Regional and Global-Scale Plate Kinematics and Dynamics From Geodetic, Geological, and Geophysical Observations Posters
Presiding: G Sella, Northwestern University; L Lavier, Institute for Geophysics, University of Texas at Austin
G43B-01 1330h
On the S-approximations of the Earth's gravity field
The so called method of linear analytical representations elaborated by V.N.Strakhov seems to be an effective tool for finding the anomalous field elements in local, regional and global cases. Suppose the "ideal" Earth is the interior of the ellipsoid with the parameters a,b,c. Then the "real" Earth can be treated as a domain restricted by piecewise continuous surface S which slightly deviates from this ellipsoid and contains the ideal Earth. Approximate values of a function G(x) which is harmonic outside S are assumed to be given in a set of points (arbitrary) on the surface S. These values can be represented by the sum of simple and double layers distributed on the ellipsoid. The densities of these layers (singular sources) are to be determined. In the local case we can take as the support of simple and double layers an infinite plane, i.e. XOY. The determination of densities (singular sources) reduces to solving a linear algebraic equation system with the approximately given right-hand side. For solving such systems there was used the new method of finding stable approximate solutions of linear algebraic equation systems with the approximately given right-hand side. The computer technologies designed by the author of this contribution have been tested on some models and turn out to be effective.
G43B-02 1330h
Relating Tide Gauge Records of Sea level Change to Local Vertical Motions
We relate local rates of relative sea level change to ITRF2000 vertical velocities via a simple function involving a regional rate of sea level change G and the components Vx, Vy, and Vz of a constant 3-D velocity vector referred to the X-, Y-, and Z-axis, respectively, of ITRF2000. Using rates from 36 tide gauges spanning the U.S. coast (Atlantic, Gulf, and Pacific as well as the coast along southern Alaska, Hawaii, Kwajelein, Guam, and Bermuda) together with ITRF2000 vertical velocities for 38 continuously operating GPS reference stations (CORS), each of which is located within a few tens of kilometers of one of these 36 tide gauges, we estimate that G = 1.8 +/- 1.1 mm/yr, Vx = 2.3 +/- 0.9 mm/yr, Vy = -0.2 +/- 1.1 mm/yr, and Vz = -6.0 +/- 1.4 mm/yr, where uncertainties represent standard errors. The misfit between these local sea level rates and the corresponding ITRF2000 vertical velocities at their associated CORS has a WRMS value of 1.6 mm/yr. This misfit is due primarily to uncertainties in the CORS-derived vertical velocities because of the relatively short history of these sites (2 - 9 years). Our results imply that a few years of GPS data from a CORS located along the U.S. coast may be used to predict the rate of local sea level change with a standard error of 1.6 mm/yr. Conversely, a few decades of tide gauge data may be used to predict local ITRF2000 velocities with this same level of accuracy. Moreover, rates of relative sea level change can also be related to vertical velocities expressed in any other spatial reference frame whose velocities are related to ITRF2000 velocities by a transformation as is the case with the realization of the North American Datum of 1983 that is known as NAD 83 (CORS96).
G43B-03 1330h
Strain Accumulation Models in the Lesser Antilles Using GPS Geodesy
In the northeastern Caribbean, convergence along the Caribbean - North American plate boundary is oblique and relatively slow. Subduction is near normal in the southern Lesser Antilles but becomes more oblique in the north, particularly near Cuba. Estimates of locking on the plate interface are controversial and vary from zero to 50% in the north but are generally agreed to be zero in the south. Compounding the difficulties in determining strain accumulation from locking, movement of the Caribbean plate is not well constrained, and errors in the Caribbean reference frame are large compared with observed site velocities. To constrain locking on the plate interface, GPS measurements have been taken since 1994 at stations installed on islands of the Lesser Antilles. In July and August of 2004, observations were made on the islands of Anguilla, St. Maarten, St. Kitts and Nevis in the northern Lesser Antilles at nine sites that were established and initially occupied in 2000 - 2002. More than 2.5 days of continuous observation using dual-frequency code-phase receivers and choke ring antennae were obtained at each site. Because stations are small and far apart and the locations less than ideal for needed measurements, observations from the 2000 - 2004 campaigns were compared with predictions from simple elastic half-space models developed using DISL. Observed velocities vary significantly from the models in both magnitude and direction. Preliminary models predict a very small signal from strain accumulation related to subduction (on the order of less than 2 mm/a to the southwest). Observed velocities have larger magnitudes, however, which may indicate contamination from local tectonics, slope instabilities, and volcanic deformation. Velocity orientations show an overall consistency with locking on the plate interface. Removing the effects of strain accumulation from observed velocities should help constrain local tectonic effects and allow us to test models of upper plate deformation and Caribbean plate rigidity.
G43B-04 1330h
A semi-analytic evaluation of the effect of second-order ionosphere term on GPS positioning
We developed a method to evaluate the effect of the second-order ionosphere term on GPS positioning. The method is based on the semi-analytic positioning error simulation method developed by Geiger (1988), which assumes the continuous distribution of the GPS satellites and maps the ranging error to the positioning error using the normal equation. We expanded the method to incorporate the satellite positioning error due to the second-order ionospheric term, which is estimated in a similar manner as the site positioning error, assuming the continuous distribution of the ground tracking stations instead of the continuous satellite distribution in the case of the site positioning error estimation. The method is first applied to simulate the positioning errors on three IGS sites (BAHR, COCO, GALA) which were investigated in Kedar et al. (2002) by analyzing the observed GPS data using the GIPSY software with the correction for second-order ionospheric term. We considered three cases, namely, 1) without satellite positioning error, 2) with satellite positioning error, and 3) with satellite positioning error whose spatial average of each component is corrected for. The third case corresponds to the situation where there are other observations available such as SLR and the center of the mass of the GPS satellite network is corrected properly. For the first case, we found that our method reproduced the positioning errors observed at these stations as well as Kedar et al. (2002). For the second case, however, we found that the positioning error is almost canceled. For the third case, we found that the error is reproduced as well as in the first case, though the spatial distribution of the error is different. These results indicate that 1) the semi-analytic method developed in this paper is accurate enough to simulate the position error due to the second-order ionospheric term, and 2) the satellite positioning error due to the second-order ionospheric term may have significant effect on the site position error and should be evaluated properly. Next, we applied the method to investigate the seasonal coordinate timeseries variability of the GEONET sites in Japan. Heki et al. (2001, 2003) fully analyzed the GEONET seasonal coordinate variability and found that most of the variability is explained by the snow loading, except the large annual scale change (5.6 ppb) whose source is unknown. We re-analyzed the GEONET data using the GIPSY software and found that the large annual scale change is actually semi-annual rather than annual. We simulated the position error time series of the GEONET sites using our method, and determined the scale parameter. We found that our method successfully reproduced the characteristics of the observed scale change of the GEONET, indicating that the large seasonal scale change of the GEONET is artifact caused by neglecting the second-order ionospheric term.
G43B-05 1330h
Investigation of Earthquakes in Turkey Using GIS
There are three main active faults in Turkey: the North Anatolian Fault (NAF), the Northeast Anatolian Fault (NEAF) and the East Anatolian Fault (EAF). NAF is one of the most active seismic regions over the world, runs along the northern part of Turkey about 1500 km, from the Aegean to the Karliova triple junction in the eastern Turkey. It has been the source of numerous large earthquakes throughout the history. Geographic Information Systems (GIS) are increasingly used to visualize elements associated with seismicity. It enables an efficient interactive exploration and spatial analysis of attributed geographical data. The database of this study consists of the earthquakes that occurred in Turkey since 1970 and the active faults in Turkey, which may cause the huge earthquakes in the future. Seismicity data were obtained from National Earthquake Monitoring Center of Kandilli Observatory and Earthquake Research Institute of Bogazici University. These data were in a text file format with each epicenter identified by latitude and longitude coordinates. This file also includes date, time, depth, magnitude and location information of the earthquakes. These data were then projected onto the map of the faults using GIS software and knowing the projection of the fault data set. The databases and analysis results are visualized by using the GIS software.
G43B-06 1330h
Monitoring Crustal Deformation on the Silent Strand of the Western North Anatolian Fault Zone
The Western Part of the North Anatolian Fault Zone (NAFZ) has been selected as a working area by many scientists due to the higher seismic activity on it. After the two big catastrophic earthquakes in 1999: the August 17 earthquake in Izmit, and the November 12 earthquake in Duzce, Turkey, many investigations have been pursued in the area. NAFZ branches from 30.6 E westward into three active fault zones. The northern strand is called Izmit-Sapanca fault zone, the middle strand is called Iznik-Mekece fault zone, and the southern strand is called Mudurnu fault segment. The middle branch, which is known seismically low active than the other fault zones, extends westward via the southern coast of Lake Iznik to the Marmara Sea; it consists of simple strand almost over its length. According to historical documents, the 29 AD event is the only large earthquake known in the Iznik area. Presence of seismic gaps in Iznik-Mekece fault zone implies potential sites for future large earthquakes. In 1990, a geodetic network installed on the Iznik-Mekece fault zone by Geodesy Department of Kandilli Observatory and Earthquake Research Institute. First GPS measurements carried out in the year of 1994. The aim of these geodetic studies is to detect the crustal deformations using space techniques in this area. Nine-point GPS network has been measured annually. This study presents the analysis of these measurements and the evaluation of recent crustal movements in the area.
G43B-07 1330h
Modeling Uncertainty in Fault Locations and Earthquake Rupture Configurations in Northwest Anatolia
Active faulting and earthquakes almost always pose uncertainties in the subsurface configuration of faults and the extent of rupture. We here identify three geometric uncertainties in the location and extend of rupture due to incomplete historical data and/or submarine environments along the North Anatolian Fault and discuss an approach for evaluating endmember fault geometry and rupture scenarios. The first case is that of the uncertainty in the location of the western end of the 1967 Mudurnu earthquake due to limited data. We identified three potential rupture geometries based on field observations and tested them using a boundary element model and a posteriori constraint provided by the next rupture, the 1999 Izmit earthquake. We find that for three different potential 1967 Mudurnu Valley rupture scenarios there are positive Coulomb stress changes at the hypocenter of the 1999 Izmit earthquake. We also find, however, that only for the scenario where subsurface rupture in the 1967 event extends towards Lake Sapanca does an Izmit rupture segment receive greater Coulomb stress change than any other neighboring fault. The second case deals with the uncertainty in the western end of the 1999 Izmit earthquake. We have used ground surface displacements from InSAR interferometry and GPS data to constrain the distribution and western termination of Izmit slip and evaluate the most likely stress change scenarios on faults in the eastern Marmara Sea. The fault slip inversions indicate that Izmit earthquake slip extended at least twelve kilometers west of the Hersek Delta into the Marmara Sea, but that the orientation of the rupture past Hersek Delta is not well constrained. Testing several possible western rupture configurations, we find that either the Princes' Islands or Cinarcik fault may be most susceptible to future failure. The third case is the uncertainty associated with the fault configuration under the Marmara Sea. Using mechanical modeling, we test three configurations of the fault system within the Marmara Sea fault proposed by others and show that an interpretation with a series of pull-apart basins along a master strike-slip fault best produces the observed deformation pattern within the Marmara Sea. In the model, crustal faults within the Marmara Sea slip in accordance with GPS-constrained slip rates along deep plate boundary dislocations. The locations and relative subsidence of the basins along the northern Marmara trough are well matched by our model results. This method shows the ability for mechanical modeling to evaluate fault configuration models that are otherwise equally justified by interpretation of seismic reflection data.
G43B-08 1330h
Spatial Variation on a Direction of Maximum Horizontal Compression in and around the Atotsugawa Fault System, Japan, Inferred from Shear Wave Splitting
We investigated a shear wave splitting of micro earthquakes occurred in and around the Atotsugawa fault system, Central Japan, to infer a spatial variation of the direction of maximum horizontal compression (Shmax). Leading shear wave polarization directions (LSPD) indicate that the maximum compressional axis trends WNW-ESE, which is nearly consistent with the direction of the regional Shmax inferred from geodetic observations and focal plane solutions of micro earthquakes. In addition, we find a spatial varation on a direction of Shmax along and across the fault. The angle between the direction of Shmax and the fault strike ( φ ) ranges 20 - 40 degrees for the most of stations at the eastern part of the fault, and 40 - 45 degrees at the central part of the fault. Moreover φ changes across the central part of the fault. Whereas φ ranges 55 - 70 degrees to the fault strike at the stations away from the fault, φ takes 40 - 45 degrees at the vicinity of the fault. In order to clarify the main cause of the spatial variation of the direction of Shmax across and along the fault, we performed a finite element modeling of the stress field in and around the Atotsugawa fault. Since the displacement field has been investigated by GPS observation at the central part of the Atosugawa fault, we incorporated it into a boundary condition of model space. Although the maximum stress direction at the central part of the Atotsugawa fault can be reproduced with the stable sliding at the deep extension of the fault, the direction of Shmax at the eastern part of the Atotsugawa fault cannot be explained by any possible models. Improving a boundary condition of the eastern part of the Atotsugawa fault or incorporating internal forces at the eastern part of the Atotsugawa fault is necessary to perform more sophisticated modeling on the variation of the direction of Shmax for the Atotsugawa fault area.
http://staff.aist.go.jp/takashi.mizuno/
G43B-09 1330h
3-D Gravity modeling of the crust and uppermost mantle beneath southern California: implications for the deep structure of the San Andreas continental transform zone.
We have used a regional compilation of both onshore and offshore gravity data to investigate variations in crustal and upper mantle structure beneath southern California in order to better understand the tectonic development of the San Andreas continental transform zone. The seismic velocity model developed by the Southern California Earthquake Center (SCEC - version 3) was used to construct an initial 3-D, multilayer density model of the region for comparison with observed gravity data. The model uses vertical triangular prisms with non-parallel upper and lower surfaces to simulate lateral density variations. The model was divided into an upper portion that incorporates near surface lithologic information and topography and a lower portion that includes density variations in the deeper crust and uppermost mantle derived from the seismic velocity data. The broad scale agreement between the anomaly field generated by our model and the major features in the observed gravity data indicates that the overall structure is well defined. In several areas, however, differences between the modeled and observed data are significant. To address these differences we have modified our initial model using densities obtained from a range of published velocity-density relationships and rock sample measurements. Differences between the modeled and observed gravity are most apparent in the high frequency portion of the gravity spectra and can best be explained by the poorly constrained near surface density distribution. Other sources of the differences include those due to edge effects and those imposed by limits in the SCEC model data. Details regarding along-strike variations in the deep structure beneath the transform zone will be presented.
G43B-10 1330h
Changes in rock-mechanical properties, local stresses, and displacements during the evolution of strike-slip faults
During their early evolution, strike-slip faults rapidly develop zones of rocks with widely different mechanical properties. These properties, in turn, largely determine the local stresses inside and around the fault zone and, thereby, the subsequent slips during fault rupture. It is common to distinguish between two main mechanical units: a fault core and a fault damage zone. The damage zone, which is normally much thicker than the core, contains some lenses of breccia, but is characterized by fractures of various types and frequencies. By contrast, the core is primarily composed of breccia and gouge. In an active seismogenic fault, the core is normally soft (with a low Young's modulus). The Young's modulus (stiffness) of the damage zone, however, depends on the fracture frequencies and trends in relation to the main direction of loading. As a rule, the higher the fracture frequency the lower is the effective stiffness in a direction perpendicular to the main fracture direction. Since fracture frequency tends to increase on approaching the core-damage zone boundary, it follows that the stiffness of a damage zone normally decreases toward that boundary. Field studies indicate that a damage zone can often be divided into several units based on fracture frequencies, each with a different stiffness. Observations also show clearly that the cores and particularly the fault-damage zones tend to grow thicker with increasing total fault displacement. We use these field results as a basis of numerical models of strike-slip faults using the boundary-element method. In these models we vary the stiffnesses of the cores and divide the damage zones into several units, each with a different stiffness. For active strike-slip faults, the stiffest units of the damage zone are at its contact with the host rock and gradually decrease in stiffness toward the contact with the core. In some of the models, the lateral tips of the strike-slip faults end inside, or near to, soft inclusions. Such inclusions include various soft rock bodies such as (for near-surface faulting) young volcanoes or (for transform faults) young volcanic zones or ridge segments. The three main results may be summarised as follows. First, when the fault tip is nearby, or inside, a soft inclusion, the slip, for given loading conditions and fault geometry, is much larger than when the tip is far away from the inclusion, or the inclusion is absent. Second, in models where soft inclusions are absent but the damage-zone thickness around the fault increases with time, the maximum displacement (u) on a strike-slip fault of a given length and with constant loading conditions gradually increases. It follows that for a strike-slip fault of a rupture (trace) length (L), the ratio u/L decreases with time. In other words, the slip in individual earthquakes in relation to the rupture length increases with time. Third, even when the fault slip in individual earthquakes increases, the displacement (slip) profile remains similar as regards shape. Thus, for this type of loading and with gradually increasing damage-zone thickness, the displacement profile remains a smooth, convex curve with a maximum slip at the fault center.