campaign observations within one year could measure the velocity of crustal deformation associated with subduction at the sites. However, the detected velocity of the deformation was limited in horizontal component. The repeatability of seafloor positioning (rms) was about 3 cm. In addition, according to the result of numerical experiment using this simultaneous method, the positioning might be affected by bias error using travel-time dataset made under the temporal variation of acoustic velocity with dual-period. The most probable cause of the bias error was misestimation of acoustic velocity was its uneven location in space. Thus to avoid misestimation, a constraint is added on the estimate of the acoustic velocity to the simultaneous method. With this constraint, the result in numerical experiment showed that the positioning by the improved method has high repeatability in three components. The constraint velocity was obtained by the continuous measurement of temperature and pressure in seawater. In parallel with the acoustic measurements, continuous measurements of acoustic velocity conducted by towing ten sensors of temperature and pressure attached on rope with interval of 50 m. For reduction of the measurement noise, fitting acoustic velocity to smooth surface with minimization of ABIC was carried out. This smoothed acoustic velocity was used for constraint in the improved simultaneous method. In this presentation, we will report the positioning repeatability of the weighted center by using actual observation data obtained from five campaign observations at the each site for the period from 2006 to 2008.
G21A-0686
A Bayesian Approach for Inter-plate Coupling Models in Subduction Zones
We aim to characterize the extent of apparent plate coupling on subduction zone megathrusts with the eventual goal of understanding variations in fault zone rheology. In this initial study, in order to demonstrate the basic approach, we adopt a simple kinematic back-slip model (Savage, 1983). This study differs from most (but not all - e.g., see Segall, 2002) analogous studies in that we use a Bayesian approach wherein we ask not for a single optimum model, but rather for a-posteriori estimates of the range of allowable model parameters. This approach also allows us to explicitly define physically plausible a-priori information on data uncertainties and model parameters, as opposed to assuming that both follow Gaussian statistics. The Bayesian approach inherently depends on an ability to routinely compute millions of forward models that are consistent with a-priori constraints and available geodetic measurements. Such computations are now viable with available computational resources. We apply this methodology in the Chilean-Peruvian subduction zone with the desire to understand the state of inter-seismic coupling in that margin.
G21A-0687
Meshfree Finite Element Technique for Geoscience Research: A New Tool for Modeling Earthquake-induced Crustal Deformation
Meshfree (or meshless) Finite Element (FE) may sound as an oxymoron, but it is an emerging modeling technique with many advantages. The method still uses elements, just as the standard (meshed) techniques. However, the elements are set by a grid that does not necessarily conform to the geometry of the modeled object. In the contrast to standard FEMs, in our meshfree method the object's geometry is represented by the distances to the boundaries. Distance to the boundary is an intrinsic property of any geometric object, and it provides a natural way to represent the geometric information and satisfy boundary conditions exactly. Because meshfree methods do not require spatial meshing that conforms to the geometric model, they provide the needed geometrical flexibility lacking in standard FE methods. However, the treatment of boundary conditions becomes more challenging. The meshfree FE technique is used in variety of engineering applications, such as heat transfer, structural analysis, shape-material optimization and stress analysis of acquired models. So far, this very promising technique had a very minor impact in geoscience research. Only three geoscience papers used the meshfree technique in their research. We have started utilizing this flexible numerical modeling tool to study earthquake-induced crustal deformation. So far, we used 3-D elastic models to simulate interseismic deformation due to a buried dislocation. Our preliminary results show an excellent fit with analytical dislocation models. We plan to continue developing the application of the meshfree technique to crustal deformation studies, in order to generate a modeling tool that can easily incorporate complicated geometries (faults, topography, and crustal units), geometrical changes over time, complicated rheologies, and heterogeneous crustal properties.