T23F-01
A Eurasian and North African Crust and Upper Mantle Model for Regional Seismic Location
We are developing a velocity model of the crust and upper mantle across Eurasia and North Africa to reduce event location error by improving regional travel-time prediction accuracy. The model includes both P and S velocities, enabling prediction of Pn, Pg, Sn, and Lg phases. Here, we report on the Pn travel-time prediction effort. The model parameterization is a global tessellation, which allows extensibility beyond the current focus area, with node spacing of approximately 1°. The model is defined on a WGS84 ellipsoid that includes topography, thus eliminating the need for conventional corrections to base model travel-time calculations. A velocity profile with depth is defined at each node that includes a crustal stack, mantle velocity at the Moho (Pn velocity), and a linear mantle gradient term. Parameterizing mantle velocity structure with a linear term allows computation of Pn and Sn travel times in approximately one millisecond, which is needed for real-time seismic location. Velocity and mantle gradient at arbitrary points can be calculated by interpolating values at nodes. The starting model is a geophysical compilation that constrains Moho depth and depths of crustal layers, as well as best estimates for crust and mantle velocities. The starting model itself is demonstrated to improve travel-time prediction over iasp91, and we use a tomographic method to further refine Pn velocity, mantle gradient, and a node-specific crustal slowness. Our tomographic data set consists of approximately 3.5 million regional arrivals from events that meet criteria that ensure location accuracy. Non-seismic constraints on event location are used when available. Each datum is tested to meet strict quality control standards that include comparison with established distance-dependent travel-time residual populations relative to the iasp91 model. In addition to bulletin measurements, nearly 12,000 arrival measurements were made at the national laboratories. This work was performed under the auspices of the U.S. Department of Energy by University of California Lawrence Livermore National Laboratory under contract No. W-7405-Eng-48.
T23F-02
Structure Of The Elevated Precambrian Terranes Rising Above The Brahmaputra Plains In Northeastern India.
We present new evidence for a thinner crust beneath most of the Shillong plateau as well as its northeast extension in Mikir Hills of northeastern India.Both these Precambrian terranes rise above the Brahmaputra plains whose crust is thicker in comparison by atleast 4~km. Although Bouger gravity over the Mikir Hills still remains to be determined, its near zero value over the ~1 km high plateau and the near normal upper mantle beneath the region, require that these elevated terranes must have been uplifted between reversed faults and continue to be supported by them under compression. The southern edge of the Shillong plateau is indeed marked by the prominent Dauki fault which swerves northeastward at the south eastern margin of the plateau to merge with the Naga thrusts that bound the Mikir Hills on the east. A similar fault bounding the plateau on the north as hypothesized by Bilham et al (2000) -the Oldham fault- is therfore required to swerve northeastward near the northeastern margin of the plateau to demarcate the Mikir Hills from the thicker crust Brahmaputra plains to its north and west. This could be explained by a strike slip offset of the Oldham fault caused by the as yet obsure but active tectonics of the NNW trending Kopili lineament that ensues from the inflexion in the Dauki-Naga thrust fault system.
T23F-03
Seismic Anisotropy In India: Lithospheric vs Asthenospheric Origin
Due to the paucity of available seismological data in the public domain, the structure of the Indian lithosphere is little known. Seismic anisotropy underneath India and Sri Lanka is investigated here using shear wave splitting measurements on core refracted phases. Seismic anisotropy can provide insights into the lithospheric structure and the possible crust/upper mantle mechanical coupling. No such study has yet been done across the whole Indian sub-continent. As a result, some authors have extrapolated to the entire plate results obtained at very sparsely located stations and inferred the northern marker of the Indian plate underthrusting Eurasia, based on unconstrained data. We introduce a very comprehensive dataset recorded at 21 seismological stations deployed from the Himalaya to Sri Lanka. An unprecedented number of null results have been computed and we stress the too often neglected importance of this type of results. The 90° ambiguity inherent to the definition of a null prevents any interpretation without the addition of non-null results, but the added piece of information they carry is very valuable. The consistency in terms of the orientation of the two potential polarization planes of the fast S-wave (phi) is striking across the entire Indian plate. One possible orientation of phi closely follows the trend of the NUVEL1A Indian plate motion (NNE-SSW) with respect to a fixed Eurasian plate. By addition of a reasonable amount of non- null results, the NNE-SSW orientation of phi is confirmed under the majority of the stations spread from Sri Lanka to the northern part of the Dharwar craton. Towards the Himalaya and the Shillong plateau, the orientation of phi rotates clockwise to become EW. The transition zone between those two orientations might be located in the Gangetic plain, which is unfortunately not instrumented. The clockwise rotation in the orientation of phi observed between peninsular India and the Himalayan front seems to counterbalance the counterclockwise rotation of the Indian plate colliding with Eurasia. This response in terms of lithospheric deformation suggests a lithospheric dominated source of anisotropy consistent with the internal strain of the lithosphere due to the active deformation of the plate induced by its collision with Eurasia.
T23F-04 INVITED
Surface Wave Dispersion Measurements and Tomography from Ambient Seismic Noise in China
Recent laboratory and theoretical studies have shown that the Green functions of a structure can be obtained from the cross-correlation of diffuse wavefields. The idea has now found rapid applications in seismology, in particular, surface waves have been found to be most easily retrievable from the cross-correlations of seismic coda or ambient noise. Dispersion measurements made on the estimated Green functions present significant advantages over traditional measurements that are based on earthquakes with limited distribution. The method is particularly useful in surface-wave path calibration and tomographic mapping in aseismic regions and at short periods (below 20 s). We have recently performed ambient noise tomography of China using the new national China Seismic Network and surrounding global and regional stations. For most station pairs, we can retrieve very good Rayleigh waves from ambient noise correlations using 12-months of continuous data at all distance ranges (100 km to 5000 km) and for periods down to about 8 s. The combination of stations achieves good and relatively uniform coverage for most parts of China. We obtain Rayleigh wave group velocity dispersion measurements (a total of about 1400 dispersion curves so far) using a frequency-time analysis method and construct Rayleigh wave group velocity maps for periods from 10 s to 60 s. Our preliminary dispersion maps show significant features that correlate with surface geology. Two major features stand out in particular. First, the major basins with thick sediment deposition correlate well with slow group velocities at shorter periods (10 to 20 s). The major basins, including Tarim, Junggar, Qadaim, Sichuan, Bohai, Songliao, Southern North China, and Jianghan, are all well delineated by slow velocities. Second, variations in crustal thickness correlate with group velocities for periods around 30 s. The major trend of crust thickening from the east to west is well represented by the velocity decreases from east to west.
T23F-05 INVITED
Lithospheric Structure, Crustal Kinematics, and Earthquakes in North China: An Integrated Study
The North China block (NCB) is geologically part of the Archaean Sino-Korean craton. But unusual for a craton, it was thermally rejuvenated since late Mesozoic, and experienced widespread extension and volcanism through much of the Cenozoic. Today, the NCB is characterized by strong internal deformation and seismicity, including the 1976 Tangshan earthquake that killed ~250,000 people. We have started a multidisciplinary study to image the lithospheric and upper mantle structure using seismological methods, to delineate crustal kinematics and deformation via studies of neotectonics and space geodesy, and to investigate the driving forces, the stress states and evolution, and seismicity using geodynamic modeling. Both seismic imaging and GPS results indicate that the Ordos plateau, which is the western part of the NCB and a relic of the Sino-Korean craton, has been encroached around its southern margins by mantle flow and thus is experiencing active cratonic destruction. Some of the mantle flow may be driven by the Indo-Asian collision, although the cause of the broad mantle upwelling responsible for the Mesozoic thinning of the NCB lithosphere remains uncertain. At present, crustal deformation in the NCB is largely driven by gravitational spreading of the expanding Tibetan Plateau. Internal deformation within the NCB is further facilitated by the particular tectonic boundary conditions around the NCB, and the large lateral contrasts of lithospheric strength and rheology. Based on the crustal kinematics and lithospheric structure, we have developed a preliminary geodynamic model for stress states and strain energy in the crust of the NCB. The predicted long-term strain energy distribution is comparable with the spatial pattern of seismic energy release in the past 2000 years. We are exploring the cause of the spatiotemporal occurrence of large earthquakes in the NCB, especially the apparent migration of seismicity from the Weihe-Shanxi grabens around the Ordos to the North China plain in the past 200 years.
T23F-06
Crustal and mantle velocity models of southern Tibet from finite frequency tomography
The formation of the Himalayan Range and Tibetan plateau has broad implications for earth sciences. Yet the exact mechanisms of the mountain-building processes, in particular, the roles of the Tibetan mantle lithosphere and upper mantle in the rise of the plateau, remain poorly constrained. Most of the previous seismic experiments in southern Tibet consisted of north-south trending linear arrays, thus provided very limited constraints on the east-west variations in the crustal and mantle structure beneath the region. As an addition to the international Hi-CLIMB project, a temporary, 2D array of 37 seismic stations was installed between Tingri and Xigaze in southern Tibet and operated from June 2004 to August 2005 by Peking University, China Academy of Geological Sciences, and Institute of Earthquake Sciences, Academia Sinica. We measured traveltimes of teleseismic body waves recorded by this array by waveform cross-correlation and carried out finite-frequency tomography inversions to image the 3-D velocity structure beneath southern Tibet. The preliminary results reveal strong lateral P-velocity variations (3-4%) at depth above 200 km, and a region of relatively high P-wave velocity extending continuously from the uppermost mantle to 350 km depth beneath southern Tibet. Interpretation of these findings and their implications for the current dynamics models of Tibetan plateau will be discussed.
T23F-07
Attenuation Tomography of Northern California and the Yellow Sea / Korean Peninsula from Coda-source Normalized and Direct Lg Amplitudes
We perform inversions for regional attenuation (1/Q) of Lg in two different regions. The path attenuation component of the Lg spectrum is isolated using the coda-source normalization method, which corrects the Lg spectral amplitude for the source using the stable, coda-derived source spectra. Tomographic images of Northern California agree well with one-dimensional (1-D) Lg Q estimated from five different methods. We note there is some tendency for tomographic smoothing to increase Q relative to targeted 1-D methods. For example in the San Francisco Bay Area, which contains high attenuation relative to the rest of it's region, Q is over- estimated by approximately 30. The over-estimation effect is investigated via the trade-off between resolution and error. We compare the coda-source normalized inversions with amplitude ratio tomography. Preliminary results in Northern California show that coda-source normalized amplitude tomography may produce better-resolved attenuation structure. We extend the analysis to the Yellow Sea / Korean Peninsula and compare with 1-D techniques.
T23F-08
Computational Improvements for Deriving a 3-D Shear-Wave Model for the Tarim Basin, China, Using Multiple Geophysical Observations
We derive a 3-dimensional Vs model for the Tarim Basin, China, using surface wave dispersion, teleseismic receiver functions, gravity and body-wave travel times. High-resolution Rayleigh wave slowness maps provide 23 group velocity dispersion curves in the period range between 8 and 100 s for a grid of locations across central Asia. Receiver functions for 38 stations and body wave travel times for 4315 seismic events are included. GRACE satellite gravity observations constrain simultaneous information from neighboring map cells and layers to properly account for density contributions to seismic velocity. The combined data sets produce a model fitting diverse geophysical parameters, and allow us to constrain the model in places where one parameter may be poorly represented; however, the diversity of parameters and size of the problem pose difficulties regarding stability and limited computing resources. We use an interative, LSQR conjugate gradient solver in lieu of singular value decomposition (SVD). LSQR is ideal for large, overdetermined sparse systems. Or optimized inversion requires 0.05% of the memory for SVD and runs ~1000 times faster, permitting multiple runs within a tractable time frame. Preliminary results agree with the SVD findings, indicating very low upper crust and high upper mantle shear velocities beneath the Tarim basin; these values exhibit variations between the eastern and western parts of the Tarim.