G13A-0784 1340h
Temporal Clustering of Earthquakes due to Stress Transfer in Viscoelastic Layers
Postseismic processes can have a significant effect on the reloading rate of the seismogenic portion of active faults. As a result, the rheology of the non-seismogenic lower crust and mantle lithosphere may play a role in determining earthquake recurrence times. Previously, \emph{Kenner and Simons} (2004) used a one dimensional spring-dashpot-slider analogue model to investigate fault reloading due to viscous relaxation. They found that the system's behavior was controlled by a non dimensional number called the Wallace number, \emph{W}. \emph{W} is equal to the average earthquake stress drop divided by the product of the applied geologic strain rate and the effective viscosity of the system. In the presence of a small amount of normally distributed environmental noise, temporal clustering of earthquakes was observed when \emph{W} was high. To study a more physical system we expand the model beyond a one-dimensional analogue. When more than one viscoelastic layer is present, the effective viscosity of the entire system is a function of both the thicknesses and the viscosities of the viscoelastic layers. We use a two-dimensional, out-of-plane finite element model of an infinite strike slip fault. The model consists of an elastic layer over one or more Maxwell viscoelastic layers. A far-field, constant velocity boundary condition drives the interseismic strain accumulation. The seismogenic portion of the fault is allowed to slip freely and instantaneously when its yield stress is exceeded. When this occurs, stress accumulated on the seismogenic fault is shed to the viscoelastic layers below; in this way, the relaxation time of the viscoelastic layers effects the stress reloading rate of the seismogenic fault. This stress is then recycled back to the seismogenic fault as the lower layers relax. The two-dimensional finite element model exhibits the same clustering behavior and dependence on \emph{W} as the spring-dashpot-slider analogue model. We determine the dependence of the effective viscosity on the thicknesses of the viscoelastic layers. \section*{References} Kenner, S.J. and Simons, M., 2004. Temporal cluster of major earthquakes along individual faults due to postseismic reloading, \emph{Geophysical Journal International}, submitted.
G13A-0785 1340h
Local Postseismic Relaxation Observed After the 1992 Landers (M=7.3), 1999 Hector Mine (M=7.1), 2002 Denali (M=7.9), and 2003 San Simeon (M=6.5) Earthquakes
The U. S. Geological Survey has observed the local postseismic deformation following the 1992 Landers (M=7.3), 1999 Hector Mine (M=7.1), 2002 Denali (M=7.9), and 2003 San Simeon (M=6.5) earthquakes. The observations consist of repeated campaign-style GPS surveys of geodetic arrays (aperture $\sim$ 50 km) in the epicentral area of each earthquake. The data span the intervals from 0.037 to 5.6, 0.0025 to 4.5, 0.022 to 1.6, and 0.005 to 0.55 yr postearthquake for the Landers, Hector Mine, Denali, and San Simeon earthquakes, respectively. We have reduced the observations to positions of the monuments measured relative to another monument within the array. The temporal dependence of the relative displacements for each monument can be approximated by {\it a}+{\it bt}+{\it c(1-exp[-t/d])} where {\it a}, {\it b}, {\it c}, and {\it d} are constants particular to that monument and {\it t} is the time after the earthquake. The relaxation times {\it d} were found to be 0.367±0.062, 0.274±0.024, 0.145±0.017, and 0.032±0.002 yr for the Landers, Hector Mine, Denali, and San Simeon earthquakes, respectively. The observed increase in {\it d} with the duration of the time series fit suggests that the relaxation process involves more than a single relaxation time. An alternative function {\it a'}+{\it b't}+{\it c'log(1+t/d')} where {\it a'}, {\it b'}, {\it c'}, and {\it d'} are constants particular to each monument furnishes a better fit to the data. This logarithmic form of the relaxation (Lomnitz creep function), identical to the calculated response of a simple spring-slider system subject to rate-state friction [{\it Marone et al.}, 1991], contains a continuous spectrum of relaxation times. In fitting data the time constant {\it d'} is determined by observations within the first few days postseismic and consequently is poorly defined. Adequate fits to the data are found by simply setting {\it d'}=0.001 yr and determining {\it a'}, {\it b'}, and {\it c'} by linear least squares. That the temporal dependence is so readily fit by both exponential and logarithmic functions suggests that the temporal dependence by itself may not be very diagnostic of the causative mechanism.
G13A-0786 1340h
FEM modeling of postseismic deformation of poroelastic material
Following a large earthquake, postseismic deformation in the focal region has been observed by GPS, leveling measurements and the other geodetic measurements. To explain the postseismic deformation, researchers have proposed and well investigated two physical mechanisms of afterslip and viscoelastic relaxation. In some cases, however, there have been observed postseismic deformation which can not be explained by these mechanisms. Therefore, another mechanism has been proposed, where the crust is treated as "poroelastic material". This concept is called "poroelasticity". In this concept, postseismic deformation is caused by pore fluid flow due to the coseismic stress redistribution. We explored, therefore, the postseismic deformation due to pore fluid flow in a poroelastic material using finite element method (FEM), which can easily handle lateral variations of hydraulic diffusivity and elastic or plastic property. We used the FEM program 'CAMBIOT3D' originally developed by Geotech. Lab. Gunma University, Japan (2003). Because this program was developed for soil mechanics, we must have modified so as to calculate deformation due to earthquake faulting. We implemented the 'split node technique' (Melosh and Refsky, 1981) to calculate the coseismic deformation. In addition to this, we modified the program to calculate the deformation taking into account the Skempton's B. This coefficient B determines what fraction of the coseismic stress due to an earthquake is allotted to pore pressure. Without Skempton's B, coseismic pore pressure becomes too large and hence postseismic deformation is calculated too large. We evaluated the postseismic deformation in a poroelastic material to show that the poroelastic deformation is quite different from that of afterslip and viscoelastic relaxation models. In this presentation, we show the postseismic deformation due to pore fluids flow in a poroelastic material and the effect of Skempton's B. Especially, we discuss what different pattern of postseismic deformation is produced depending on the lateral variation of hydraulic diffusivity structures in and around the fault zone, which structures have been differently inferred from fault zone core sampling researches and so on.
G13A-0787 1340h
Crustal Deformation in the Kanto District, Central Japan, Following the 2000 Seismo-volcanic Activity of the Izu Islands
Starting from June 26, 2000, an unprecedented seismic activity occurred around the Miyake-jima, Kohzu-shima, and Nii-jima Islands, in the northern Izu islands. This seismic swarm activity was initiated by the volcanic magma intrusion beneath the Miyake-jima volcano. An intrusion of massive (about 1km$^{3}$) magma caused the seismic swarm activity and magnificent crustal deformation in the surrounding area within about 200km from the source region. After the seismic swarm activity calmed down, we detect a change in crustal displacement rates in the southern Kanto region from daily coordinate solutions of the continuous GPS network. Interestingly, the change appears mostly in the E-W components. Comparison of GPS velocity data for two time periods (1996-200 and 2001-2002) indicate that the westward displacement rate decreased by about 25% (from 23 mm/yr to 17 mm/yr) at Tateyama, the southern tip of the Boso Peninsula. On the other hand, we do not see significant changes in the N-S and vertical components. Continuous monitoring of crustal displacements with GPS has revealed that the post-swarm deformation is now coming back to the pre-swarm steady state. That is, the time series of E-W component show transient curves, converging into the original steady state. The transient curve can be equally well reproduced by an exponential decay or a logarithmic function. The relaxation time for the exponential curve is estimated as about 3 years. One possible explanation for this transient deformation is viscoelastic relaxation. Since the Izu Islands are situated on the oceanic Philippine Sea plate, the upper mantle with a low viscosity would response to the huge stress change cause by the magma intrusion. The other possibility is a change of frictional property on the plate interface between the Philippine Sea and the Pacific plate. Under the southern Kanto area, the subducted Philippine Sea slab leans on the subdcted Pacific slab. Interaction between these two oceanic plates is still not understood well. But the massive dyke intrusion strongly pushed the subducted Philippine Sea slab, changing the frictional status at the bottom of the Philippine Sea plate. Since the motion of the Pacific plate subduction is nearly westward, this idea can be a solution for the observation that only the E-W components are affected.
G13A-0788 1340h
Estuarine Evidence Of Postseismic Transients In 17th-century Hokkaido And 20th-century Chile
We studied postseismic transient of two giant interplate earthquakes; one is a 17th-century earthquake along the southern Kuril trench, eastern Hokkaido, and the other is the 1960 Chilean earthquake. The 17th-century earthquake along the southern Kuril trench induced postseismic uplift that probably lasted for decades, as judged from stratigraphy and paleoecology at two estuaries along the Pacific coast of eastern Hokkaido. The estuaries, at Mochirippu and Kiritappu, were invaded by an unusually large tsunami before their tidal flats emerged and became lowland forests. The tsunami is marked by a sand sheet mainly 2-5 cm thick. The sand is both underlain and overlain by tidal-flat mud. However, the mud above the sand grades upward into peat that contains volcanic ash layers from late in the 17th century. The upward sequence of mud, sand, mud, and peat implies that an earthquake (marked by the tsunami deposits) predates gradual coastal uplift (marked by the transition from mud to peat). We quantified the land-level history by means of fossil diatom assemblages in the mud and peat. The reconstructed land-levels show several decimeters of preseismic subsidence, no coseismic change, and at least 1 m of postseismic uplift. The volcanic ash layers, erupted from southwest Hokkaido and dated with the local historical records, show that the uplift started before 1667 and ended by 1694. Postseismic uplift of the 1960 Chilean earthquake provided stratigraphic and paleoecological evidence. The delta of the Rio Coihuin is in middle along the 1960 rupture zone and 10 km east of Puerto Montt, which was at the inland limit of the coseismic downwarp in 1960. The delta contains three terraces, each underlain by a different stratigraphic sequence. The_@highest terrace, now above all tides, is underlain by lahar-runout sand_@and silt derived from Mt. Calbuco, 20 km to the northeast. Inset into_@this terrace are at least lower ones that form tidal marshes. The marshes_@are underlain by intertidal deposits dominated by mud and peaty mud. A family that has farmed a remnant of the lahar-runout terrace provided_@detailed accounts of changes in land level since the 1960 earthquake._@The highest terrace had been inundated by tidal water during the monthly high tide and dominated by a tidal marsh plant, Salicornia sp., before the 1960 event. The middle terrace had been also dominated by other tidal marsh plant, Juncus balticus and Scripus americanus that the family used for their craft. However, after the event, the terraces emerged and the tidal vegetation changed. The highest terrace became freshwater upland forest, and the middle terrace increased Agrostis alba and Salicornia sp. Tidal flat changed to the present lowest terrace dominated by Puccinellia sp. Guided by this testimony, we checked deposits beneath the middle terrace beside the family's house. These deposits consist of peat, mud, sand, and volcanic ash layers. _@The uppermost peat-over-mud contact probably represents emergence that the family has been watching after the 1960 earthquake, though we cannot yet date the contact. Above the peat-over-mud contact, the peat contains rhizomes of Spartina densiflora and plant macrofossils of salt-tolerant vascular plants. Judging from the present vegetation around the delta, this change in vascular plants shows at least 1m of emergence.
G13A-0789 1340h
Post-seismic deformation across the central Nevada Seismic Belt, USA, and rheology of the lithosphere.
We investigate the post-seismic surface deformation in the Central Nevada Seismic Belt (CNSB), located at the south western end of the Basin and Range, USA. The CNSB was the location of major normal and strike-slip earthquakes in 1915 (Pleasant Valley M7.6), 1932 (Cedar Mountain M7.2) and in 1954 (Fairview Peak M7.2 & Dixie Valley M6.8). These earthquakes represents an interesting set of events to study post seismic deformation because of their magnitude, number and temporal repartition. Moreover, GPS measurement over the CNSB shows abnormal pattern, first the rate of present deformation is higher by ~2mm/yr in comparison with the quaternary strain rate obtained by paleoseismology, then, the GPS shows compressionnal pattern in an extensional context, which suggest transient post-seismic deformation following the earthquakes. We measure surface deformation in the CNSB using Interferometric Synthetic Aperture Radar (InSAR). We obtained 9 interferograms from 18 SAR images each covering four adjacent frames (~400 km in along-track direction), spanning 4-7 years using ERS SAR imagery acquired between 1992 and 2001. A precise displacement map obtained by stacking (averaging) the interferograms shows shortening of the distance between the ground and the satellite at a rate of ~2-3mm/yr (range change). The maximum range changes occur in the epicentral area of the 1915 Pleasant Valley, 1954 Fairview Peak and 1954 Dixie Valley earthquakes. We modeled the post-seismic deformation expected after the earthquakes using VISCO1D code (Pollitz F.) which allow to compute the surface deformation caused by viscoelastic relaxation within the lithosphere. We include the stress imposed by the four earthquakes, tested different lithosphere models and invert for the viscosity of both lower crust and upper mantle, elastic thickness and earthquakes magnitudes. We found that post-seismic deformation explain the observe deformation for an elastic thickness equal to the crustal thickness and a viscoelastic upper mantle with low viscosity value of ~2e18 Pa.s. Both abnormal high extension rate and compressionnal pattern measured by GPS are explained by our model for the post-seismic deformation.
G13A-0790 1340h
Transient uplift since the 1960 Chilean earthquake
Postseismic uplift midway along the 1960 Chile rupture appears to be slowing down. We studied the uplift near Puerto Montt, 70 km east of the Pacific coast. Puerto Montt is located on the boundary between the 1960 earthquake's coseismic and a coseismic and/or postseismic upwarp farther inland. This deformation was mapped by George Plafker in 1968, and the postseismic uplift was checked by him and others in 1989. The postseismic uplift has been attributed to afterslip on the plate interface, stress relaxation in the forearc mantle, or both. We obtained further information about shoreline change before and after the earthquake from the Pacific coast to 120 km eastward inland, about midway along the length of the 1960 rupture. Following in Plafker's footsteps, we used several criteria for distinguishing between coseismic and postseismic movement. We asked eyewitnesses whether sea level changed at the time of the earthquake, and whether it has changed in the decades since. We also looked for emerged landforms and for environmental changes in vegetation. Where reliable testimony or natural features provided exact reference points, we measured differences in height by means of total station. Twenty-seven places yielded numerical estimates of land-level changes since 1960. The largest total amount of displacement during past 44 years was estimated to be 2.1 m in Chamiza, 10 km east from Puerto Montt. According to the interviews, most of the uplift occurred during or in the first month the earthquake. The total amount of uplift tends to decrease eastward from Chamiza. The easternmost point (60 km east from Puerto Montt) did not rise or fall during the 1960 earthquake, but gradually rose nearly 1 m in the first 40 years thereafter. The postseismic uplift has probably been canceled by subsidence later in earthquake deformation cycles. The Puerto Montt tide gauge record shows little or no uplift in the 1990s. Similarly, some eyewitnesses in and near Chamiza say that in the past decade, the sea has begun to rise. Although historical (1575) and prehistoric earthquakes perhaps comparable to the 1960 event produced stratigraphic evidence for tsunami and coseismic subsidence 55 km to the west, near Maullin, we have not found marine terraces that would indicate net late Holocene uplift in the Puerto Montt area. George Plafker, through Brian Atwater, kindly gave us copies of his field notes from 1968 and 1989. Marco Cisternas organized the project. Youlton's work supported by Fondecyt 1020224.
G13A-0791 1340h
InSAR Observations of Time-Dependent Postseismic Deformation in the Mojave Desert: Resolving Tectonic From Non-tectonic Processes
We examine InSAR observation of the postseismic deformation pattern following the 1992 Landers and 1999 Hector Mine earthquakes to help differentiate the various tectonic and non-tectonic processes. Viscoelastic relaxation of the upper mantle or lower crust, poroelastic response of the crust, and afterslip have all been used to explain various features of the geodetic data following the Landers and Hector Mine events. Our objective is to use the InSAR data to decipher the relative magnitude of these tectonic processes, all of which have distinct spatial and temporal deformation patterns. In this work we emphasize the use of the temporal pattern of deformation to further constrain the postseismic sources of deformation. We processed 250 interferograms along track 127 in the Mojave Desert spanning the time period from August 1992 through the end of 2000. We perform a least squares inversion of the InSAR data to solve for a time series that describes the temporal evolution of the deformation. Unfortunately, several non-tectonic processes complicate the InSAR data. Lateral variations in atmospheric water vapor, which acts to delay the propagation of the radar signal, produce regional artifacts in the line-of-sight displacement. Atmospheric artifacts are reduced by imposing temporal smoothing in the inversion. Additionally, confined aquifers located in the Mojave Desert produce a land subsidence pattern as the water table fluctuates both on a long term and seasonal timescale. Well level data in the basins are analyzed to address the component of land subsidence in the range-change time series. The land subsidence signal can be removed from the time series by assuming that the resulting deformation is elastic. The well level data are linearly scaled and subtracted from the range-change time series leaving the non-linear tectonic signal. The remaining range-change signal is then quantitatively compared to the predicted postseismic deformation for models of the viscous, poroelastic, and afterslip response.
G13A-0792 1340h
Viscoelastic Deformation Model of the Western United States Instantaneous Velocity Field
Constraints on long-term deformation characteristics of the continents are provided by geologic slip rates on major faults, paleomagnetic rotations, and estimates of instantaneous velocity as measured by Global Positioning System (GPS) or other geodetic measurements over a short time span. Geologic slip rates and paleomagnetic rotations represent the long-term velocity field (i.e., that averaged over a timescale much longer than the earthquake cycle on an individual fault) in the vicinity of the corresponding faults or blocks. On the other hand, instantaneous velocity measurements correspond to a timescale that is much shorter than a typical earthquake cycle on a major fault. Such measurements are gaining importance with ever-increasing coverage of continental regions with geodetic networks. In order to satisfactorily explain the instantaneous strain rate field, we require estimates of slip of significant historic earthquakes, the mechanics of tectonic loading, and the underlying rheology. These are also critical ingredients to understand the long-term deformation pattern. We present and extend an existing relationship between the long-term dislocation rates and instantaneous velocities, allowing separate treatments of faults with known slip history, creeping faults, and dislocation sources distributed between the faults. For faults with known slip history, the relationship depends on a viscoelastic Earth model and thus accounts explicity for viscoelastic cycle effects. For distributed dislocation sources, we calculate the average interseismic velocity produced by viscoelastic relaxation over an entire cycle. We apply this relationship to the GPS velocity field in the western United States in order to test the importance of the relaxation from historic events and characterize the pattern of distributed deformation and the tectonic forces imposed by the bounding Pacific and Juan de Fuca plates. By accounting for viscoelastic cycle effects integrated over the totality of dislocation sources in the western United States, the relative contributions of discrete (fault-like) and distributed dislocations may be quantified. Our modeling of the western US strain rate field shows that relaxation following major earthquakes (M $>$ 7.5) strongly shapes the present strain rate field over most of the plate boundary zone. Relaxation following minor earthquakes is well detected within smaller regions around secondary active seismic zones (e.g., Central Nevada Seismic Zone, East California Shear Zone).
G13A-0793 1340h
Post-Seismic Crustal Deformation Following The 1999 Izmit Earthquake, Western Part Of North Anatolian Fault Zone, Turkey
The North Anatolian Fault is an about 1500 km long, extending from the Karliova to the North Aegean. Turkey is a natural laboratory with high tectonic activity caused by the relative motion of the Eurasian, Arabian and Anatolian plates. Western part of Turkey and its vicinity is a seismically active area. Since 1972 crustal deformation has been observed by various kinds of geodetic measurements in the area. Three GPS networks were installed in this region by Geodesy Department of Kandilli Observatory and Earthquake Research Institute( KOERI ) of Bogazici University: (1) Iznik Network, installed on the Iznik-Mekece fault zone, seismically low active part, (2) Sapanca Network, installed on the Izmit-Sapanca fault zone, seismically active part, (3) Akyazi Network, installed on their intersection area, the Mudurnu fault zone. First period observations were performed by using terrestrial methods in 1990 and these observations were repeated annually until 1993. Since 1994, GPS measurements have been carried out at the temporary and permanent points in the area and the crustal movements are being monitored. Horizontal deformations, which have not been detected by terrestrial methods, were determined from the results of GPS measurements. A M=7.4 earthquake hit Izmit, northern Turkey, on August 17, 1999. After this earthquake many investigations have been started in the region. An international project has been performed with the collaboration of Massachussets Institute of Technology, Turkish General Command of Mapping, Istanbul Technical University, TUBITAK-Marmara Research Center and Geodesy Department of KOERI. Postseismic movements have been observed by the region-wide network. A GPS network including 49 well spread points in Marmara region was observed twice a year between 1999 and 2003 years. During these surveys, another network with 6 points has been formed by using 2 points from each 3 microgeodetic networks on NAFZ with appropriate coverage and geometry. These points have been connected by GPS observations to monitor the deformations. This expanded microgeodetic network has been occupied with Istanbul-Kandilli continuous GPS station (KANT). The objective of this paper is to present the post-seismic crustal deformation obtained from the GPS observations at the Western Part of the North Anatolian Fault (NAF) in Turkey.
G13A-0794 1340h
Models of Afterslip and Viscoelastic Response following the Landers and Hector Mine Ruptures
We evaluate the postseismic deformation following the 1992 {\it M}$_{w}$ 7.3 Landers and 1999 {\it M}$_{w}$ 7.1 Hector Mine earthquakes using a fault slip distribution that includes both strike-slip and dip-slip. Observations of coseismic slip show significant vertical deformation that has not yet been included in models of postseismic activity. The rheological models include both deep afterslip and the viscoelastic response of a half space beneath an elastic layer. We simulate the complex geometries of the ruptures using hundreds of vertical fault patches as constrained by the surface rupture and aftershock studies. The viscoelastic response is modeled using a 3-D coseismic slip model obtained from inversions of InSAR and GPS observations [{\it Simmons et al.}, 2002; {\it Fialko}, 2004]. The post-seismic afterslip models which include strike-slip, dip-slip, and shallow fault-normal displacement are developed using the coseismic rupture geometry. The 3-D deformation is computed at a 500-m grid spacing and compared with available postseismic interferograms in both the near field and far field allowing us to consider both long wavelength and short wavelength behavior. The objective of the analysis is to evaluate the relative importance of afterslip and viscoelastic response and to determine whether a model with a thick elastic plate ($\sim$50 km) and moderate mantle viscosity (10$^{19}$ Pa s) is compatible with geodetic observations.
http://topex.ucsd.edu
G13A-0795 1340h
Primary Results Related To Crustal Deformation Study Of The Eastern Part Of NAFZ, Turkey (By GPS Measurements)
Eastern sector of North Anatolian Fault Zone (NAFZ) is an ideal region to investigate crustal deformation and block kinematics by Global Positioning System (GPS) measurements. Due to the northward movement of the Arabian Plate, the Erzincan-Karliova region is squeezed, crushed, and expelled westward along the NAF and East Anatolian Fault Zones. The active fault pattern indicates that maximum crustal shortening and crustal deformation in Turkey takes place in this region. This region is the most tectonically active region in Turkey and it is capable of generating major earthquakes in every 3-4 years. The area needs to be monitored using seismic, geodetic, and geochemical techniques. This project are being performed with the collaboration of Kandilli Observatory and Earthquake Research Institute (KOERI) of Bogazici University and the Marmara Research Center of the Scientific and Technical Research Council of Turkey (TUBITAK-MAM). The study includes the investigation of seismicity and earthquake potential. First period and also second period of GPS measurements were performed at sixteen GPS stations in the area, despite the hard conditions. This paper presents the study and the primary results obtained from these GPS observations.
G13A-0796 1340h
Clues of Post-seismic Relaxation for the 1915 Fucino Earthquake (Central Italy) From Modeling of Leveling Data
The 1915 Fucino earthquake (Ms=6.9) was one of the largest and most destructive events in Italy during last century. The epicentral area is centered in the Abruzzi region (Central Italy), where a long historical record of large earthquakes is available. Seismotectonic studies on this region, based on instrumental seismicity (focal mechanism solutions of major events and stress analysis of background seismicity), borehole break-out studies and several geological and paleoseismological investigations, suggest NE-SW oriented active extension. The 1915 earthquake fault produced detectable surface ruptures for about 20 km along NW-SE striking SW-dipping structures. Coseismic geodetic data recorded in the epicentral area have been inverted in the past (Amoruso et al. 1998 and references therein), indicating a source fault dipping at moderate angle toward SW and a normal focal mechanism, with a non-negligible left-lateral component. Three high precision leveling lines located in a wide sector north and east of the Fucino plain were measured in 1950 and 1997-2000 by the IGM (Istituto Geografico Militare). Two consecutive lines run in a NW-SE direction along the chain, and form a ``T-shape'' net together with a third line SW-NE striking, towards the Adriatic sea. The total length is about 360 km with a mean benchmark density higher than 0.5 bm/km. The relative elevation changes recorded during this time interval show maximum values between 7 and 12 cm with a signal wavelength of 40-70 km. The observed elevation changes stand significantly above the calculated total error of 1.13 $\sqrt{L}$ mm $\sqrt{\rm km}$. A sharp gradient has been noticed east of the earthquake epicenter, where we observe peculiar elevation changes along a 40 km long section of the leveling line. The observed elevation changes in the Fucino earthquake area seem to comprise both regional tectonic deformation and post-seismic relaxation. The former and the latter effects are expected to dominate along sections of the leveling lines which are respectively about perpendicular and parallel to the Apennines. Since we compare measurements performed in 1950 and 1997-2000, relaxation effects refer to a late stage of the process. We have used Pollitz (1997) code for computing gravitational-viscoelastic postseismic relaxation on a layered spherical Earth. Different Earth models, characterized by different thicknesses and viscosities of crustal layers and of the upper mantle, have been considered. Even if S/N ratio of expected post-seismic effects is not high, comparison between predictions and observations allows to constrain regional crustal structure. Best-fit seismic moment is in good agreement with Amoruso et al. (1998) and residuals are fully consistent with expected regional tectonic deformation in Central Apennines.
G13A-0797 1340h
Postseismic and Interseismic Deformation of Large Normal-Faulting Earthquakes in the Basin-Range
We studied the change of surface deformation after the 1959 Ms=7.5 Hebgen Lake, Montana, earthquake, measured by trilateration and GPS from 1973 to 2000 and the only postseismic observation of large normal-faulting earthquake in the Basin-Range. Time-dependent changes of baseline length across the fault were used to constrain the rheological structure of the crust and upper mantle beneath the Hebgen Lake fault zone. A Monte Carlo approach revealed less viscous, or weaker, lower crust and upper mantle southeast of the fault zone compared with those implied by baseline changes northwest of the fault. The local tectonism can explain this lateral variation, of which high heat flow and a thin brittle layer has been observed at the nearby Yellowstone caldera, locating 30 km southeast of the Hebgen Lake fault. We also applied the postseismic deformations measured by leveling in the Yellowstone-Hebgen Lake region and five GPS sites near epicentral area of the Ms=7.3, Borah Peak, Idaho, earthquake to examine the Hebgen-Lake rheological model. This model, moreover, is similar to the lithospheric structures of the eastern-Basin-Range implied by the long-term deformation of the lacustrine shoreline caused by the Lake Bonneville rebound, and thus was used to estimate combined postseismic responses caused by the most recent paleoearthquakes (M$>$7.0) on the Wasatch and the East Great Salt Lake faults and large historic earthquakes (M$>$5.5) in the Wasatch Front area, Utah. Contemporary surface deformation of the Wasatch Front area measured by campaign and continuous GPS was used to model the interseismic fault-loading rate of the Wasatch fault, where paleoseismic data from trenching revealed long-term fault-slip rate of the fault. Combining these results provide important new insights on temporal variations of the seismic cycle and constraints on related earthquake hazards.
G13A-0798 1340h
Realistic Post-seismic deformation in a spherically symmetric Maxwell Earth
The realistic post-seismic deformation in a spherically symmetric earth is obtained for the first time. The calculation is done by the new algorithm which makes it possible to consider the self-gravitation, the compressibility, and a continuously varying structure of the Earth without any approximations. This makes a remarkable contrast to the previous methods which have neglected such effects for the intrinsic numerical difficulties. The essential point of the new algorithm is to evaluate numerically the Laplace integration without taking the sum of the {\it innumerable} poles. Using this method, a complete set of the Green's function is shown; time variations of displacement, gravity, geoid height at the surface for a strike-slip, dip-slip, horizontal and vertical tensile point dislocation. As an earth model, we employ the 1066A model and the standard viscosity profiles. The result shows a diverse spatial pattern due to a viscous structure or a source depth. The considerations on the result suggests that what determines those patterns is mainly the thickness of the lithosphere and a relative location of the source. Of particularly interested is that the near filed deformation is mostly prevented by the lithosphere. This makes the {\it post}-seismic deformation be distinctive in an epicentral distance of a few hundreds km. This indicates that a post-seismic gravity change is possibly detected by the satellite mission because the wavelength exceeds 100 km, which strongly encourages us to observe it to infer the viscosity under the lithosphere.
G13A-0799 1340h
Temporal and Spatial Variations of Afterslip Following the 1999 Chi-Chi, Taiwan Earthquake
GPS data collected following the 1999 Chi-Chi, Taiwan, earthquake (M$_{w}$ 7.6) reveal noticeable postseismic deformation, with maximum displacements of 25 and 23 cm in horizontal and vertical components during the 15 months after mainshock. Postseismic transients mainly occurred on the hanging wall causing east-west extension, similar to the deformation pattern during mainshock. Distributed afterslip reproduces the overall features in the GPS observations (Hsu et al., 2002, GRL; Yu et al., 2003, JGR). In this study, we try to understand to what extent space-time inversion of afterslip alone can explain the postseismic deformation. The Extended Network Inversion Filter (McGuire and Segall, 2003, GJI) is employed to reveal the fault-slip history. This approach is based on a Kalman filter and capable of extracting spatially coherent signals, such as fault slip transients, and separating these from incoherent signals, such as local benchmark motions. We assume a two-segment fault geometry with a 100 km long, 24$\deg$ dipping and N3$\deg$E trending shallow rupture segment connected to a 40 km wide, horizontal decollement at a depth of 10 km. Our results show that the slip-rate distribution remains roughly stationary over the 15 month period, while the magnitude of the slip rates decay rapidly with time. The average slip rates on the shallow fault decay more rapidly than the deep slip. The slip rate on the shallow dipping fault began at about 0.45 m/yr and decreased to 0.1 m/yr in 200 days, while that on the decollement started at 0.55 m/yr and took about 450 days to decay to 0.1 m/yr. The cumulative slip on lower decollement over 15 months is 1.5 times larger than that on the dipping fault. The total accumulated afterslip moment is 5.78$\times$10$^{19}$ N-m, much more than seismic moment, 2.0$\times$10$^{19}$ N-m, released by aftershocks. In contrast with the limited coseismic slip on the decollement, afterslip is prominent there. The deep postseismic slip is consistent with models including velocity-strengthening friction below the rupture zone.
G13A-0800 1340h
What is the Strength of the Plastic Lower Crust? Comparing the Evidence From Postseismic Deformation and Laboratory Experiments
The plastic strength of many rock types has been well studied in the laboratory. These data suggest that stresses in excess of a few 100 MPa are required to allow deformation at tectonic strain rates near the brittle-ductile transition. The microstructures observed in naturally-deformed rocks agree with such stress levels. By contrast, earthquake-induced stress perturbations, which scale with stress drop, are not more than 10 MPa. Moreover, they decrease rapidly with distance from the edges of an earthquake rupture. Hence, the earthquake cycle produces stress variations of only a few percent in the aseismic plastic crust and mantle. Unless the rock rheology is very different from what is observed in the laboratory, such small stress variations should not produce large creep transients. Yet, transient postseismic deformation in the lower crust and upper mantle is now commonly observed in the aftermath of earthquakes of magnitude 6 and higher. This implies that the strength of the aseismic regions of the lithosphere is much less than inferred from laboratory. We tested the validity of this reasoning with numerical models of postseismic deformation using one of the weakest laboratory-determined flow laws, that of quartzite. The resolution of this paradox may reside in the presence of localized shear zones that differ in their mineral assemblage, fabric, or grain size from the host rock, or possibly a temperature anomaly associated with the downward continuation of brittle faults.
G13A-0801 1340h
GPS Observations Indicate Postseismic Deformation in the Vrancea Region, Romania
The Vrancea region in the South-East Carpathians of Romania is a seismically active area related to past or ongoing lithosphere subduction. One of the outstanding questions is whether the subducting slab is attached, detached or in the process of being detached from the overlying lithosphere. GPS campaigns of 3-D crustal displacements in this region, performed by ISES (Netherlands) in collaboration with SFB-461 (Germany) and the University of Bucharest (Romania), are expected to assist in solving this issue. Measurements have been performed in this area since 1995 and the current GPS network consists of more than 50 campaign points and six permanent GPS stations. However, ongoing deformation related to large ($M_w$ $>$ 7) earthquakes that struck the Vrancea region in 1977, 1986 and 1990 could have a considerable contribution to GPS-observed displacement rates. This so-called postseismic deformation results from viscoelastic relaxation of crustal and shallow upper-mantle low-viscosity zones due to the redistributed stress and strain after faulting. Results from the GPS campaigns combined with numerical postseismic deformation modelling indicate that postseismic relaxation is even the dominant mechanism for present-day horizontal crustal displacements in the Vrancea region. This suggests small ($<$ 1 - 2 mm/yr) to insignificant ongoing plate convergence. Vector solution directions are not very sensitive to the earth model, while magnitudes agree for a seismically and tectonically consistent earth model for the Vrancea region. This has important implications for interpreting neotectonic processes in the Carpathians.
G13A-0802 1340h
Post-seismic deformation following the M7.1 Hector Mine earthquake: An integrated approach utilizing InSAR and seismic data
The post-seismic period is a critical stage of the earthquake cycle, indicating how a fault regains strength following a large magnitude earthquake. Additionally, regions of large post-seismic deformation may reflect the variable stress field created during rupture. Interferometric images provide excellent spatial and temporal coverage of deformation in areas of high coherence such as the Mojave Desert, where the Hector Mine earthquake occurred. We analyze high quality interferometric pairs covering days to years after the Hector Mine mainshock. The InSAR data reveal a complex deformation field with patterns of small (0.1 km) to medium (10 km) spatial scales. We see large deformation to the north, where the fault bifurcates into two strands and significant deformation concentrated along the majority of the surface rupture. We assume a diffusive process is governing the deformation and find a diffusion time constant for each image pixel. The resulting map suggests a spatially variable recovery time around the Hector Mine rupture zone. We compare deformation patterns highlighted by InSAR with seismic array data. Following the Hector Mine mainshock portable seismic array deployments took place in 1999, 2000, and 2001, recording thousands of aftershocks as well as active source detonations within the fault zone. Seismic data were used to probe the main rupture plane of deformation via fault zone trapped waves. These data delimit a 75-100 m wide low velocity zone that experienced some velocity recovery in the years subsequent to rupture [Vidale and Li, 2003; Li et al., 2003]. In addition, a study of shear-wave anisotropy along the Hector Mine rupture zone determined orientation of microcracks in the region around the surface rupture [Cochran et al., 2003]. Preliminary results suggest a good correlation between the InSAR results and seismic data, with areas of high surface deformation corresponding to zones of anomalous seismic velocity and/or large anisotropy. However, each process suggests post-seismic relaxation occurs over different time scales and we investigate different models to explain the observations.
G13A-0803 1340h
Along-strike variations in post-seismic deformation in northern Chile and southern Peru
We use InSAR and GPS data to constrain the spatio-temporal evolution of post-seismic after-slip following three large subduction zone earthquakes in South America. Post-seismic deformation following the 1995 ${M}_{w}$~8.1 Antofagasta, Chile earthquake is barely above the InSAR detection limit, but by combining GPS observations with 36 interferograms we model the post-seismic deformation between the years 1995-2000. If the deformation is after-slip on the fault interface, the equivalent moment magnitude is 10-20% of the co-seismic moment, with the maximum deformation occurring 100 km from the maximum co-seismic slip. There is a pulse of deformation between large aftershocks in 1996 and 1998 that appears distinct from the general decay of after-slip with time. The low magnitude of post-seismic deformation is anomalous compared to other recent subduction zone earthquakes, including the nearby 2001 ${M}_{w}$~8.4 Arequipa, Peru earthquake. After-slip following the 2001 earthquake between 2001-2004 equals about 20-40% of the co-seismic moment. There is no definitive post-seismic deformation following the 1996 ${M}_{w}$~7.7 Nazca, Peru earthquake between 1997-1999, although there is no data spanning the first 51 days after the earthquake. Variations in the depth of rupture during these three earthquakes cannot solely explain the variations in after-slip. We suggest that variations in sediment subducted in each location may control the magnitude of after-slip. The larger thickness of sediments in the region of the 2001 earthquake might also explain why observed after-slip is located in the same area as the co-seismic slip, instead of down-dip, as in other subduction zones.