S23B-1885
A Comparison of two Recent Damaging Earthquakes in the UK
We compare two damaging moderate size British earthquakes. The first occurred near the town of Folkestone on the southeast coast of England on 28 April 2007 with a moment magnitude of 3.8 (4.2 ML). The earthquake was shallow at about 5 km. The event caused considerable damage near the epicenter and was felt throughout south-eastern England. The second earthquake was slightly larger with a moment magnitude of 4.4 (5.2 ML) and occurred on 27 February 2008 in Lincolnshire (eastern central England). Incidences of damage were sporadic but observed over a relatively wide area and the earthquake was felt over much of the UK. This earthquake was a lot deeper at about 20-25 km. Depths are constrained using array data recorded at teleseismic distances. For both earthquakes we are able to invert for the moment tensor using regional broadband data. Both mechanisms are predominantly strike-slip and consistent with the regional stress pattern. We compare the aftershock sequences, source parameters, ground motion observations and distribution of intensity. We determine stress drop and rupture dimensions for each event by modelling the source displacement spectra. We correct for attenuation using a new regional average QLg(f) model for the British Isles. We find that the deeper earthquake is characterized by a significantly higher stress drop. This corresponds to a larger slip relative to the shallower earthquake, while the fault area is about the same for the two events. We use the source parameters and QLg(f) model to compute peak ground acceleration for both earthquakes using the stochastic method. We are able to reproduce the ground motions from the shallower event with this model but for the deeper event, the modelled ground motions are almost a factor of ten lower than observed. This unexpectedly high ground motion does not seem to be caused by local site effects.
S23B-1886
Fault Plane Determination And Possible Triggering Of The 2007 Kuril Islands Earthquake (Mw 8.1)
Two great earthquakes occurred in the Kuril Islands area in 2006 and 2007. The 2006 earthquake (Mw 8.3) occurred on 15 November and was located on the subduction boundary between the Pacific and Okhotsk plates. About two month later, the 2007 earthquake (Mw 8.1) occurred on 13 January and was located on the outer-rise portion of the Pacific Plate. It is relatively-uncommon that, great earthquakes (Mw >= 8.0) happen in the same region within a short span of time, and also great outer-rise earthquakes have occurred only three times since the early 20th century. The aim of this study is to estimate the orientations of fault planes for the 2007 outer-rise earthquake and investigate the possible of triggering for the 2007 earthquake. For the 2007 earthquake, there are currently not clear indications of the geometry so it is difficult to understand the relation between the 2006 and 2007 earthquakes. To relocate the associated seismicity, we used a master event technique. The depths of the master events were determined by teleseismic waveform modelling and other earthquakes were located relative to the master events. For the relocations of seismicity, we used P wave arrival times compiled by the United States Geological survey (USGS) during the period from 20 September 2006 to March 2007. The results of fault plane determinations show that, the fault plane of the 2006 earthquake has a shallowly northwest dipping plane and the 2007 earthquake has more steeply dipping plane to the southeast. To investigate the possible triggering, we estimate the static stress changes with the Coulomb Failure Function Change (Δ CFF). Using the slip distribution for the 2006 earthquake, we calculated the static stress changes in the hypocenter area of 2007 earthquake. The results of the determination of Δ CFF show positive values (conductive to failure) for the hypocenter area of the 2007 earthquake and surrounding area where there was active seismicity during the few months before the 2007 earthquake. From these results, the static stress changes appear to be consistent with triggering of the 2007 earthquake.
S23B-1887
Fault plane Determination for the Niigataken Chuetsu-oki Earthquake Using Tsunami Simulations
On July 16th 2007, the Niigataken chuetsu-oki earthquake occurred along the Japan Sea coast of central Japan. This strike of the fault was estimated to be approximately northeast-southwest but from the preliminary aftershock distribution and crustal deformations, it was difficult to determine if the fault dipped toward the northwest or toward the southeast. A tsunami occurred due to the sea bottom deformation caused by this earthquake. This tsunami was not so large, so it did not cause significant damage along the Japan Sea coast, however, it was clearly recorded by a number of tide-gauge stations along the Japan Sea coast. In this study, we use the focal mechanisms from F-net and estimate the location of the fault plane from arrival times of the tsunami waves at Kashiwazaki station in the southern direction and Sado station in the northern direction. We calculated the sea floor deformation assuming both northwest and southeast dipping fault planes, and then calculated the resultant tsunami waveforms for tide-gauge stations that clearly recorded the tsunami. Using comparisons of the model and observed tsunami waveforms at 6 stations, we try to estimate whether the major slip was on a westward or eastward dipping fault plane. The results show that a fault that dips to the south-east fits both the arrival times and the shape of the tsunami waveforms, better than for a fault that dips toward the northwest.
S23B-1888
Slip Distribution of the 2008 Iwate-Miyagi Nairiku, Japan, Earthquake Inverted from PALSAR Data
On 14 June 2008, the Iwate-Miyagi Nairiku earthquake struck northeast Japan, where active seismicity has been observed under east-west compressional stress fields. According to the Japan Meteorological Agency, the magnitude and the hypocenter depth of the earthquake are 7.2 and 8 km, respectively. The earthquake is considered to have occurred on a west dipping reverse fault with a roughly north-south strike. The earthquake caused significant surface displacements, which were detected by PALSAR, a Synthetic Aperture Radar (SAR) onboard the Advanced Land Observing Satellite (ALOS) employed by the Japan Aerospace Exploration Agency (JAXA). Several pairs of PALSAR images are available to measure the coseismic displacements. InSAR data show up to 1 m of line-of-sight displacements both for ascending and descending paths. The pixel matching method was also used to obtain range and azimuth offset data around the epicentral region, where displacements were too large for the interferometric technique (see Fukushima (this meeting) in detail). We inverted the obtained SAR interferometric and pixel matching data to estimate slip distribution on the fault. Since the geometry of the fault are not well known, the inverse problem is non-linear. If the fault surface is assumed to be a flat plane, however, the non-linearity is weak. Following the method of Fukahata & Wright (2008), we resolved the weak non-linearity based on ABIC (Akaikefs Bayesian Information Criterion). That is to say, the fault parameters (e.g. strike, dip and location) as well as the weight of smoothing parameter were objectively determined by minimizing ABIC. We first estimated slip distribution by assuming a pure dip slip for simplicity, since it has been reported that the dip slip component is dominant. Then, the optimal fault geometry was dip 26 and strike 203 degrees with the location passing through (140.90E, 38.97N). The maximum slip was more than 8 m and most slips concentrated at shallow depths (< 4 km). Without fixing the rake, a large slip area with the maximum slip of about 8 m concentrated in the shallow region was obtained again.
S23B-1889
Source Rupture Process of the 2008 Iwate-Miyagi Nairiku, Japan, Earthquake Revealed from Near Fault Strong Motion Records
We reveal the source rupture process of the 2008 Iwate-Miyagi Nairiku earthquake through the waveform inversion of the near fault strong motion records. At 8:43, June 14, 2008 (JST), a M7.2 shallow inland crustal earthquake struck Iwate and Miyagi Prefectures, northeast Japan. This earthquake, the 2008 Iwate-Miyagi Nairiku earthquake, is a reverse fault event, whose fault plane strikes to the southwest and dips to the northwest according to moment tensor solutions, aftershock distribution, and surface fault break. Strong motion networks, K-NET and KiK-net, operated by the National Research Institute for Earth Science and Disaster Prevention (NIED) obtained lots of the strong motion records at the near source region, which provide the information of the detailed source rupture process. We employ the multi-time-window linear waveform inversion method (Hartzell and Heaton, 1983). S-wave portion of the velocity waveforms (0.1-1 Hz) are used for the inversion. We use mainly borehole data at KiK- net stations. Green's functions are calculated using the discrete wavenumber method (Bouchon, 1981) and reflection/transmission coefficient matrix method (Kennett and Kerry, 1979) on the assumption of 1-D layered velocity structure. The rupture propagation effect within the subfault is included in the Green's function following Sekiguchi et al. (2002). Some stations are located on thick sediment. In order to take into account their effect on the observed waveforms, we construct the velocity structure model for each station referring to a 3-D subsurface structure model compiled for strong motion evaluation of whole Japan (Fujiwara et al., 2006). Validity of the assumed velocity structure is confirmed by simulating the aftershock records. The rupture starting point is set on the hypocenter relocated by Sekine et al. (2008) using double-difference method (Waldhauser and Ellsworth, 2000). Assumed fault plane covers a space of 40 km × 18 km. Its strike angle is 209° referring to F-net moment tensor (NIED) and dip angle is 40° considering the location of the observed surface fault trace. The fault plane is divided into 2 km × 2 km subfaults. The slip history on each subfault is represented by 7 smoothed ramp functions whose duration is 0.8 s at 0.4 s interval. The weight of each function, model parameter in the inversion, is obtained by minimizing the L-2 norm between the observed and synthetic waveforms using the least square method. Restriction of the variation in the rake angle and the spatio-temporal smoothing of the model parameter are applied as constraint conditions to stabilize the inversion. The obtained rupture model shows that the rupture propagated mainly southward. There are two large slip areas. One is located at the hypocenter and the other lies at the shallow part of the fault, which is to the south of the hypocenter. The extent of the large slip area at the hypocenter is rather small. The southern large slip area occupies larger area, from which the seismic moment is largely released between 4 s and 7 s after rupture started. The fitting between the observed and synthetic waveforms is favorable except for IWTH25, which is closest to the epicenter (epicentral distance is 3 km). For further study, we will improve the waveform fitting of IWTH25 since the data of this very near source station constrain the source rupture model better.
S23B-1890
Kinematic Source Rupture Process of the 2008 Iwate-Miyagi Nairiku Earthquake, a MW6.9 thrust earthquake in northeast Japan, using Strong Motion Data
The 2008 Iwate-Miyagi Nairiku earthquake (MJMA7.2) on June 14, 2008, is a thrust type inland crustal earthquake, which occurred in northeastern Honshu, Japan. In order to see strong motion generation process of this event, the source rupture process is estimated by the kinematic waveform inversion using strong motion data. Strong motion data of the K-NET and KiK-net stations and Aratozawa Dam are used. These stations are located 3-94 km from the epicenter. Original acceleration time histories are integrated into velocity and band- pass filtered between 0.05 and 1 Hz. For obtaining the detailed source rupture process, appropriate velocity structure model for Green's functions should be used. We estimated one dimensional velocity structure model for each strong motion station by waveform modeling of aftershock records. The elastic wave velocity, density, and Q-values for four sedimentary layers are assumed following previous studies. The thickness of each sedimentary layer depends on the station, which is estimated to fit the observed aftershock's waveforms by the optimization using the genetic algorithm. A uniform layered structure model is assumed for crust and upper mantle below the seismic bedrock. We succeeded to get a reasonable velocity structure model for each station to give a good fit of the main S-wave part in the observation of aftershocks. The source rupture process of the mainshock is estimated by the linear kinematic waveform inversion using multiple time windows (Hartzell and Heaton, 1983). A fault plane model is assumed following the moment tensor solution by F-net, NIED. The strike and dip angle is 209° and 51°, respectively. The rupture starting point is fixed at the hypocenter located by the JMA. The obtained source model shows a large slip area in the shallow portion of the fault plane approximately 6 km southwest of the hypocenter. The rupture of the asperity finishes within about 9 s. This large slip area corresponds to the area with surface break reported by the field survey group (e.g., AIST/GSJ, 2008), which supports the existence of the large slip close to the ground surface. But, most of surface offset found by the field survey are less than 0.5 m whereas the slip amount of the shallow asperity of the source inversion result is 3-4 m. In north of the hypocenter, the estimated slip amount is small. Slip direction is almost pure dip-slip for the entire fault (Northwest side goes up against southeast side). Total seismic moment is 2.6× 1019 Nm (MW 6.9). Acknowledgments: Strong motion data of K-NET and KiK-net operated by the National Research Institute for Earth Science and Disaster Prevention are used. Strong motion data of Aratozawa Dam obtained by Miyagi prefecture government is also used in the study.
S23B-1891
Characteristics of Foreshocks to Moderate-sized (M > 5.5) Earthquakes in Eastern Taiwan
We investigate earthquakes (M > 2.0) occurred in a seismic zone near the transition corner from subduction to collision in eastern Taiwan from January 1991 to April 2007 by using the double-difference earthquake relocation algorithm. This seismic zone is twelve kilometers long, three kilometers wide and having a depth range between eight to twelve kilometers in the northern end of the Taiwan collision plate boundary. We systematically identify seventeen swarms that have more than fifty earthquakes occurring within one month interval and find that four out of seventeen earthquake swarms located at similar area and formed a specific seismic zone occurred before moderate-sized earthquakes with distances less than thirty kilometers. The moderate-sized earthquake in general located on the east of the seismic zone between eighteen and fifty-nine days after individual swarm. The spatial distribution of swarms is aligned into a plane and their focal mechanism solutions are similar, showing a strike of north¡Vnorthwest and dipping 20° to 30° to the east. As to which structure related to the moderate-sized earthquakes remains unclear. We shall investigate how the specific earthquake swarms relate to the moderate-sized earthquakes nearby.
S23B-1892
InSAR Search for Earthquakes Across the Sierra Nevada-Basin and Range Transition Zone
We utilized interferometric synthetic aperture radar (InSAR) to search for contemporary crustal deformation within the Sierra Nevada-Basin and Range Transition Zone at the latitude of Reno, Nevada. In total, more than 110 interferometric pairs were processed using ERS-1, ERS-2, and Envisat data from the WInSAR and GeoEarthscope archives for the 1992-2008 period in an effort to detect any ground deformation signals associated with multiple M >4.5 earthquakes. In particular, InSAR analysis was focused on a series of events occurring north of Lake Tahoe in 1998, 2003, and 2005, and on a swarm of earthquakes at Reno in 2008. The analyses further included a search for post-seismic deformation associated with the M 5.9 Double Spring Flat event, including several earthquakes in the M 4.5-5.0 range. The principal findings of the study showed that of 12 earthquakes of M >4.5 occurring in the region since 1992, we could identify InSAR ground deformation signals with only three sequences: the 1994 Double Spring Flat (M 5.9), the 2004 Adobe Hills (M 5.0-5.6), and the 2008 Reno-Mogul (M 4.7) events. The remaining earthquakes, which ranged in magnitude from M 4.6 to 5.0, did not show any evidence of ground deformation on multiple InSAR pairs covering the time of the events. We could find no visible evidence of the 1998 (M 4.9) and 2005 (M 4.8) Lake Tahoe events, eliminating the one visible InSAR signal as being related to groundwater pumping. We attribute the general lack of surface signals to the small magnitudes and deep focal depths (> 10 km). The 2008 Reno-Mogul swarm did, however, exhibit a prominent deformation signal with as much as 4 cm of LOS change over a 10-15 km area, and it is the smallest magnitude earthquake that we have detected with InSAR in the western Basin and Range. We attribute the visible ground deformation to the shallow depth of the earthquake (< 5 km). Differences between the LOS changes observed on ascending and descending pairs indicate that the main event was related to right-normal oblique-slip motion on a northwest-oriented fault, in close agreement with the aftershock pattern. We used a simulated annealing algorithm to estimate the location, orientation and slip parameters of a best-fitting single plane offset that explains the InSAR and the GPS measured surface deformation. We found numerous groundwater pumping signals in many of the alluvial basins. These signals are strongly controlled by Quaternary faults, and the signals can be mistaken for tectonic deformation. Most of the groundwater signals are seasonal in nature exhibiting alternating cycles of elastic subsidence and uplift. These observations together with wavelength filtering of radar phase data allowed us to segregate the seasonal signals and to determine that the ground deformation along the faults was not tectonic.
S23B-1893
Strong Crustal Earthquakes in Central Aleutian Islands in 2006-2008: Implications for the Block Rotation Model
Between 2006 and 2008 six strong crustal earthquakes occurred in central and western Aleutian Islands. The series started with a M6.5 event on June 14, 2006 that occurred immediately west of Kiska Island and had a M6.0 aftershock. Two weeks later a M6.2 event occurred near Buldir Island. On April 15 and 16, 2008 a M6.4 and M6.6 earthquakes occurred in the Amchitka Pass area and on May 2, 2008 a M6.6 earthquake occurred between the Kanaga and Tanaga Islands. The events are located along a 450 km long segment of the Aleutian arc and all have strike-slip faulting mechanisms with varying orientations of the focal planes. The Alaska Earthquake Information Center reported hundreds of aftershocks for each event. Due to the seismic network limitations, however, the event locations are poorly constrained and the fault planes can not be easily determined. Geist et al. (1988) proposed a block rotation model for the central and western Aleutians. Five blocks of various sizes have been identified based on geomorphic evidence. Boundaries between the blocks were delineated based on the submarine fault-controlled canyons. Northern boundaries were defined as the southern edge of the corresponding summit basins, which were formed as result of rotation of the blocks. Southern boundaries were prescribed along the seaward edge of the arc massif. The block boundaries were meant to delineate regions of cohesive movement. They are bounded by zones that are significantly disrupted by normal and strike-slip faults. The 2006-2008 earthquakes are located either on or north of the delineated northern boundaries of the Buldir, Rat and Delarof blocks. A search of the Global CMT catalog produced a handful of crustal strike-slip events in the region. The most notable are the earthquakes that occurred after the great megathrust events in the region, such as a Ms7.2 event in 1966 that followed the M8.7 1965 Rat Islands earthquake and a series of strong crustal shocks (up to M6.6) near Atka Island following the M8.0 1986 Adak earthquake. Most of the recorded crustal strike-slip events are concentrated near the block boundaries. We use waveform cross-correlation and double-difference relocations to identify aftershock distribution and determine preferred fault planes of the 2006-2008 earthquakes to identify how these events characterize the crustal block model.
S23B-1894
Relative Source Time Function Studies of Earthquakes in Southeastern Alaska
We are using the Relative Source Time Function (RSTF) method to determine the source properties of earthquakes within southeastern Alaska, a region extending from the Queen Charlotte Islands to Yakutat Bay. In our approach we deconvolve the spectral quotient of the P-arrival of a small event from that of a larger event. The arrivals are selected using a tapered cosine window, then we use a water level technique to stabilize the quotient in the frequency domain, and we apply a bandpass or highcut filter before inverse transforming our result. Our first goal is to compare the source processes of earthquakes along the Queen Charlotte-Fairweather fault system to those occurring off these major plate bounding faults to determine if there are differences in stress drop and source duration between these regions. Secondly, we hope to determine if observed differences in source processes of events along the Queen Charlotte-Fairweather fault system may be related to the location of the events relative to known fault asperities and segment boundaries. We have identified about twenty events occurring between 1995 and 2008 distributed throughout the study area. In addition, we have digitized and analyzed seismograms of older events occurring in July 1973 located in the Cross Sound area.
S23B-1895
Background seismicity study of Southeastern Alaska
Southeastern Alaska is dominated by strike-slip motion along the Queen Charlotte-southern Fairweather fault (QC-SFF) system in the south and transitions to oblique convergent motion partitioned between strike-slip motion along the Denali and northern Fairweather fault systems and thrusting along faults of the St. Elias region. Geologic complications are further increased by the subduction of the Yakutat microplate beneath North America and glacial processes. By studying regional background seismicity we intend to better determine the current state of stress of southeastern Alaska from the Dixon Entrance to Yakutat Bay. We gathered phase data for over 3000 earthquakes of magnitude <5 that occurred from 1973-2005 from Alaskan and Canadian databases. We relocated these earthquakes using the Double-Difference joint hypocenter method. We used these relocations and first motion data to estimate the regional stress field. We combined stress information with GPS, magnetic and gravity datasets in order determine how plate motion is partitioned in this region and to identify other potentially active faults.
S23B-1896
Using Relative Source Time Functions to Calculate Stress Drops for Earthquakes in South-Central Alaska
We present results of a study of stress drops of intraslab earthquakes in South-Central Alaska (M≥5, depths of 25-100 km) in order to examine the effect of interface coupling on the source properties of these events. The stress drops are calculated using the Relative Source Time Function (RSTF) approach. The RSTF for a larger event is obtained by deconvolving a small event (which is considered its EGF, empirical Green's function) from the spectral quotient of the windowed first P wave arrival of a moderate to large event. First arrivals were selected using a tapered cosine window. We then used a water level technique to stabilize the quotient in the frequency domain. Next we applied a narrow bandpass in the low frequency range and/or a Gaussian filter and finally inverse Fourier transformed the RSTF to the time domain. Analysis of RSTF at different stations provide information on directivity of the ruptures, as well as stress drop variation between intraslab events along portions of the Alaska-Aleutian megathrust having different amounts of coupling. The stress drop is directly proportional to the energy (integral of squared function) under the RSTF. We have focused on events occurring since December 1997 to the present recorded on broadband or strong motion instruments.
S23B-1897
The near-sonic rupture velocities of the intracontinental Kokoxili Mw 7.8 and Denali Fault Mw 7.9 strike-slip earthquakes imaged with teleseismic P-waves
The Denali and Kokoxili strike-slip earthquakes are two of the longest intracontinental ruptures in recent history. Previous studies report a range of rupture velocities. We image these earthquakes by reverse-time migration of the intermediate-frequency P wavetrain recorded by global broadband seismometers. This technique provides a relatively direct measure of rupture velocity (speed and direction) as constrained by the radiated seismic energy--immune from restrictive assumptions or rupture-speed bounds placed on the solution. We compare our results with published seismic, GPS-displacement, and surface-slip inversion results. Both ruptures were initially subshear and transitioned over a distance no longer than 40 km to supershear speeds close to the P-wave speed of ~5.6 km/s. We investigate the accuracy of our results with synthetic data and experiment with using different imaging parameters and seismic subnetworks. These tests allow us to rule out the possibility of subshear speeds along the supershear segments. Although we cannot exclude supershear speeds of 4.5 to 6.5 km/s, our most reliable rupture velocities of ~5.6 km/s are close to the local P-wave velocities. We hypothesize that these intracontinental faults have weak shear strengths or high breakdown slips or crustal rigidities, and experience at least moderate slip or slip-rate weakening. Our observations and previous published results lead us to speculate that very long, surface- extending faults with general homogeneity in pre-stress and fault strength, together with smaller adjacent fault segments to provide triggering, may be necessary ingredients for the sub-Rayleigh to supershear rupture speed transition in strike-slip earthquakes.
S23B-1898
The Rupture Processes of the 1958 Fairweather Fault, Alaska, Earthquake Sequence
The 1958 Fairweather fault earthquake sequence included one of the largest strike-slip events (M~7.9) to occur in North America over the past century. Fletcher and Freymueller (2003) have suggested only 75 years is needed to build up slip for a similar sized event along this northernmost segment of the fault. I have re-examined the rupture process of the 1958 sequence through body waveform modeling, intensity studies, aftershock relocations, and modeling of induced static stress changes. The mainshock ruptured primarily to the northwest with an aftershock zone extending over 400 km. Most aftershocks are located southeast of the region of maximum moment release during the mainshock and some appear to have activated faults of the Cross Sound/Transition zone system. Aftershocks also extend westward to the Icy Bay region and northward to the Denali and Totschunda faults. The intensity data are consistent with waveform modeling results that suggest maximum moment release in the mainshock occurred along the fault between Mt. Fairweather and Dry Bay.
S23B-1899
Hypocenter Relocation and Source Parameters of the 2006 Kiholo Bay, Hawaii Earthquake Sequence
We studied aftershock distribution of the Kiholo Bay earthquake (MW 6.7), which occurred on October 15, 2006 (UTC) beneath the NW part of the Island of Hawaii, to determine the fault plane from the two nodal planes. To more precisely locate aftershocks relative to one another, we used the cross correlation technique [e.g., Wolfe et al., 2003] and double-difference hypocenter determination [Waldhauser and Ellsworth, 2000]. The relocated cloud of aftershocks has a NW-SE trend with a SW dip, which is consistent with one of the nodal planes given in the Harvard moment tensor solution. The hypocenter of the main shock is located close to the eastern edge of the activity, suggesting that the rupture was almost unilateral to the NW. We also estimated corner frequencies of 31 earthquakes (2.5 < ML < 5.0) which occurred in 2006 and 2007 near the source region of the Kiholo Bay earthquake to investigate the spatial variation and scaling relationship of stress drop. We used strong-motion and short-period seismic stations operated by the USGS Hawaiian Volcano Observatory (HVO). We picked earthquakes with ML 3.7 and 2.0 as empirical Greenfs functions for strong-motion and short-period waveforms, respectively. We fitted the source spectra with the omega squared model and estimated corner frequencies of P waves from the deconvolved waveforms. We then calculated stress drops from the model of Madariaga [1976] assuming that the local magnitude determined by HVO is equal to the moment magnitude. The stress drops range from 0.040 to 23 MPa. We found that earthquakes close to hypocenters of the main shock and the largest aftershock (ML 5.0) have smaller stress drops. No significant relationship between stress drop and seismic moment was observed but it is difficult to argue that the stress drop is constant because of the large variability. Two of the 31 earthquakes occurred before the Kiholo Bay main shock and their stress drops have no systematic differences compared to the values of the 29 aftershocks.
S23B-1900
Comparisons of corner frequencies and strain-drops from P and S waves generated by earthquakes along the Karadere-Düzce branch of the North Anatolian fault
We estimate the strain-drops of aftershocks of the 1999 İzmit and Düzce earthquakes using S waves recorded by a local seismic array along the Karadere-Düzce branch of the North Anatolian fault in the 6 months following the İzmit mainshock. The analysis method is associated with separation of source, travel-time and station spectral terms and stacking results at several stages to enhance the signal-to-noise ratio. The strain-drops are obtained by fitting iteratively the separated source spectra of 201 nearest neighboring events in different amplitude bins to the ω 2 source spectral model. The strain-drops obtained from S waves generally match those derived in an earlier study from P waves (Yang et al., 2008). However, the calculated ratio of P/S corner frequencies is nearly 1.0, instead of the 1.5 value predicted by Madariaga (1976). The discrepancy could be caused by the existence of P coda waves inside the S wave window. We also measure the rise times and pulse widths of P and S phases from stacked waveforms generated by 36 repeating clusters (Peng and Ben-Zion, 2006). The median ratios of the corner frequencies of the P and S waves from pulse width of repeating clusters at the different stations range from 1.1 to 1.7. The variations of values for the same source at different locations suggest that these results are affected to some extent by propagation and site effects. To further reduce the effects of the P wave coda and propagation path, we plan to apply our stacking technique to the S coda waves in multiple windows. Since the S coda waves primarily consist of S-to-S scattering and are further from the P waves, they are expected to contain valuable information on the S source spectrum with reduced effects from the P coda waves.
S23B-1901
Spatial aftershock distribution of the 26th December 2004 great Sumatra-Andaman earthquake in the northern Sumatra area
In July 2005, 20 ocean bottom seismometers were deployed in the northern Sumatra area. More than 1000 events were identified during the 12 days recording period. After relocation, the seismicity shows different patterns on each side of the Upper Splay Fault. East of this boundary, beneath the Aceh basin, the 30-60 km deep earthquakes depths outline the subduction interplate boundary. These earthquakes have been attributed earlier to after slip activity. West of it, the aftershock distribution is strongly influenced by the N-S oceanic fracture zones of the subducted plate. Two N/S trending clusters of 10 to 50 km deep earthquakes observed below the lower wedge along the 93.2 and 93.6°E meridians are interpreted as reactivated fracture zones. The post-seismic activity suggests a significant influence of the N-S lineaments on the oceanic plate, the toe of the wedge and the lower wedge. These oceanic fracture zones and overlying sediments of the wedge may play a role of barrier to stop the co-seismic rupture propagation during the 2004 great Sumatra-Andaman earthquake. The geological features determined in the upper plate are dominated by the NS and NNW trending directions, which show the influence of the underlying oceanic fracture zones on the structure of the upper plate. The dextral wrenched system of thrust faults within the wedge would be affected by the differential stress accumulation along the diffuse India/Australia plate boundary. The relatively high free-air gravity, the co-seismic slip distribution and the geological structural study support the presence of the fracture zones in the Outer Ridge. We propose that the inter-plate boundary does not cut through the Fracture Zone Ridges to emerge along the Frontal Thrust but instead ramps upward along the Lower Splay fault.
S23B-1902
Spatial Study of November 2004 seismic event in Costa Rica (Mw=6.4)
The Costa Rica seismic network OVSICORI-UNA located a Mw=6.4 (Harvard CMT) earthquake that occurred on November 20, 2004 at 08:07 UTC. The seismic activity was relocated using a double difference technique and a probabilistic non-linear global search method. The seismic event was located in the central Pacific part of Costa Rica with a focal depth of 25-km and about 100-km inland from the Middle America trench. The seismic event occurred in the deepest part of the crust, with aftershocks distributed to the surface. The local fault delineated by the seismic sequences is part of the fault system that delineates the boundary between the Caribbean plate and Panama Block. We used local data from this seismic sequence to invert for a moment tensor solution of some events magnitude Mw=4.5. We used CMG-6TD intermediate velocity instruments installed by OVSICORI-UNA network in 2003 for the moment tensor inversion.
S23B-1903
Seismicity and Stress in the Southern Jalisco Block, Mexico.
Fault plane solutions for a group of 41 earthquakes with Mw magnitudes between 4.9 and 8.0, detected between November 1980 and May 2007, were extracted from the CMT catalog and used to investigate possible spatial and temporal variations of stress in the south part of the Jalisco Block. Using rigorous statistical tests we decide on the quality and variability of the earthquake source mechanisms and with the use of FMSI and LSIB stress tensor inversion techniques and cumulative misfit analysis find, at the 95% confidence level or more, significantly different regions and periods of stress regime. An important difference in seismicity and stress between the southern and northern regions of the Jalisco Block is highlighted by the regular occurrence of earthquakes to the south, in stark contrast with the scarcity of seismicity with Mw > 4.9 north of 19.36°N, a region that is illuminated with smaller earthquakes when local temporal or permanent deployments of seismographs are used, as noted by previous studies. We conclude that the region located near the coordinates 105°W and 18.43°N appears as a particularly diffuse zone affected by various stress styles and that the period following the great Mw 8.0 earthquake on October 9, 1995 appears as a short-lived change in stress induced by the great shock.
S23B-1904
Local Seismicity Analysis of the Ometepec, Guerrero Area
The Ometepec Guerrero area in southern Mexico (16° - 17.5° N, 99° - 97.5° W) is a region of high seismic potential. In spite of the geological importance of the area, few works have been related to the correlation between local seismic distribution and geological features, this motivated us to the installation of a local seismic network that has been in operation since early May 2008; consisting of four GURALP broadband three component stations. The signals are recorded continuously at 100 samples per second. The network is complemented with a nearby permanent station (PNIG) from the National Seismologic Service (SSN). We intend to maintain the network operating for approximately three years. We present such a preliminary analysis of results of the first 4 month period of observations. Most of the recorded events have epicenters on land-and within the network. A few percent were located offshore with a wider margin of location error. The distribution of epicenters delineate several clusters that could be probably related to the surface expression of geological features; however, this situation will be clarified as more data is gathered.
S23B-1905
The 2005 Tarapaca, Chile, Intermediate-depth Earthquake: Evidence of Heterogeneous Fluid Distribution Across the Plate?
The physical mechanism of intermediate-depth earthquakes remains unsolved, and dehydration embrittlement in subducting plates is a candidate. An earthquake of Mw7.8 occurred at a depth of 115 km beneath Tarapaca, Chile. In this study, we suggest that the earthquake rupture can be attributed to heterogeneous fluid distribution across the subducting plate. The distribution of aftershocks suggests that the earthquake occurred on the subhorizontal fault plane. By modeling regional waveforms, we determined the spatiotemporal distribution of moment release on the fault plane, testing a different suite of velocity models and hypocenters. Two patches of high slip were robustly obtained, although their geometry tends to vary. We tested the results separately by computing the synthetic teleseismic P and pP waveforms. Observed P waveforms are generally modeled, whereas two pulses of observed pP require that the two patches are in the WNW-ESE direction. From the selected moment-release evolution, the dynamic rupture model was constructed by means of Mikumo et al. (1998). The model shows two patches of high dynamic stress drop. Notable is a region of negative stress drop between the two patches. This was required so that the region could lack wave radiation but propagate rupture from the first to the second patches. We found from teleseismic P that the radiation efficiency of the earthquake is relatively small, which can support the existence of negative stress drop during the rupture. The heterogeneous distribution of stress drop that we found can be caused by fluid. The T-P condition of dehydration explains the locations of double seismic zones (e.g. Hacker et al., 2003). The distance between the two patches of high stress drop agrees with the distance between the upper and lower layers of the double seismic zone observed in the south (Rietbrock and Waldhauser, 2004). The two patches can be parts of the double seismic zone, indicating the existence of fluid from dehydration, whereas the region of negative stress drop is in the absence of fluid. In the background environment of negative stress drop, fluid can change the negative stress drop to positive, due to pore pressure variation (e.g. thermal pressurization).
S23B-1906
Detailed source process of the 2007 Tocopilla earthquake and its main aftershocks
We investigated the detail rupture process of the Tocopilla earthquake (Mw 7.7) of the 14 November 2007 and of the main aftershocks that occurred in the southern part of the North Chile seismic gap using teleseismic broadband and strong motion data. The earthquake happen in the middle of the permanent broad band and strong motion network IPOC newly installed by GFZ and IPGP-CNRS, and of a digital strong- motion network operated by the University of Chile. The Tocopilla earthquake is the last large thrust subduction earthquake since the major Iquique 1877 earthquake which produced a destructive tsunami. The Arequipa (2001) and Antofagasta (1995) earthquakes already ruptured the northern and southern parts of the gap, and the intraplate intermediate depth Tarapaca earthquake (2005) may have changed the tectonic loading of this part of the Peru-Chile subduction zone. The Tocopilla earthquake raises some disturbing questions: why this earthquake didn't extent further north ; what has been the role of the Mejillones peninsula in the south which seems to act as a barrier? We studied the detailed source process using the strong motion data available. The strong-motion data show clearly two S-waves arrivals, allowing the localization of two sources. The main shock started north of the segment close to Tocopilla. The rupture propagated southward. The second source was identified to start about 20 seconds later and located 50 km south from the hypocenter. The earthquake ruptured the interplate seismic zone over more than 150 km and generated several large aftershocks, mainly located south of the rupture area with the same focal mechanism, except for the largest one that took place on the 16 December. This event is a down-dip compressional event (slab push) placed down dip of the main interplate coupling zone at the southern end of the main event rupture zone. Finally in order to understand whether the northern gap has actually been reduced or not by the occurrence of the Tocopilla earthquake which was smaller than expected, we will discuss different spontaneous rupture scenarios.
S23B-1907
Slip Rates along the Yunnan Fault Zones Estimated From Repeating Microearthquakes
Fault slip rate has been widely estimated by geologic and geodetic observations. Most of the geologic and geodetic observations are, however, surficial measurements which require an assumed rheology to estimate the slip rate at seismogenic depth. The discovery of repeating microearthquakes has provided a new means for inferring the slip rate at depth along active faults as well as for imaging potential asperities. Repeating earthquakes are a series of seismic events regularly occurring on a patch of a fault plane. These earthquakes usually generate nearly identical waveforms when recorded at the same station. In this study, we examined the waveform data recorded by the Yunnan digital Seismic Network (YSN) between 1999 and 2006 to estimate slip rates along active fault zones in the Yunnan province, south China. A cross- correlation based analysis yielded the identification of 412 doublets and 353 multiplets, which count up to ~45% of the local seismicity. Most of the sequences are aperiodic with recurrence intervals varying from a few minutes to hundreds of days. There are, however, several quasi-periodic sequences which allow us to infer slip rates. For example, using two clusters occurring in the Lijiang and Ninglang areas, we obtained a slip rate of approximately 5 mm/yr at depth of ~23 km. This estimate is consistent with geological and surface GPS measurements. We also found some variations in slip rate in the study region. Different segments along the same fault could have distinct difference in slip rate. For example, the Xiaojiang fault is characterized by a low slip rate in the northern segment, and a high slip rate in the central and southern segments that is consistent with the local seismic activity.
S23B-1908
Finite Fault Modeling for Moderate-sized Earthquakes (M~5) in Southern California
We develop a finite fault inverse system to study the rupture process of moderate-sized earthquakes within a regional network, such as the California Integrated Seismic Network (CISN). The slip model is constrained by matching the body waves recorded at local distances. We pre-calculate the earth response using a 1D local structure model but correct the 3D structure effects by adding time shifts and wavelet-based scale- dependent amplitude corrections, which could be either directly calculated using a precise 3D earth structure or calibrated by systemically analyzing background seismicity and aftershocks near the hypocenter. Comparing with the empirical Green's function approach, this method does not require a small event that has the same focal mechanism as the mainshock. We adopt this method to study the 2008 Mw 5.4 Chino Hills California earthquake. Our preliminary result showed that the optimal fault plane of this event has a strike of 291 degrees and dips 59 degrees to the northeast. The rupture initiated at a depth of 14.75 km and then propagated bilaterally for 1.5 sec. The inverted total seismic moment is 2.02e+17 Nm (Mw 5.47), consistent with the CISN moment tensor solution.
S23B-1909
Spatial variations of seismic parameters and seismic properties in California
The b value in Gutenberg-Richer relationship is critical for both hazard analysis and physical understanding of earthquakes. In California, people usually assume that the magnitude distribution is independent of location, but it might not be apply to other regions such as creeping zone, geothermal and volcanic regions. We divide California into 1x1 degrees grid and calculate the b value in each cell and estimation the variation of b value caused by magnitude errors using simulated earthquakes. The spatial variation of p value in Omori law will help people to find the whether and how the stress field, geographical condition and temperature influence the aftershock sequences. We use ETAS model to estimate the p value in each cell and find the statistical spatial correlations. We explore the relationship between b, p values and heat flow in California to find whether there are more small earthquakes in high temperature area and whether the high temperature shortens the stress relaxation time. We explore the background rate in California. We find the stationarity of background rates in southern and northern California are significant different. And we also explore the variation of the ratio of background events to total events across different stress regimes to discuss how stress influences the earthquake physicics.
S23B-1910
Rupture models for the recent Chino-Hills Earthquake(SC)
Broadband regional and teleseismic data are used to determine a kinematic model of slip. The fault parameters are Mw=5.4, strike of 285°, dip of 62°, and a rake of 131°, as determined by modeling a large number of TriNet waveforms with the Cut-and-Paste(CAP) methodology. The prediction of teleseismic waveforms confirms the depth estimate of 15km with source duration of 1s. These waveforms showing clear depth phases are similar to a strong aftershock following the 1987 Whittier earthquake, Bent and Helmberger (1989), with the same strike located about 30km northward on the Whittier fault zone. We use the CAP+ technique [Tan and Helmberger (2006,2008)] to model aftershocks and foreshocks from the Chino-Hill region to calibrate paths and establish amplitude corrections in the (.5 to 2Hz) frequency band. Directivity can then be determined accurately. The combination of teleseismic and regional data enables estimation of both vertical and horizontal directivity. Preliminary models indicate rupture towards the northwest.
S23B-1911
Source Properties and Early Aftershocks of the July 29 2008, Mw 5.4, Chino Hills Earthquake
On July 29 2008 a Mw 5.4 earthquake shook Southern California. Its epicenter, near Chino Hills, lies in an area of dense seismological instrumentation, and creates a special opportunity to further understand the source properties of moderate sizes earthquakes. We present an analysis of strong motion records of the Chino Hills earthquake. Using empirical Green's functions, we estimate the apparent source time function (ASFT) from P- and SH-waves recorded within 30 km epicentral distance by the Southern California Seismic Network, with waveform data available through the Southern California Earthquake Data Center. ASTFs are obtained by applying a regularized time-domain deconvolution between mainshock and aftershocks records, with positivity constraints. We investigate the second-order spatio-temporal moments of the source, following the method proposed by McGuire (2004). Our analysis reveals a source duration significantly shorter than 1 second, in the short end of the usual value range observed for earthquakes of this magnitude. Owing to the particularly complete azimuthal coverage of the available recordings, the method can potentially yield insights on rupture directivity and help resolve the fault plane ambiguity. Indeed the aftershock locations lie on a deep sub-horizontal plane, and do not help differentiating which of two possible dipping fault planes was activated during the mainshock. Moreover, we are identifying and relocating early aftershocks buried in the first 5 minutes of the mainshock coda. We are investigating whether these early aftershocks delineate better the mainshock fault plane.
S23B-1912
Earthquake-induced tendency of the Earth Oblateness change
It has been known (since Chao and Gross, 1987) that globally earthquakes have a strong tendency in making the Earth less oblate, or rounder; that is, the Earth¡¦s oblateness parameter J2 has been decreasing slightly but secularly because of the cumulative effect of earthquakes. As the dynamic explanation is still outstanding, we calculate the earthquake-induced, cumulative-to-date change in J2 as a function of the radius in the Earth interior, ΔJ2(r), of all earthquakes according to the Global CMT catalog (close to 30,000 events from 1976 to October 2007). The calculation is via the normal-mode summation scheme using the CMT moment tensor for the seismic source mechanism on the SNREI Earth model PREM. The radial profile function ΔJ2(r) thus has the physical meaning of the contribution, per unit (normalized) radial distance, to the whole-Earth oblateness change cumulated of earthquakes since 1976. The result for ΔJ2(r) indeed shows a decrease through the mantle and core (including the core-mantle boundary), except near the crustal depth where most earthquakes happen. We then attribute ΔJ2(r) to two major earthquake types: normal faulting and thrust faulting. The result shows that the secular tendency of ΔJ2(r) are distinctive according to source mechanisms: decreasing ΔJ2(r) by thrust faulting and increasing ΔJ2(r) by normal faulting. The thrust faulting dominates the ΔJ2(r) tendency over that of normal faulting by about six times. We further examine the implications of this result in terms of global earthquake dynamics.
S23B-1913
Aftershock distribution relative to main shock slip and trench parallel gravity anomalies
A recent study by Llenos and McGuire (2007) uses second moments to create rupture models for 15 great subduction zone earthquakes and compares the centroid and rupture extent results relative to trench parallel gravity anomalies (TPGA). They find that for 75 percent of the events studied, TPGA increases between earthquake centroid and the edges of rupture, indicating structural control on rupture arrest. As an independent test of this relationship, we are using a modified Joint Hypocenter Determination code to relocate aftershocks of the events studied by Llenos and McGuire (2007). Available arrival time data have proven adequate to robustly relocate 8 of the 15 aftershock sequences. Our preliminary results show good agreement between the aftershock distributions and the Llenos and McGuire (2007) rupture extent and directivity results for 5 events. We do not observe a clear-cut relationship between TPGA features and the aftershock zone, however. We will explore the use of waveform cross-correlation to provide additional and more accurate arrival time picks for first arrivals and depth phases in order to extend our analysis to the remaining sequences.