S51A-0972
Nonisotropic Scattering of S-Waves in the Crust Around the San Jacinto Faulting Zone, Southern California
We studied the energy distribution of seismic scattering based on the source factor results ( SF in the single{\it -}backscattering model of Aki and Chouet, 1975, {\it JGR} ) obtained from polarization{\it -}dependent {\it S-}coda amplitudes. We have tested our algorithm on band-passed ( 2 {\it -} 4 Hz ) waveforms of a {\it M$_{L}$}= 5.1 local crustal earthquake ( {\it h} = 18.7 km ) recorded by the ANZA broadband network in southern California. We analyzed the records of 10 stations, of which 7 are < 20 km away. Over all stations {\it S-}coda waves with a polarization azimuth of $73° ( clockwise from North ) and vertical angle of $9° ( positive values downward ) have the maximum value of SF. We find the polarization direction of the maximum energy of backscattering {\it S-}waves is consistent with the orientation of a normal plane of the scattering patterns, strike $164° and dip $81°. This suggests the distribution of the scatterers is nonuniform. They are predominantly oriented S$16°E and are responsible for the maximum value of the backscattering energy of {\it S-}waves in the crust beneath ANZA. This orientation is consistent with the axes of compression obtained from results of seismogenic deformation field ( Unruh {\it et al.}, 1996, {\it JGR} ) in southern California. The scatterers ( opened cracks? ) are parallel the maximum compressional stress.
S51A-0973
Shear-wave splitting in the source region of the 2004 M6.8 Niigata-ken Chuetsu earthquake, central Japan
A large earthquake with a magnitude of 6.8 occurred in the central part (Chuetsu district) of Niigata Prefecture, central Japan, on October 23, 2004. A temporary seismic network composed of 54 stations was installed by Tohoku University two days after the main shock. All stations have 3 components of seismographs with a sampling rate of 100 Hz. The observation lasted for about a month and many small aftershocks were detected. We analyze shear wave splitting using the waveform data obtained by the aftershock observation and investigate anisotropy around the focal area of the main- and after-shocks. We also focus on the difference between the anisotropy within an earthquake fault zone and that in the surrounding area. We use M>1.5 earthquakes that have incident angle of less than 35 degrees. The leading shear wave polarization direction (fast direction) and the lag time of two quasi shear waves (DT) are obtained by applying the cross-correlation method [Ando et al., 1983]. Seismograms are band-pass filtered at 2-8 Hz at the cross-correlation computation. We search for the optimizing pair of fast direction and DT from the two horizontal component of each seismogram using a grid search, in the range of 0-180 degree with increments of 5 degrees for fast direction, and in the range of 0-1s with increments of 0.01s for DT. We set the time windows of the cross-correlation computation to be the first one cycle of the shear wave for each seismogram. We observed shear wave splitting at 37 stations, and the obtained results showed different pattern of shear wave splitting between the northeastern and southwestern parts of the study area. Fast directions are E-W or ESE-WNW at many stations in the southwestern part of the study area. This direction is consistent with the predominant P-axis of the focal mechanisms of microearthquakes [Kosuga, 1999], which suggests that the regional stress field control the anisotropy. However, NE-SW fast directions are dominant at stations in the northern part of the study area inconsistent with the regional stress field. We compared the present results with the seismic velocity structure of the study area determined by Okada et al. [2005]. General feature of the velocity structure is that the western part of the focal area shows lower velocity compared with the eastern part of the focal area. So the low-velocity area corresponds to the area that fast directions are E-W and the high-velocity area corresponds to the area that fast directions are NE-SW. Many aftershocks occurred in the northeast of the main shock and the strikes of major active faults are NE-SW there. Our results with NE-SW fast directions observed in the northeastern part of the study area may be related to local fault alignment. The results obtained in this study suggest the small-scale (20-30km) variation in shear-wave splitting around the focal area of M6.8 Niigata-ken Chuetsu earthquake.
S51A-0974
Surface Tectonics-Matched SKS Splitting: Vertically Coherent Mantle Deformation or Crustal Anisotropy?
SKS splitting measurements are generally assumed to reflect mineral alignment in the upper mantle, which in turn is caused by mantle deformation or flow. A crustal contribution to SKS splitting is usually considered negligible. More recently, ultrasonic laboratory measurements have suggested that crustal materials can possess anisotropy strong enough for a layer of order 10 km thickness to produce splitting comparable to a mantle signature. A counterargument, against the relevance of crustal SKS splitting, is that the crust is unlikely to have homogeneous anisotropy over 10 km thick layers. We perform numerical modelling of splitting in realistic crustal models with anisotropy that varies on a < 1 km scale in depth. The crustal models are built by taking elastic tensors from ultrasonics and varying their orientations according to geological mapping of foliation and lineation changes on a macroscopic scale, with examples from the Californian Sierra Nevada and other regions. We find significant splitting in these models. SKS splitting is often found to be related to surface tectonics; usually, this observation is interpreted as deformation of the upper mantle coherent with the crust. Our modelling offers an alternative explanation, namely that a significant part of the splitting signal may stem from the crust itself.
S51A-0975
Testing a suite of multi-layer dipping anisotropy models with teleseismic shear wave array data
While shear-wave splitting provides wonderful new constraints with respect to mantle deformation, nonetheless most datasets are limited in their ability to constrain anisotropic models more complex than a single layer with a flat (horizontal) fast velocity axis (FVA). Often this limitation results from single station data which is often contaminated by signal generated noise. In our analysis, data from a 30-station 90-km diameter broad-band array deployed in SW Montana provides 9 SKS and 7 direct-S events that are stacked to attenuate signal generated noise and provide good error estimation crucial to statistical model ranking. To quantify the resolving power of this dataset, the posterior model probability density (PPD) volumes for a suite of seven models are compared. The forward problem is calculated using ray-theory and the model is parameterized as a set of flat velocity layers with possible dipping FVA. Source normalization is performed using the cross-convolution method which approximates the source waveform as linearly polarized before entering the anisotropic receiver-side structure. To search the N-dimensional model space for low misfit models, the Neighborhood Algorithm is used. The seven models tested are the combination of one, two, or three layers with each layer either possessing a flat FVA or a dipping FVA. Comparison of the different model PPD volumes (using 1- and 2-D probability density marginals) and F-test statistics shows that models more complex than a single layer of flat FVA anisotropy are required by the data. Our best model is two layers of dipping anisotropy whose fit to the data are better than a single flat layer model at >80% confidence. The PPD for this model is compact and uni-modal and can be reasonably approximated as a multivariate Gaussian. However, a three layer dipping anisotropy model produces non-compact multi-modal marginals with large covariance between model parameters. We conclude that: 1) use of direct S- arrivals provides a significant increase in model resolution; 2) models with a lower layer FVA that dips 20-40-a down to the SW is a robust requirement of the data; 3) array data provide improved resolution with respect to single station data. The dipping FVA found may be attributed to the Yellowstone plume (see additional URL) that rises 100 km to the SE of our array.
S51A-0976
Crustal Attenuation within the Turkish Plateau and Surrounding Regions
We have imaged lateral variations in the seismic attenuation in the crust from the inversion of inter-station Lg Q measurements in Turkey and the surrounding regions. We used data collected from three temporary arrays (ETSE, WTRN, ASN) and three permanent networks (KOERI,LLNL-JSO, GSN). We used approximately 2300 waveforms from ~200 events recorded at 75 stations in the region. Our tomographic model of Lg Qo is consistent with previous more qualitative Lg attenuation models that showed inefficient or blocked Lg across the Eurasian-Arabian plate boundary. We have also found, similar to the previous models, efficient Lg propagation throughout much of the Arabian plate. A substantial variation in Lg Qo values was also found for the Arabian peninsula (~250-800). The northern Arabian platform is generally had low to normal Lg Qo values (~250 to 350). In addition, we observe high Qo (~670-800) values for the southern Arabian Plate. Additionally, we have found a dramatic decrease in Lg Qo across the Bitlis suture; the plate boundary between the Arabian and Eurasian plates. Beneath the Turkish Plateau, our high to modestly high Lg attenuation values (Qo ~100 to 200) probably originate from both scattering and intrinsic attenuation due to the tectonic complexity and the wide spread young volcanics in the region. However, we think that the lowest Qo values for the East Anatolian plateau (~70 to 100) and the portion of western Turkey around the Menderes Massif (~60 to150) are most probably caused by intrinsic attenuation. Beneath the part of the Taurus Mountains in western Anatolia, normal Lg Qo values (~200-300) have been found. These relatively higher values may be related to the nature of the crust in the Tauride mountain belt that has created a stable continental crustal waveguide for east-west ray paths. For northeastern Turkey, the Caucasus and Azerbaijan, we also found some low to normal Lg Qo values (~170-180). We have also observed a fairly consistent negative correlation between the frequency dependence of Lg Q and the magnitude of Lg Qo in the northern Middle East.
S51A-0977
Attenuation Properties of Regional Phases from Peaceful Nuclear Explosions in Northern Eurasia
In this study, we analyze the data from Peaceful Nuclear Explosion (PNE) profiles in Russia for amplitude ratios and coda properties, with an objective to provide a detailed characterization of regional phase propagation in northern Eurasia. Logarithms of Lg/Sn and Lg/Pcoda amplitude ratios at frequencies of 0.5-3 Hz show a consistent decay with the source-receiver distance, with regionally-dependent decay rates. These decay rates appear to correlate best with the variations of two crustal parameters: 1) Moho depth gradient measured away from the source and 2) the travel-time within the sediments (product of the slowness and thickness of the sedimentary cover). The variation of the crustal thickness is the more significant factor affecting Lg propagation and is interpreted as causing defocusing of Lg energy. Moreover, crustal thinning affects Lg propagation stronger than thickening, apparently causing the energy to leak into the mantle or become backscattered). The correlation of the Lg decay rates with the travel times within the sedimentary layers could be explained by seismic attenuation within the sediments. Another striking observation from comparative analysis of the PNEs is the difference of Lg coda decay characters across the study area. Within the East European Platform and south-west West Siberian Basin, Lg coda amplitude decays exhibit a relatively low frequency dependence of about power 0.13 of the frequency. By contrast, within the Siberian Craton, the coda exhibits a nearly constant decay rate for all frequencies, corresponding to the Q vs. frequency exponent of 0.8-1.0. However, such a strong frequency dependence of Q could actually be due to frequency-independent attenuation caused by geometric spreading and leakage of the seismic energy from the crust. Using this model, the geometric spreading parameter was found to be constant for the entire region, and crustal quality Q=470 west of the Urals and an infinite Q within the Sibarian Craton. Such low attenuation within the Siberian Craton is also supported by the observed Pg propagating to the distances of 1600-1700 km in this area.
S51A-0978
Dense Seismic Recordings at Hengill Geothermal
A temporary array composed of 21 seismic stations was deployed in the Hengill geothermal field in SW Iceland during the months of July and August 2005. The Hengill experiment was designed to closely monitor seismic activity related to cold water injection in two nearby wells (HE-8 and HE-16) within the geothermal area. Nineteen of the seismic stations installed were L-28 4.5-Hz sensors while the remaining two were Guralp CMG-40T broadband seismometers. The seismic array recorded an average of approximately 7 events (microearthquakes) per day at a sampling rate of 500 samples per second. The high sampling rate is critical for shear-wave splitting analysis, which requires accurate measurements of fast shear-wave polarization directions and time delays between the arrivals of the fast and the slow shear-wave. Based on the high-resolution seismic data produced, it is possible to track temporal as well as raypath-dependent variations in polarizations and time delays. Such changes are used to model the response of the geothermal reservoir (in terms of fracture geometry and fracture density) to cold water injection activities. More specifically, we are interested in studying the response of subsurface cracks to fluid pressure changes as injection is started, halted, and then resumed again.
S51A-0979
Power-law decay characteristic of coda envelopes revealed from the analysis of regional earthquakes
So far, coda envelopes for wide frequency bands with long lapse time range have been described by bending curves which have several systematic changes in decay rate with lapse time. However, for NS component seismograms of regional earthquakes, applying the Hilbert Transform to make envelope curves and taking regression analysis on the envelopes in period bands from 24 s to 1/6 s for a wide lapse time range up to 4,000 s instead of measuring the peak amplitudes of the coda used previously, we found coda envelopes show the feature of ''power-law'' decay and there is no systematic change of decay gradient except for once occurring around {\it ScS} arrival with lapse time. 157 seismic events recorded by 11 IRIS broadband seismic network stations during the period from 1988 to 2005 with focal depths shallower than 50 km (shallow events) and deeper than 150 km (deep events) and with epicentral distance less than 1,000 km from individual seismic station were collected in this study. The moment magnitude of selected events ranges from 4.7 to 7.8. The ''power-law'' decay characteristic of seismic envelopes indicates that the power spectrum of seismic coda could be simply expressed as of the form {\it P}({\it T}, {\it t}) ~ {\it t} $^{-{\it α}}$, where {\it t} is lapse time and {\it T} is central period. We find that {\it α} = 1.6-6.2 for before {\it ScS} arrival in all period bands and 0.6-5.3 for after {\it ScS} arrival ranging from 24 s to 3/2 s periods. The coda decay gradient at short periods is steeper than that at longer periods for both cases of before and after {\it ScS} arrival. In particular, a clear offset of coda amplitude associated with {\it ScS} arrival appears in Turkey, Kyrgyzstan and Bolivia at 4-8 s and 8-16 s period bands. These coda decay gradient change and offset behavior in coda envelopes around {\it ScS} arrival are not clear in seismograms of shallow events. The simple and distinct characteristics of seismic coda envelopes could provide reliable information to determine physical values such as seismic source parameters and to identify the regional difference of medium heterogeneity in the deep Earth.
S51A-0980
Thermal Implications for Regional Wave Propagation
The new heat flow map of North America is used to compute a complete Moho temperature map for the entire US to compare to the Pn velocity variations. Thermal conductivity models and an exponential radioactive heat generation model allow the computation of the entire 3D temperature field. To a first order the high temperature areas at the top of Moho correlate to low Pn velocity, while the low temperatures correlate with high Pn velocity. For example, the high Pn velocities in southern Oklahoma are associated with low Moho temperatures, primarily due to the low surface heat flow. Due to the large crustal thickness, up to 50 km, and high surface heat flow values, modeled Moho temperatures in the Southern Rocky Mountains are anomalously high, well above the melting temperatures for typical lower crust composition. The Moho temperatures in the Basin and Range are lower than those in Southern Rocky Mountains due to smaller crustal thicknesses, only 30 km, but they approach melting temperatures for lower crustal compositions. These results suggest that partial melting could be responsible for the abnormal low Pn velocities in Southern Rocky Mountains and Basin and Range. In addition the high Moho temperatures inferred for the Southern Rocky Mountains suggest a simple conductive model might not be adequate for this area.
S51A-0981
Spatial Distribution of Medium Heterogeneity Beneath Eurasian Continent and Western Pacific as Revealed From Teleseismic P-coda
We systematically characterize the heterogeneous structure of the lithosphere by analyzing transverse amplitude of teleseismic P-coda, since the transverse components of P-wave are generated from lateral heterogeneity of the structure. We use 1619 bandpass filtered seismograms (0.5-1, 1-2 and 2-4 Hz) of deep and shallow events recorded at IRIS GSN stations. For each event-station pair, we calculate the energy partition of P-coda waves into the transverse component, and then we take its root-mean-square amplitude that represents normalized transverse amplitude. We further average the transverse amplitudes from deep and shallow events, respectively, at each station. We call this average transverse amplitude. The results show a clear positive correlation between the average transverse amplitude from deep events and those from shallow ones for all of the stations where both events are observed. This result strongly suggests that the average transverse amplitude estimated from shallow events can be used to characterize the heterogeneity of the lithosphere beneath the stations. That is, we are able to characterize the medium heterogeneity in wide area because shallow events are well recorded at many stations around the world. The observed average transverse amplitudes of shallow events show significant regional difference of heterogeneity that relate to tectonic conditions. Most of large average transverse amplitudes are observed at stations located on active tectonic regions along the collision zones in the southern Eurasia and the subduction zones in the western Pacific. A polar station located on the north Atlantic ridge-transform system also indicates large average transverse amplitude. The mid Eurasia and Australia, which are tectonically quiet regions on stable continents, show small average transverse amplitude. We further compare the average transverse amplitudes with the number of earthquakes within 300 km distance from each station for the past 30 years. The results show that seismically active regions are always characterized by large amplitudes. However large average transverse amplitudes do not always indicate high seismicity. This implies that the average transverse amplitudes detect medium heterogeneity that does not relate with current seismicity.
S51A-0982
Seismic imaging of a subducted ridge, southern Chile
Recent thermal modeling suggests that subduction of an active spreading ridge or a young oceanic plate can cause extensive magmatism, generating granitic melts and metamorphism which may lead to a significant volume of new continental crustal material. We report on a seismic experiment designed to constrain a number of input parameters to this thermal modeling such as the geometry of the subducting plate and the fine scale seismic velocity and attenuation structure. The simple plate geometry of the Chile triple junction makes it an ideal place to study the effects of ridge subduction. We have installed a network of 60 broadband seismometers in southern Chile just east of the triple junction to study the seismicity, the plate geometry and the velocity and attenuation structure of this region. The seismographs are installed from La Tapera (~44.6°S) in the north to Villa O'Higgins (~48.3°S) in the south. Here we report preliminary results from this seismic network. Local events in this region fall into three categories - those associated with the Liquie-Ofqui fault, those associated with the ridge subduction and a concentration of events beneath Volcano Hudson. The joint inversion of short period surface wave dispersion and receiver function data show the existence of a pronounced crustal low velocity zone beneath western Lago General Carrera region which coincides with the projected location of the most recent subducted ridge segment.
S51A-0983
Seismic Tomography of Central S\^{a}o Miguel, Azores Islands(Portugal)
The Azores Archipelago consists of nine volcanic islands, located at about $38° N and $28°W, in the triple junction of the American, Eurasian and African plates. The largest island is S\^{a}o Miguel. It has rift zones mainly trending NW-SE and E-W; calderas, as Fogo and Furnas, at the intersection of these tectonic lineaments; and thermal springs and fumaroles distributed along these fault systems. Furnas, Sete Cidades and Fogo are the most active volcanic complexes: central volcanoes with a dominantly trachytic production. Furnas is the youngest and consists of a steep-sided caldera structure formed during several collapses. The most important thermal features lie on an E-W lineament which cuts the Furnas caldera complex. The Fogo volcanic edifice is built over an older submarine lava basement and composed by lava flows, domes and pyroclastic flows deposits, with the summit truncated by a caldera.Thermal manifestations are associated with a NW-SE fault system and consist mainly of fumarolic activity. S\^{a}o Miguel was selected as a site for a seismic experiment in a European Union-sponsored project with the aim of quantifying the seismicity of various quiescent volcanoes in inhabited areas. The 3D distribution of P- and S-wave velocities is derived for central S\^{a}o Miguel, by traveltime tomography. We use P- and S-wave arrival times of 289 local earthquakes by a network of 23 seismometers. The model has good resolution in the shallowest 5 km. There are several Vp anomalies, referred to a composite picture of geologic deposits, volcanic structures and tectonic features. Furnas caldera has a shallow, low Vp value probably marking volcaniclastic sediments. A negative Vp anomaly is associated with the geothermal field of Ribeira Grande. Another low Vp area is related to the highly fractured NW-SE tectonic lineament connecting two geothermal areas in central S\^{a}o Miguel. Conversely, high velocity zones mark a central seismogenetic zone at 4-5 km of depth and the Altiprado region. The S\^{a}o Bras high Vp is instead interpreted in terms of high-density deposits. These interpretations are supported by distribution of Vp/Vs in the area.
S51A-0984
Variation of fundamental mode Rayleigh wave group velocity dispersion in Iran and the surrounding region
We present group velocity dispersion results from a study of regional fundamental mode Rayleigh waves propagating across Iran and the surrounding region. Data for these measurements come from field deployments within Iran by the University of Cambridge (UK) and the Universite Joseph-Fourier (FRA) in conjunction with International Institute of Earthquake Engineering and Seismology (Iran), within Oman by the Universite Pierre et Marie Curie-Paris (FRA), in addition to data from IRIS and Geofone. 1D path-averaged dispersion measurements have been made for ~800 source-receiver paths using multiple filter analysis. We combine these observations in a tomographic inversion to produce group velocity images between 15 and 60~s period. Because of the dense path coverage, these images have substantially higher lateral resolution for this region than is currently available from global and regional group velocity studies. We observe variations in short-period group velocity which is consistent with the surface geology. Low group velocity (2.45-2.55~km/s) at short periods (15-20~s) is observed beneath the south Caspian Basin, northern Iran, the Persian Gulf, the Zagros, the Makran, northern Afghanistan and southern Pakistan. Somewhat higher group velocity (2.60-2.70~km/s) at these periods occurs in central Iran. At intermediate periods (30-40~s) group velocities over most of the region are low (2.90-3.10~km/s) compared to Arabia. At longer periods (50-60~s) group velocities remain low (3.35-3.45~km/s) over most of Iran but there is a suggestion of higher group velocities beneath the northern and central Zagros.
S51A-0985
Three-Dimensional Crustal Structure in Southeastern Sicily (Italy) by Applying the Adaptive Mesh Seismic Tomography Method
Absolute and differential P- and S- wave data were used to invert Vp, Vp/Vs on a 3D grid of nodes and event locations in southeastern Sicily (Italy). The events selected have magnitude ranging between 1.0 and 4.2 and were recorded between 1994 and 2003 by a local network, consisting of nine digital 3-component stations. The adaptive mesh seismic tomography method (Zhang and Thurber, 2005) was applied by combining 3,474 absolute travel times and ~11,250 catalog-derived differential times of about 360 earthquakes. Unlike the most common tomography methods which use regular three-dimensional grid approaches, the adaptive mesh method is able to adapt the mesh nodes according to the changes of event locations and velocity structure. This capability is particularly useful when poor starting event locations and velocity model are available, since in these cases the ray distribution may change greatly during the inversion. The quality of the 3D model resulting from the tomographic inversion was evaluated on the basis of some synthetic tests. Moreover, comparison of the tomographic results with the tectonic map of the region has allowed satisfactory correlation of velocity anomalies with main tectonic faults. In particular, low velocity zones are found in the Ionian coast (inland and offshore), where several structures with NW-SE and NNE-SSW direction are present.
S51A-0986
Crustal Thickness and Poisson's Ratio of Continental Crust
An important component to the understanding the evolution of the continental lithosphere is to improve our knowledge on lower continental composition. Contribution towards this goal, we perform receiver function analysis using teleseismic waveforms recorded at permanent and temporary broadband seismic stations located in Middle East, Europe, Asia, and north Africa. Two hundred and twenty six stations recording a total of about 6,000 teleseicmic events producing more than 100,000 seismograms have been investigated. The distribution includes 72 stations in the Middle East, 57 stations in Europe, 60 stations in Asia, and 37 stations in central and north Africa. We have examined receiver functions for 213 of stations (best data) in the period of 1990-2004 and applied the receiver function stacking procedure of Zhu and Kanamori [2000] to estimate Poisson's ratio and crustal thickness. We have divided the research area according to five tectonics environments, explicitly Shields, Platform, Paleozoic orogenic belts, Mesozoic-Cenozoic orogenic belts, and rift zones based on Condie's [1989] simplified classifications. The results from this study shows lower value of Poisson's ratio σ=0.25 for Shield and Platform compare to the Orogenic-belts with σ=0.27. Crustal thickness for Shield and Platform show the value of 38km and 43km respectively, while for the Orogenic belts we found a value of 37km for Paleozoic belts and 39km for Mesozoic-Cenozoic belts, although the range of thicknesses for the younger active regions is large. Since our ultimate goal is to provide an improved imaged of global continental structure and composition, we combine our observations with receiver functions results from other published analysis. In total we have integrated observations from 374 stations located in different geologic setting and the results indicate the value of σ=0.26 for Poisson's ratio and H=38km for crustal thickness in Shield, σ=0.27 with H=43km for the Platform, and σ=0.28 with H=36-39km for the Orogenic belts. We will compare crustal thickness and Poisson's ratio estimates for the crust beneath the stations with the recent global studies.
S51A-0987
Shear-Wave Velocity Structure of the New Madrid Seismic Zone From Rayleigh Wave Group Velocity Dispersion Using Regional and Local Earthquake Data
Regional and local Rayeligh-wave group velocity dispersion data were used to extract shear-wave velocity structure of several propagation paths from earthquakes into the New Madrid Seismic Zone. The dispersion data are limited at high frequency by the mid-crustal depth of the earthquakes and at low frequency because of the relatively short propagation paths involved. Inferred dispersion curves also show irregularities possibly due to velocity heterogeneity along the path. The working passband occurs between frequencies of 0.05 and 0.5Hz. The dispersion data were inverted for shear-wave velocity structures. Our starting model for the inversion was a crustal model obtained from a seismic refraction experiment performed in the Eastern Tennessee Seismic zone. Preliminary results indicate a high shear-wave velocity layer at a depth of about 12-15km within the crust. The high velocity layer is seen to occur at a shallower depth compared to the result of the refraction data anlaysis of Catchings (1999) and the resolution is relatively poor for deeper parts of the crustal model.
S51A-0988
Crustal Structure Beneath the Kanto-Koshinetsu Region, Central Japan Inferred from Receiver Functions
We applied the receiver function technique using multiple-taper method to estimate crust, uppermost mantle, and the subducting plate structure in the Kanto-Koshinetsu region, central Japan. This region is characterized by many faults and tectonic lines, and moreover the Philippine Sea plate subducts northward and the Pacific plate subducts westward. We analyzed P-wave seismograms recorded by regional seismographic networks from August 2002, a dense seismographic array which consists of 30 stations in the Boso Peninsula from February 2004, and temporary seismographic stations which installed in the source region of the 2004 Mid Niigata prefecture earthquake (M6.8) for a month. This study mainly used teleseismic events in the range of epicenter distances between 30 to 90 degrees. Moreover, we used local intermediate and deep events. They have almost the same characteristics as teleseismic traces and we can apply in the higher frequency, although they can be analyzed only in the short time width. A thick sediment layer with low velocity near surface may be preventing to resolve deep structure. To investigate the effect of sediment layer beneath stations, we make the synthetic receiver function traces assumed four layers structures with a sediment layer and compared them with observed trace. Estimated average S-wave velocities from surface to 5 km depth indicate very low velocity structures in the Mid-Niigata area and the Boso-peninsula and in some stations locally. Next, we estimated the structure beneath each station by forward modeling and investigated the depth of the Moho discontinuity. The Moho depth is the range of about 25 to 40 km, and increases in the mountain district. This result is consistent to the other previous studies. They also show a tendency to change around the tectonic lines. On the other hand, depth change of the southern part in this analysis region may be corresponding to the oceanic Moho discontinuity of the northward subducting Philippine Sea plate. However, receiver function traces in many stations have variety azimuthally and estimated structure in some stations was unstable for high noise level, the effect of a thick sediment layer or heterogeneous structure. We will have to investigate these in detail in consideration of the tangential component.
S51A-0989
Three-Dimensional Velocity Structure and Precise Hypocenters in the Nikkou-Ashio Area, Central Japan Revealed by the DD-Tomography Method
The Nikko-Ashio area, Central Japan, is one of the most seismically active regions in Japan._@It locates on the volcanic front and a couple of volcanoes and faults exist. We detect a large amount of earthquakes about 8,000 a year by routine observation. The specific characteristics of the activities are as follows: 1) Shallow earthquakes occur within 3-10 km depth. 2) Those earthquakes separate into clusters. 3) Remarkable reflectors are detected about 15 km depth. 4) Deep low frequency earthquakes occur beneath that area. In this report we investigate the relationship between those seismic activities and velocity structure. We have clarified three-dimensional velocity structure in the Nikkou-Ashio area in January, 2002-March, 2005 from DD-tomography method applied to 76,580 absolute times and 656,369 double differences by routine observation at 57 stations in and around the Nikkou-Ashio area. Applying this velocity model to all arrival time data in January, 1991-June, 2005, we have also obtained precise hypocenters from DD-tomography method. The results indicate that a clear velocity image from the top of the crust to the upper-most mantle and the connection between the velocity structure and hypocenters. 1) Micro-earthquakes occur in the narrow range of 5-10 km depth. In that seismogenic zone, P-velocity is relatively high ( 5.8-6 km/sec ) inserted into two low velocity zones, the upper-most crust and the middle crust. 2) Low velocity dominates in 20-30 km depth. 3) About 30 km depth, in which deep low frequency earthquakes rise, S-velocity is extremely low and the ratio of Vp/Vs is high.
S51A-0990
Three Dimensional P and S Wave Velocity Structure and Vp/Vs Ratios for the Eastern Tennessee Seismic Zone
Three dimensional P and S wave velocity models have been constructed for the active eastern Tennessee seismic zone (ETSZ) using travel time tomography. The inversion utilized 5758 P wave and 4212 S wave arrival times, respectively, from 473 earthquakes recorded over the time period 1983 to 1994. Block size was 12 by 12 km horizontally by 4 km vertically. Model resolution was examined using chessboard and image reconstruction tests. Ray path coverage is similar in both models and both models contain similar velocity anomalies. The primary difference between the models is in the magnitude of the anomalies; P wave velocity anomalies are in the range +5 to -5% while S wave velocity anomalies are in the range +3.5 to -3.5%. The most significant velocity anomaly imaged is a region of low velocity that trends parallel to the seismicity and lies just to the SE of the prominent NY-AL magnetic lineament. The NY-AL magnetic lineament is associated with Grenville basement rock and coincides with the NW boundary of the ETSZ. Vp/Vs values for the low velocity region are either normal (1.72) or low, reaching 1.68. The imaged velocities and Vp/Vs ratios can be attributed to rock lithologies such as gneiss and do not require the region to be extensively fractured. Anomalously high P and S wave velocities are associated with a prominent mafic intrusion in eastern Kentucky and with portions of the crust in easternmost TN. The regions of high P and S wave velocities are also associated with high Vp/Vs ratios, suggesting the presence of mafic intrusions. The initial data set used for the inversion will be increased by approximately 400 earthquakes recorded from 1994 to 2002. The additional data will allow reduced block size and improved model resolution.
S51A-0991
Finite Frequency Tomography of the Patras Area (Gulf of Corinth,Greece)
The Gulf of Corinth is the most active rifting zone of the Mediterranean region. Despite many studies, the crustal structure of this region is still a debating subject. In the western part of the Gulf where we focus our study, previous tomographic studies point out a complex crustal structure and a low-dip surface that may accommodate the deformation. In this study, we include the sensitivity kernels in a 3D tomographic method. The influence of the frequency content will be analyzed when considering variable velocity structure. An accurate reconstruction of a 3D high-resolved images of the mid-crust beneath the Gulf of Corinth will improve our understanding of the rifting process in this area. In order to do so, we perform accurate estimation of both rays and travel-times between source/receiver and diffraction points using graph theory and an additional bending. Then, we estimate the amplitude using paraxial theory on these rays, amplitude estimation which are required in the sensitivity kernel. We invert both the velocity and the hypocenter parameters, using these so-called banana-doughnut kernels. Our dataset comprises of 451 local events with 9236 P- and 7523 S- first-arrival times. The events were recorded in the western part of the Gulf (Aigion area) in the framework of the 3F-Corinth project. We first analyze the impact of the crustal heterogeneities on the sensitivity kernels. Then, we compare our images with previous tomographic results in order to illustrate what brings inside our images the finite-frequency effects compared to features introduced by the high frequency limit which is the ray theory.
S51A-0992
Propagation of S-waves Through the Sediments in the Mississippi Embayment
S body waves from microearthquakes in the New Madrid Seismic Zone (NMSZ) are investigated at selected broadband station sites to understand wave propagation through the Mississippi embayment sediments. Earthquake body waveforms display distinctive features that constrain the nature of the body wave local site response and wave propagation within the unconsolidated Mississippi embayment sediments. S-wave resonance effects may infer near-site conditions. Site resonance effects change between individual receivers because of velocity heterogeneity. Travel times of observed S-phases such as S, Sp, and SsShs (the first S-wave reverberation) can be used to estimate the average S-wave slowness and Poisson's ratio within the embayment sediments. An average Poisson's ratio in the range of 0.34 to 0.45 is obtained for selected sites within the central NMSZ. Use of well log data in wave calculations shows that 1-D heterogeneity can be the first-order influence on seismic wave propagation within the Mississippi embayment sediments.
S51A-0993
Assessment of Finite-Frequency Theory for Regional Surface Wave Tomography
To map small scale heterogeneties more reliably, many recent works propose to go beyond ray theory using more accurate finite-frequency propagation theories. However, while any finite-frequency theory is better than ray theory in forward modelling, it is not obvious that a finite-frequency theory will lead to a better image through an inversion process. We compare here ray theory and finite-frequency theory for surface wave tomography both with synthetic data and with a regional real dataset. Our finite-frequency theory only considers near forward, single scattering without mode coupling. Moving from ray theory to finite-frequency theory also requires to reconsider the regularization of the inversion. For a better imaging of small scale structures, finite-frequency theory should ideally be applied without a priori lateral smoothing. This works well with synthetic noiseless data, but does not lead to satisfactory results when applied to real data, showing that finite-frequency inversions are still ill-posed, likely because of the noise in the data. In the real data case, a priori lateral smoothing is thus necessary which implies no improvement in imaging. This suggests that refining the theory only is not sufficient to significantly improve tomography.
S51A-0994
Development of a Regional Velocity Model Using 3D Broadband Waveform Sensitivity
We are developing a new approach which relies on a cascade of increasingly accurate theoretical approximations for computation of the seismic wavefield to develop a model of regional seismic velocity structure for eastern Eurasia using full seismic waveforms. The selected area is particularly suitable for the purpose of this experiment, as it is highly heterogeneous, presenting a challenge for standard modeling techniques, but it is well surrounded by earthquake sources and a significant number of high quality broadband digital stations exist, for which data are readily accessible through IRIS (Incorporated Research Institutions for Seismology) and the FDSN (Federation of Digital Seismic Networks). The initial model is derived from a large database of teleseismic long period waveforms (surface waves and overtone wavepackets) using well-developed theoretical approximations, the Path Average Approximation (PAVA) and Nonlinear Asymptotic Coupling Theory (NACT). These approaches assume waveforms are only sensitive to the 1D (PAVA) and 2D (NACT) structure in the vertical plane between source and receiver, which is adequate for the development of a smooth initial 3D velocity model. We refine this model using a more accurate theoretical approach. We utilize an implementation of a 3D Born approximation, which takes into account the contribution to the waveform from single scattering throughout the model, giving full 3D waveform sensitivity kernels. We perform verification tests of this approach for synthetic models, and show that it can accurately represent the wavefield as predicted by numerical approaches in several situations where approximations such as PAVA and NACT are insufficient. The Born 3D waveform sensitivity kernels are used to perform a higher resolution inversion of regional waveforms for a smaller subregion between longitudes 90 and 150 degrees E, and latitudes 15 and 40 degrees N. To further increase the accuracy of this model, we intend to utilize a very accurate numerical approach, the coupled spectral elements method (Capdeville et al., 2003), for the forward calculations of synthetics. This accounts for nonlinear effects of the structure on the seismograms in addition to the 3D broadband sensitivity.
S51A-0995
Constraints on the Interpretation of S-to-P Receiver Functions
We present results from a forward modeling study to asses the feasibility of using S-to-P converted phases to image the seismic discontinuity structure of the crust and upper mantle. We show that a significant level of interfering P-wave energy arriving before the direct S-wave arrival can overprint the S-to-P converted phases of interest produced by the direct S-wave arrival and may result in Sp receiver function phases that do not represent true Earth structure. The source of this P-wave energy is attributable to a number of phases, including those that have undergone multiple reflections off the Earth's surface. For deep focus earthquakes (300-600km deep), a significant amount of P-wave energy is observed from pPPP, pPPPP, and sPPPP phases, and arrives within the same time window as predicted for S-to-P converted phases from the direct S phase arrival. Furthermore, for earthquakes at all depths, interfering P-wave energy arrives within the same time window as predicted for S-to-P converted phases from the SKS phase arrival, limiting the usefulness of SKSp receiver functions for upper mantle imaging. To isolate true Sp receiver function phases from contamination due to other P-wave phases, we find it necessary to: 1) restrict the depth of earthquakes that we use to less than 300 km, and 2) restrict the earthquake epicentral distance range that we use to between 60-75 degrees and ensure that the observed phases have consistent moveout characteristics.
S51A-0996
Rayleigh and Love Wave Tomographic Study of the Upper Mantle Beneath the Reykjanes Ridge
General insight into the existence and role of mantle plumes in mantle dynamics may be obtained by studying how melting and flow patterns beneath mid-ocean ridges are perturbed near hotspots. The Iceland-Reykjanes Ridge system is a well-studied example of this interaction. Extending southward from Iceland, the Reykjanes Ridge exhibits an anomalous character for its slow spreading rate suggesting southward spreading of a plume material beneath the ridge. One end-member model for this flow suggests that low-viscosity plume material is narrowly channeled beneath the ridge axis; an alternative model suggests that it spreads radially outward beneath the lithosphere. We analyzed Love and Rayleigh waves that were generated by earthquakes located to the south of Iceland and that were recorded by 46 broadband seismometers located across Iceland. Over 12000 measurements of the phase, group travel time, and amplitude of narrow-pass filtered waveforms (10-50s Love; 14-100s Rayleigh) were included in an inversion for mantle and crustal shear anisotropic velocity structure. Amplitudes of the waveforms are sensitive to the velocity structure across the ridge and are largest at stations located near the ridge axis. This is due to lateral refraction of surface wave energy into a wide low-velocity region beneath the ridge axis that acts as a lateral wave-guide, trapping the surface wave energy. The waveforms are sensitive to the magnitude and width of the low-velocity region and our surface wave inversion technique exploits this sensitivity. Tomographic images show a wide low-velocity region beneath the lithosphere (approx. 600 km across and centered on the ridge) that is consistent with high temperatures (<200\?) and a small amount of melt (<0.5%). Our results indicate that plume material spreads relatively broadly outwards to the south of Iceland, and is not restricted to a narrow, sub-ridge, channel. Our results also indicate that the spreading plume mixes with the greater part of the upper mantle, to at least 200 km depth.