Upper Mantle Structure Beneath The Chinese Capital Region From Teleseismic Finite-frequency Tomography

Chinese capital region is located in the northern part of North China block, including three tectonic units: North China basin, Yanshan and Taihang uplift regions. This region has complex geological structure and Cenozoic tectonic activity. Many destructive earthquakes have occurred in this region, such as 1679 Sanhe-Pinggu earthquake (M8.0), 1976 Tangshan earthquake(M7.6) and 1998 Zhangbei earthquake(M6.0), which killed many people and caused great economic loss. Because of important geographical location, complex geological structure and stronger seismic activity, many geoscientists are very interested in the deep structure of this region. In this study, we applied the finite-frequency seismic tomography to teleseismic events recorded by the digital Capital Seismic Network(CSN) to determine 3-D P-wave velocity model of the upper mantle under this region.Finite-frequency tomography(FFT) is a new method to invert deep structure of the earth in the age of broadband seismic observation. This new method use 3-D integral of the sensitivity kernel around the ray path to represent travel-time delay instead of use 1-D integral of the slowness perturbation along the ray path to represent travel-time in ray theory, which is the essential difference between FFT theory and ray theory. The main characteristic of FFT is that the broadband seismic waveform data are divided into different frequency bands, which make the broadband signals of the observed seismic waves be fully utilized. The deeper crossing of 3-D sensitivity kernels in different bands enlarge the sample space, reduce the affect of uniformity data coverage, and improve the resolution of the velocity model.The range of the present study area is 113.5°～120.5°E and 37.5°～41.5°N. We selected more than 300 events from all teleseismic events which have epicentral distances between 30°to 90°from the center (39.5°N, 117.0°E) of the study region and magnitudes greater 5.0, and they occurred during a period from September 2003 to December 2005. All more than 300 teleseismic events used in this study have a fairly complete azimuthal and epicentral coverage. To account for frequency-dependent of seismic wave and frequency bands of different type of seismograph, we used band-pass-filtering to divide broadband wave into high-, intermediate-, and low-frequency bands (1.0-2.0, 0.1-1.0 and 0.05-0.1 hz, respectively), and measured P-wave relative travel-times between two nearby stations by multi-channel cross correlation in each frequency band for each event. As a result, we obtained 18499 high accuracy P-wave relative travel-times (6212, 7229 and 5058 for high, intermediate and low-frequency bands, respectively). The crustal and mantle volume beneath the study region is parameterized with 3-D grids of 37×35×31 centered at (39.5°N, 117.0°E), and the grid interval in all directions is about 35 km. To get the 3-D Fréchet sensitivity kernel, ray-tracing technique is employed to calculate central geometrical ray paths from source to receiver in IASP91 model accurately, then sensitivity kernels around the central geometrical ray paths were calculated by paraxial approximation for each frequency band. We established observation equation with these measured relative travel-times and 3-D Fréchet sensitivity kernels in different bands and then inverted this observation equation. To reduce the affects of heterogeneity in the crust to the inverted upper mantle velocity structure, we first corrected travel-time shifts due to the thickness of crust and station elevation before inversion. Then, a correction term at each station was incorporated into the inversion to absorb travel-time shifts caused by other shallow heterogeneity. The damped least squares and convolutional quelling are employed to restrict velocity model and the LSQR algorithm is used to solve the large parsesystem of observation equations, a 3-D P-wave velocity model of the upper mantle beneath the Chinese capital region was determined. In order to evaluate the effects of ray coverage and spatial resolution, we conducted the detailed resolution tests by using sampling density, checkerboard resolution test and restoring resolution test.The final velocity model was determined by using damped least squares method to restrict model and LSQR algorithm to invert relative travel-times which correction of crust and elevation were take into account and a correction term at each station was added. Our results show there are distinct differences of deep velocity structure down to 150 km depth under the three tectonic units. The Yanshan uplift exhibited the high velocity (high-V) feature which is consistent with stable and aseismic character. Under the Taihangshan uplift, broad low velocity (low-V) are visible, but it also shows up as small high-V anomalies. A large scale prominent low-V anomaly is obvious under the North China basin and Bohai bay where seismic activity is higher.A large scale prominent low-V anomaly was revealed in the shallow upper mantle under the North China basin and Bohai bay. In the North China basin the low-V anomaly generally extend from 50 km to 150 km depth, but in the Bohai bay, this low-V anomaly gradually extend down 200 km depth. The depth of this low-V anomaly is 50-70 km under the Bohai bay and North China basin, which is consistent with the depth of high conductivity layer in the upper mantle determined by the measurement of magnetotelluric sounding and heat flow. This result shows lithosphere thinning in the Bohai bay and North China basin.From our tomographic images, we can see that Zhangjiakou-Penglai fault zone is a variation belt of velocity structure in the upper mantle, also a boundary of lithosphere thinning. Most of large historical earthquakes occurred in this boundary belt. We consider that these large earthquakes maybe result from the heat stress caused by the uneven asthenosphere activity in the two sides of this fault zone. So the large earthquakes occurred in this region are not only related to crustal heterogeneity but also affected by the deeper velocity structure variation in the upper mantle.Relative to the travel-time seismic tomography based on ray theory, finite-frequency seismic tomography requires seismic waveform with high quality, the inversion results are fairly large affected by data sample. In those regions with dense sample and good resolution, the results from these two methods are basically coincident, but the velocity anomalies revealed by FFT are fragmentary.