PUBLICATIONS. Geochemistry, Geophysics, Geosystems

Transcription

1 PUBLICATIONS Geochemistry, Geophysics, Geosystems RESEARCH ARTICLE Key Points: First 3-D anisotropic model of lithosphere in northern part of North China First integration of ambient noise and earthquake tomography in North China Distinct lithospheric structure implies various extension from west to east Correspondence to: Y. V. Fu, Citation: Fu, Y. V., Y. Gao, A. Li, and Y. Shi (215), Lithospheric shear wave velocity and radial anisotropy beneath the northern part of North China from surface wave dispersion analysis, Geochem. Geophys. Geosyst., 16, , doi:1.12/ 215GC5825. Received 31 MAR 215 Accepted 17 JUN 215 Accepted article online 19 JUN 215 Published online 21 AUG 215 VC 215. American Geophysical Union. All Rights Reserved. Lithospheric shear wave velocity and radial anisotropy beneath the northern part of North China from surface wave dispersion analysis Yuanyuan V. Fu 1, Yuan Gao 1, Aibing Li 2, and Yutao Shi 1 1 Institute of Earthquake Science China Earthquake Administration, Beijing, China, 2 Department of Earth and Atmospheric Sciences, University of Houston, Houston, Texas, USA Abstract Rayleigh and Love wave phase velocities in the northern part of the North China are obtained from ambient noise tomography in the period range of 8 35 s and two plane wave earthquake tomography at periods of 2 91 s using data recorded at 222 broadband seismic stations from the temporary North China Seismic Array and permanent China Digital Seismic Array. The dispersion curves of Rayleigh and Love wave from 8 to 91 s are jointly inverted for the 3-D shear wave structure and radial anisotropy in the lithosphere to 14 km depth. Distinct seismic structures is observed from the Fenhe Graben and Taihang Mountain to the North China Basin. The North China Basin from the lower crust to the depth of 14 km is characterized by high-velocity anomaly, reflecting mafic intrusion and residual materials after the extraction of melt, and by strong radial anisotropy with Vsh > Vsv, implying horizontal layering of intrusion and alignment of minerals due to vigorous extensional deformation and subsequent thermal annealing. However, low-velocity anomaly and positive radial anisotropy are observed in the Fenhe Graben and Taihang Mountain, suggesting the presence of partial melt in the lithosphere due to the mantle upwelling and horizontal flow pull. 1. Introduction The North China, part of the Sino-Korean Craton, consists of the Ordos Plateau in the west and the North China Plain in the east (Figure 1). Since the Mesozoic, the North China has undergone significant lithospheric thinning and extension, producing volcanism, rifting, and large earthquakes [e.g., Fan and Menzies, 1992; Zheng et al., 1998; Liu et al., 24]. This tectonic reactivation has been a target for multidisciplinary scientific studies because the region offers a good opportunity to study the evolution and recycling of cratonic material. Despite decades of research, how this thinning occurred is still not well understood. The contribution of lithospheric delamination [Tian et al., 1992; Gao et al., 29], or thermal and chemical erosion [Xu, 21; Zheng and Wu, 29; Gao et al., 24; Menzies et al., 27], or mantle plume [Deng et al., 24] continues to be controversial. A firm understanding of the seismic structure of the crust and upper mantle is needed to confirm these or other suggested models. The lithospheric structure of the North China has been constrained by 2-D deep seismic soundings [e.g., Zhang, 1996; Jia and Zhang, 25] and regional 3-D seismic tomography derived from travel time analysis of body wave [e.g., Huang and Zhao, 26, 29; Tian et al., 29; Xu and Zhao, 29; Tian and Zhao, 211; Wang et al., 213] and dispersion measurement of surface wave [e.g. Friederich, 23; Huang et al., 23; Priestley et al., 26; An et al., 29; Huang et al., 29]. Although these seismic models provide some constraints on the formation and evolution of the North China, they are not at sufficiently high resolution to identify the complex structure, such as the morphology of the delaminated old lithosphere and the shape and origin of the low-velocity anomaly. In addition, almost all of previous tomographic studies make the isotropic assumption and do not consider the effect of anisotropy despite there is a significant trade-off between the lateral variation of isotropic velocity and anisotropy. This limitation can bias the inverted velocity structure when the strength of anisotropy is large. A more detailed and accurate image of lithospheric structure should resolve some ambiguity about important issues, such as the shape and origin of velocity anomaly. In this paper, we determine phase velocities of Rayleigh and Love wave in the northern part of North China by applying ambient noise tomography [Lin et al., 28] for the periods of 8 35 s and two-plane-wave tomography [Forsyth and Li, 25; Li and Li, 215] for the periods of s. A new 3-D model of shear FU ET AL. SHEAR VELOCITY AND RADIAL ANISOTROPY 2619

2 Geochemistry, Geophysics, Geosystems Figure 1. Map of topography and seismic stations in the northern part of North China. The broadband stations from the temporary North China Seismic Array and the permanent China Digital Seismic Array are indicated by red and yellow triangles, respectively. Black diamonds mark the locations of Tangshan (TS), Tianjin (TJ), Beijing (BJ), Zhangjiakou (ZJK), Datong (DT), and Yuncheng (YC). Blue large triangle shows the Datong volcano field from An and Su [28]. Green circles show the locations of earthquake with magnitude larger than 5 since 197 and the historic earthquake with magnitude larger than 7. NCB, North China Basin; TM, Taihang Mountains; FG, Fenhe Graben; LM, L uliang Mountains; YinM, Yin Mountains; YM, Yan Mountains. wave velocity and radial anisotropy in the crust and upper mantle is obtained by simultaneously inverting Rayleigh and Love wave dispersion curves. 2. Surface Wave Dispersion Analysis We use three-component seismic data (November 26 to December 27) from 222 broadband stations from the temporary North China Seismic Array and the permanent China Digital Seismic Array in the FU ET AL. SHEAR VELOCITY AND RADIAL ANISOTROPY 262

3 (a) Z-Z (b) T-T time (s) time (s) Figure 2. Cross-correlated waveforms filtered from 5 to 5 s between station A19 and other stations. (a) Record sections of vertical-vertical cross correlation. (b) Record sections of transverse-transverse cross correlation. Red lines indicate the approximate group arrival times of Rayleigh and Love wave, respectively. northern part of North China (Figure 1). We correct the misorientation of horizontal components of 1 permanent stations from China Digital Seismic Array using the method from Niu and Li [211]. Rayleigh and Love phase velocity dispersions at short periods (8 35 s) are measured from cross correlation of ambient noise. For long periods (>35 s), we apply the two-plane-wave method to teleseismic earthquakes to obtain the phase speed curves of Rayleigh and Love wave. The measurements obtained by ambient noise tomography are most sensitive to crustal structure, while information about the mantle is constrained by the teleseismic earthquake tomography. For ambient noise tomography, we follow the data processing and measurement techniques described in detail by Bensen et al. [27] and Lin et al. [28]. The empirical Green s functions (EGFs) of Rayleigh and Love wave between the stations are extracted from cross correlations of vertical component and transverse component, respectively. The arrival time of Love wave on transverse cross correlation is generally earlier than that of Rayleigh wave on the vertical cross correlation from the same station pair, illustrating that Love wave generally propagates faster than Rayleigh wave at the same frequency (Figure 2). For each period, the EGFs can be retained for the dispersion analysis only when they satisfy the far-field approximation (an interstation spacing of at least three wavelengths) and the high signal-to-noise ratio criterion (larger than 15 for our study) [Lin et al., 28]. The number of measurements decreases with period due to the far-field approximation. However, the raypath coverage is still good in the center of the study region even at longer periods. The phase velocity dispersion curves of Rayleigh and Love wave from 8 to 35 s are determined from the EGFs by an automated frequency time analysis [Levshin et al., 1989; Ritzwoller and Levshin, 1998; Levshin and Ritzwoller, 21]. These dispersion measurements are inverted for phase velocity maps on a spatial grid across the northern part of North China using the tomographic method of Barmin et al. [21]. We use fundamental mode Rayleigh and Love waves from earthquakes in a distance range of with body wave magnitude larger than 5.5. The distributions of 87 events for Rayleigh wave and 57 events for Love wave with high-quality data both show good azimuthal coverage (Figure 3). There are denser crossing Rayleigh and Love wave raypaths within the array, where the resolution is expected to be higher than in the surrounding area. Fundamental mode Rayleigh and Love wave trains are obtained at 11 frequencies, FU ET AL. SHEAR VELOCITY AND RADIAL ANISOTROPY 2621

4 Figure 3. Epicenters of earthquakes used as (left) Rayleigh and (right) Love wave sources in the epicentral distance range of 3 12 degree in this study. The triangle represents the center of the study region. ranging from 11 to 5 mhz. We first apply normalization for station instrument responses to vertical component for Rayleigh wave and transverse component for Love wave, and then filter them with a 1 mhz wide, fourth-order, double pass Butterworth filter centered at the frequency of interest. The Rayleigh and Love waves with good signal-to-noise ratio and coherence from station to station are retained and isolated with a rectangular window with cosine tapers at the two ends (Figure 4). For Love wave data, the selection is much more strict than that for Rayleigh wave since fundamental mode Love wave could be contaminated by the first and second higher modes especially for Love waves passing thoroughly oceanic plates [e.g., Nettles and Dziewonski, 211; Foster et al., 214] and multiple surface-reflected SH waves (SS, SSS, and SSSS). We assess the interference of higher modes and multiple S waves by visually checking the waveforms. No coherent waveforms indicate the existence of interference. Li and Li [215] discuss the interference and provide the criteria in selecting high-quality fundamental mode Love wave at a greater length. We basically follow their procedure to extract fundamental mode Love waves. The amplitude and phase of Rayleigh and Love wave are inverted for phase velocity by the two-plane-wave method, which uses the interference of two plane waves to represent the incoming wave field [Forsyth et al., 1998; Forsyth and Li, 25; Li and Li, 215]. For Rayleigh wave inversion, this interference is represented by the sum of the displacement of each plane wave on the vertical component. However, this summation cannot be applied to Love wave since the particle motion of each plane wave is different. Li and Li [215] provide a method to solve this problem. They decompose the amplitude on the transverse component of each plane wave to x and y components in a local Cartesian system for each earthquake and then add corresponding displacements on x or y component. The inversion procedure is the same as in Forsyth and Li [25] except that we add one component in the observed and predicted data. The final misfit is the sum for both x and y components. For each plane wave in both Rayleigh and Love wave inversion, a Gaussian sensitivity function is utilized to represent the finite width of the response of surface waves to structure along the raypath. We adopt a standard, iterative, linearized inversion technique [Tarantola and Valette, 1982] to solve for phase velocities on a grid of nodes and the incoming wavefields for each event simultaneously. There are a total of 156 grid nodes that are distributed unevenly in the region. The grid cell is in the center of the region and.758 by.758 along edges. To determine how well the inversion parameters can be retrieved, checkerboard resolution tests are performed to model with a cell size of for ambient noise tomography and for two-planewave tomography, respectively (Figure 5). The input model contains alternating positive and negative velocity perturbations of 64% with an average velocity of 4 km/s (Figures 5a and 5b). Synthetic Rayleigh FU ET AL. SHEAR VELOCITY AND RADIAL ANISOTROPY 2622

5 11mHZ 13mHZ 14mHZ 17mHZ 2mHZ 22mHZ 25mHZ 29mHZ 33mHZ 4mHZ 5mHZ time (s) 11mHZ 13mHZ 14mHZ 17mHZ 2mHZ 22mHZ 25mHZ 29mHZ 33mHZ 4mHZ 5mHZ time (s) Figure 4. Waveforms of (left) Rayleigh and (right) Love wave at station A19 from an earthquake that occurred in Japan on 13 January 27. The waveform on the top is a broadband wave train, and other waveforms are filtered with a 1 mhz width filter centered at frequencies from 11 to 5 mhz both for Rayleigh and Love wave, respectively. and Love wave phase velocities are calculated according to the actual paths at each period with a random error of.4 km/s. The pattern of the small-scale anomalies both for Rayleigh and Love wave can be largely recovered at 16 s from ambient noise tomography (Figures 5c and 5e). The checkerboard model of 18 km wavelength is also well retrieved both for Rayleigh and Love wave at 5 s from two-plane-wave tomography (Figures 5b, 5d, and 5f). The magnitude of anomaly and the overall model resolution are reduced at the edge of the study region. This synthetic modeling shows the resolving power of these two methods and data. Our lateral resolution is 1 km at short periods and 18 km at longer periods. Based on the resolution tests and the raypath distribution, the poor-resolution areas are clipped off in the phase velocity and shear wave velocity maps below. Only features in the well-resolved region are described and discussed in the following sections. Combining Rayleigh and Love wave phase velocity maps at 8 35 s from ambient noise tomography and those at 2 91 s by two-plane-wave tomography, we obtain the phase velocity maps for Rayleigh and Love wave from 8 to 91 s. For the overlapping periods (2 35 s), the phase speeds of Rayleigh and Love wave measured from ambient noise tomography (Figures 6a and 6b) generally agree with those from the twoplane-wave tomography (Figures 6c and 6d), respectively. Therefore, we average these two measurements. Both for Rayleigh and Love wave, there is a continuity of velocity patterns between adjacent periods due to the overlapping depth ranges of surface wave sensitivity to structure. The main features on the maps for Rayleigh and Love wave at each period are similar to each other (Figure 7). At the period of 8 s, pronounced low anomaly is observed at the North China Basin; while high anomalies are imaged at the Taihang Mountain, Yan Mountain, and L uliang Mountain (Figures 7a and 7b). Slow anomaly correlates with sedimentary basin. Wang et al. [212] also observed such features from the Rayleigh wave phase velocity tomography. With the period increasing, the North China Basin gradually turns into high-velocity zone and northern part of Fenhe Graben and L uliang Mountain change to low-velocity anomaly from 2 up to 71 s (Figures 7c 7h). The high velocity in North China Basin roughly agrees with a large-scale relatively high group velocity zone found by Huang et al. [23]. Rayleigh and Love waves have different sensitivities to the earth structure. A fundamental mode Rayleigh wave has its peak sensitivity at the depth of 1/3 of the wavelength, while Love wave at the same period tends to be more sensitive to slightly shallower depths [Bensen et al., 27] (Figure 8). In addition, phase velocities of Rayleigh and Love waves are most sensitive to vertically polarized and horizontally propagating shear wave velocity (Vsv) and horizontally polarized and propagating shear velocity (Vsh), respectively. Significant differences between the phase velocities of Rayleigh and Love wave at the same period are also observed at 71 s (Figures 7g and 7h). High-velocity anomaly for Rayleigh wave is imaged in the Yan FU ET AL. SHEAR VELOCITY AND RADIAL ANISOTROPY 2623

6 (a) (b) velocity anomaly (%) velocity anomaly (%) (c) 16 s (d) 5 s velocity anomaly (%) velocity anomaly (%) (e) 16 s (f) 5 s velocity anomaly (%) velocity anomaly (%) Figure 5. Resolution tests for Rayleigh and Love wave phase velocity inversions by (a, c, and e) ambient noise tomography and (b, d, and f) teleseismic two-plane wave tomography. (a and b) Two input models. (c f) Corresponding recovered models for (c and d) Rayleigh and (e and f) Love wave at period of 16 s by ambient noise tomography and 5 s by teleseismic two-plane wave tomography, respectively. FU ET AL. SHEAR VELOCITY AND RADIAL ANISOTROPY 2624

7 (a) 3 s-an (b) 3 s-an (c) 3 s-tp (d) 3 s-tp Figure 6. Phase velocity maps of (left) Rayleigh and (right) Love wave at the period of 3 s from (a and b) ambient noise tomography and (c and d) two-plane wave tomography. The perturbation is relative to the average phase velocity of Rayleigh and Love wave at 3 s period. Mountain (Figure 7g). However, it shows slow velocity of Love wave (Figure 7h). These differences could be caused by isotropic shear wave velocity variation with depth and/or the presence of radial anisotropy. 3. Inversion for Shear Wave Velocity and Radial Anisotropy The determined phase velocities of Rayleigh and Love wave at periods of 8 91 s are jointly inverted for shear wave velocity and anisotropic structure in the lithosphere. In a transversely isotropic medium, the elastic coefficients of material can be expressed by five parameters, b V, a H, n, u, and g [Takeuchi and Saito, 1972], where n 5 b H 2 /b V 2 and u 5 a V 2 /a H 2. The model parameters are b V and n. Since Rayleigh and Love wave phase velocities primarily depend on shear wave velocity, less on density and P wave velocity, we couple P wave velocity to S wave velocity using a constant Poisson s ratio, which is a reasonable approximation for the materials of the crust and upper mantle. For u and g, we scale them to n by Du 52.65Dn and Dg 52.75Dn, where the scaling factors are derived from Montagner and Anderson [1989]. The starting model contains four layers in the crust with the crustal thickness constrained by receiver function [Wang et al., 213; Li et al., 214] and four layers in the upper mantle to 14 km depth. The top layer is for sedimentary rock having a variable thickness as calculated from Wu et al. [211] and Laske et al. [213] ( igppweb.ucsd.edu/gabi/crust1.html). In the inversion, the method of Saito [1988] is used to calculate the FU ET AL. SHEAR VELOCITY AND RADIAL ANISOTROPY 2625

8 (a) 8 s-an (b) 8 s-an (c) 2 s-an (d) 2 s-an (e) 5 s-tp (f) 5 s-tp Figure 7. Map of phase velocity perturbation for (left column) Rayleigh and (right column) Love wave derived from ambient noise tomography (AN) at periods of 8 and 2 s and teleseismic two-plane wave tomography (TP) at 5 and 71 s. The perturbation is relative to the average phase velocity of Rayleigh and Love wave at given periods, respectively. FU ET AL. SHEAR VELOCITY AND RADIAL ANISOTROPY 2626

9 (g) 71 s-tp (h) 71 s-tp Figure 7. (Continued) partial derivatives of phase velocities to shear wave velocity and other anisotropic model parameters in each layer. We introduce vertical smoothing to a correlation of.3 between a layer and its nearest neighbors through off-diagonal terms in the priori model covariance matrix in order to prevent strong vertical oscillations. The model parameters are slightly damped,.2 for b V and.4 for n, respectively. Because the inversion is intrinsically nonlinear, the solutions for model parameters depend strongly on the starting model. To obtain a better starting model and show the resolving power of our methods and data, we perform a test with a synthetic model in which 5% radial anisotropy is only permitted in the upper mantle (black lines in Figure 9b). Fu and Li [215] already show the ability of this method to resolve the radial anisotropy in the crust. Synthetic phase velocities of Rayleigh and Love wave for this model are calculated (symbols in Figure 9a) and then used as data to invert shear velocity and radial anisotropy with two dc/dvsv dc/dvsh depth (km) s 16 s 25 s 5 s 71 s 91 s depth (km) (a) 2 (b) Figure 8. Sensitivity kernels for (a) Rayleigh and (b) Love wave phase speeds at a selection of periods. The kernels are calculated based on the shear velocity model AK135 [Kennett et al., 1995] using the method of Saito [1988]. FU ET AL. SHEAR VELOCITY AND RADIAL ANISOTROPY 2627

10 period (s) Shear wave Velocity (km/s) phase velocity (km/s) Love Rayleigh Depth (km) (a) 14 (b) period (s) Shear wave Velocity (km/s) phase velocity (km/s) Love Rayleigh Depth (km) (c) 14 (d) Figure 9. Synthetic tests for joint anisotropic inversion with (a and b) different initial models and (c and d) resolving ability of vertical variation of radial anisotropy. The synthetic 1-D anisotropic model in Figures 9a and 9b allows radial anisotropy with amplitude of 5% only in the upper mantle. However, it contains negative radial anisotropy (25%) in the lower crust and positive radial anisotropy (5%) in the upper mantle in Figures 9c and 9d. (a and c) Phase velocity dispersion curves. Rayleigh and Love wave observations are shown with solid circles and triangles, respectively. The blue lines are the predictions from the model (blue lines in Figure 9b) using an isotropic initial model. The red lines are the predictions from the model (red lines in Figures 9b and 9d) using an anisotropic initial model with the anisotropy constrained from isotropic Vsv and Vsh. The anisotropic parameters are damped to the initial model during the inversion. (b and d) Synthetic models (black lines) and shear wave velocity profiles obtained from anisotropic inversion (blue and red lines). Solid line is for Vsv, and dashed line is for Vsh. different initial models. The difference between these two initial models for anisotropic inversion is the radial anisotropic parameter n. The first initial model is an isotropic model with b V and n 5 1. While the second one is an anisotropic model where we keep b V the same as the first one, but n 5 Vsh*Vsh/Vsv/Vsv. Vsv and Vsh are obtained by isotropic inversion of Rayleigh and Love wave phase velocity, respectively. b V is obtained from both the Rayleigh and Love wave data by an isotropic inversion in which the start model has only one parameter with value of the average of Vsv and Vsh. The obtained model from anisotropic inversion with an isotropic starting model underestimates Love wave phase velocities at most periods (blue lines in Figure 9a). This may be due to the relative low depth resolution of n which is controlled by Love wave FU ET AL. SHEAR VELOCITY AND RADIAL ANISOTROPY 2628

11 [Nishimura and Forsyth, 1989]. The data misfit could be improved without the constraint on anisotropic parameter, but the inverted model shows strong oscillations. For anisotropic inversion with the introduction of anisotropy in the starting model, the radial anisotropy from the best fitting model (red line in Figure 9b) is about 4.3%, close to the value in the synthetic model. It also improves the data fit significantly, causing a 7% variance reduction in the misfit (red lines in Figure 9a). Our experiment indicates that the starting anisotropic model with the amount of anisotropy constrained by Vsv and Vsh could be a good reference model. The model parameters can be well determined. We also test the resolving power of vertical variation of radial anisotropy through perturbing above experiment by allowing a negative radial anisotropy in the lower crust and positive one in the uppermost mantle (Figure 9d). The strength of each radial anisotropy is 5%. We do the joint anisotropic inversion with initial model parameters constrained by Vsv and Vsh. Both the Rayleigh and Love phase velocities are fit well (Figure 9c). The inverted negative and positive radial anisotropy is 24.1% and 3.8%, respectively (Figure 9d). The rapid vertical variations of radial anisotropy can be largely recovered. Given the range of periods for the used Rayleigh and Love waves, shear wave speed in the crust and upper mantle can be well determined. Radial anisotropy is reliable to the depth of 14 km where both Rayleigh and Love waves have high sensitivity, but it cannot be well resolved at deeper depth due to the low sensitivity of Love wave at relatively long periods (71 91 s) (Figure 8). Phase velocities of surface wave can only provide integrated information about the Earth structure. They are related to the sedimentary thickness, crustal thickness, velocity, and anisotropy. A trade-off between these parameters must be present. We first test the effect of sedimentary thickness on the inverted velocity and anisotropy. The results show that the model without the sediment layer can fit the Rayleigh wave speeds very well, but the misfit to Love wave speeds remains large at short periods (<16 s). However, the misfit reduces greatly when taking into the account of sediment effect. Our tests indicate that changing the thickness of sediment in a range of 2 km only slightly changes the resultant model. Therefore, the thickness of sedimentary layer mainly affects the Vsh from Love wave. For crustal thickness, we find that its uncertainty has a very small effect on the estimate of velocity and anisotropy. The variation of 2 km in crustal thickness can cause a change of shear wave velocity less than 1% at most. 4. Three-Dimensional Structure and Discussion We first perform anisotropic inversion at each grid point with the starting model in which b V is obtained by an isotropic inversion jointly from Rayleigh and Love wave dispersion and n is constrained by Vsv and Vsh through the equation in the synthetic test. Figure 1 displays the 1-D anisotropic models at two map points in the North China Basin and Fenhe Graben, respectively. The North China Basin has a slow sedimentary layer, but both Vsv and Vsh in the crust at the North China Basin point are higher than those in the Fenhe Graben. At the depth above 6 km, radial anisotropy is not significant with the amplitudes less than 2% for the point in Fenhe Graben, while the North China Basin shows larger radial anisotropy. However, large positive radial anisotropy exists both at these two points from 6 to 14 km. Combining all the result beneath each point, we obtain a model of the 3-D shear wave velocity and radial anisotropy in the crust and upper mantle in the northern part of North China. Horizontal slices of the perturbations of Voigt average shear wave velocity (((2Vsv 2 1Vsh 2 )/3) 1/2 ) are shown in Figure 11 in eight layers from surface to 14 km. Our model reveals strong lateral and vertical variations. Shear wave velocity in the upper and middle crust correlates well with geologic provinces on the first order. The North China Basin and Fenhe Graben both show low-velocity anomaly, while the L uliang Mountain, Taihang Mountain and Yan Mountain appear as fast velocity (Figures 11a, 11b, 13a, and 13c). This pattern of velocity variation was also observed by Yang et al. [212]. Relatively thick sediments may be responsible for the observed slow anomalies. The most striking features are the high-velocity imaged beneath the North China Basin (Figures 11c 11h and 13c) and low-velocity observed in northern part of L uling Mountain and Fenhe Graben (Figures 11c 11h and 13c) from the lower crust down to roughly 14 km. This high-velocity feature has been imaged in many previous travel time tomographic studies [Tian et al., 29; Zhao et al., 29; Zheng et al., 211]. The model of Tian et al. [29] even shows this feature down to 3 km. However, Rayleigh wave studies show a much slower shear velocity from surface to the depth of 1 km through inversion of group velocity [An et al., 29; Huang et al., 29] or phase velocity [Obrebski et al., 212]. This inconsistency can FU ET AL. SHEAR VELOCITY AND RADIAL ANISOTROPY 2629

12 period (s) period (s) phase velocity (km/s) phase velocity (km/s) (a) 3. (c) Shear wave Velocity (km/s) Shear wave Velocity (km/s) Depth (km) 6 8 Depth (km) (b) 14 Figure 1. (a and c) Dispersion curves and (b and d) associated best fitting models at two grid points in the North China Basin (1116.5, 139) and Fenhe Graben (1114, 14), respectively. Observed (triangles for Love wave and circles for Rayleigh wave) and predicted (lines) dispersions from the best fitting anisotropic model in Figures 1b and 1d. The red line is for Vsh and the blue line is for Vsv. be due to the difference between the Vsv and Vsh models because the model from Rayleigh wave studies is Vsv, while our model is a Voigt average of Vsv and Vsh and strong positive anisotropy (Vsh > Vsv) is indeed imaged here (Figures 12b 12h). Our model also shows that the size of this high-velocity anomaly decreases with depth (Figures 11c 11h). A small low-velocity anomaly appears in its northeast direction in the depth range of km (Figure 11h). We associate the high-velocity anomaly to the remains of Archean lithosphere which has experienced metamorphism, chemical and thermal erosions at shallow depth, and even been replaced at deep depth in some areas during the significant tectonothermal rejuvenation since the Mesozoic [e.g., Xu, 21; Zheng and Wu, 29]. The prominent low-velocity anomaly in the Taihang Mountain and Fenhe Graben is also imaged by Zhao et al. [29] even down to the transition zone and may be due to the high temperature associated with the sublithospheric warm mantle flow which probably plays an important role during the Mesozoic reactivation. FU ET AL. SHEAR VELOCITY AND RADIAL ANISOTROPY 263

13 (a) -15 km B (b) km A B A shear velocity anomaly (%) shear velocity anomaly (%) (c) 25-Moho km (d) Moho-6 km shear velocity anomaly (%) shear velocity anomaly (%) (e) 6-8 km (f) 8-1 km shear velocity anomaly (%) shear velocity anomaly (%) Figure 11. Maps of shear wave velocity perturbation in the crust and upper mantle. The velocity perturbations are relative to the average values in each layer. FU ET AL. SHEAR VELOCITY AND RADIAL ANISOTROPY 2631

14 (g) 1-12 km (h) km shear velocity anomaly (%) shear velocity anomaly (%) Figure 11. (Continued) Besides shear wave velocity, variation of radial anisotropy provides additional information about subsurface structure and deformation mechanisms of the lithosphere. The best resolved anisotropy is shown in Figure 12 and the strength of radial anisotropy is calculated by 2(Vsh-Vsv)/(Vsh1Vsv) 3 1. A prominent feature of radial anisotropy is the strong positive anisotropy (larger than 4%) imaged beneath the North China Basin in the mid-lower crust (Figures 12b, 12c, 13b, and 13d). This observation that strong positive radial anisotropy in the crust appears in the area that has experienced significant extension is similar to that found in the Basin and Range Province and the Rio Grande rift in the western U.S. [Moschetti et al., 21; Fu and Li, 215]. This positive crustal anisotropy could be explained as subhorizontal alignment of crustal minerals or thin layers and sills of melt intrusions due to lithosphere extension. The northern part of Fenhe Graben shows negative anisotropy in the middle crust but positive in the lower crust (Figures 12b, 12c, and 13b). Chang et al. [213] obtain similar pattern of the radial anisotropy but with smaller amplitude along a profile across this region. The amplitude difference may be due to the method used for radial anisotropy. They obtain Vsv and Vsh from isotropic inversion of Rayleigh and Love wave separately while we calculate Vsv and Vsh from anisotropic joint inversion of Rayleigh and Love wave simultaneously. We account for the trade-off between the velocity and anisotropy. However, they ignore the effect of anisotropy to get the velocity first and then calculate anisotropy by such inaccurate velocity. Our result provides the first 3-D radial anisotropy model with high accuracy. In the upper mantle, the variation of radial anisotropy is smaller than that in the crust. Strong positive anisotropy with Vsh > Vsv is observed in almost the entire region except beneath the Yan Mountain (Figures 12d 12h and 13d). The Taihang Mountain and North China Basin show distinct velocity pattern but almost the same radial anisotropy in the upper mantle (Figures 11d 11h and 12d 12h), which may suggest different thermal state between these regions. The strong magmatism in the North China Basin due to the lithospheric extension was in Mesozoic [Xu et al., 24]. The lithospheric thickness is less than 1 km in this area from S receiver function study [Chen, 29]. However, the new lithosphere might be formed by the materials of mantle flow through thermal annealing. The large amount of mafic intrusion and mantle residual after extraction of melt produce the high-velocity anomaly, while subhorizontal alignment of mineral or thin layers and sills of melt intrusions cause the positive radial anisotropy in the upper mantle (Moho to 14 km) during the Mesozoic extension. The high velocity and positive radial anisotropy suggest a depleted mantle source for Cenozoic basalts [Zhou and Armstrong, 1982] and the lack of recent anthesnosphere upwelling as proposed by Tian and Zhao [213] in the North China Basin. For the Fenhe Graben and Taihang Mountain, Zhao and Xue [21] propose a mantle flow model in which the ascending warm material due to the subduction of the Pacific Plate may flow up along the basement of the eastward thinning lithosphere according to the P and S wave velocity images [Zhao et al., 29] and SKS measurements. This hot material may raise the mantle temperature, cause the Quaternary rift in the FU ET AL. SHEAR VELOCITY AND RADIAL ANISOTROPY 2632

15 (a) -15 km (b) km (c) 25-Moho km (d) Moho-6 km (e) 6-8 km (f) 8-1 km Figure 12. Maps of radial anisotropy in the crust and upper mantle. Positive radial anisotropy is for Vsh > Vsv, and negative anisotropy is for Vsh < Vsv. FU ET AL. SHEAR VELOCITY AND RADIAL ANISOTROPY 2633

16 (g) 1-12 km (h) km Figure 12. (Continued) region [Xu et al., 24], and produce the low velocity we observed. The subhorizontal mantle flow may pull the weak lithosphere and make mantle lithospheric materials aligning horizontally, causing the positive radial anisotropy with Vsh > Vsv. The low velocity and positive radial anisotropy suggest that the rifting in the Fenhe Graben and Taihang Mountain is active. A LM FG NCB A A LM FG NCB A depth (km) (a) depth (km) (b) longitude longitude shear velocity anomaly B NCB YM B B NCB YM B depth (km) (c) depth (km) (d) latitude latitude shear velocity anomaly Figure 13. Vertical profiles of (a and c) shear wave velocity perturbation and (b and d) radial anisotropy in the crust and upper mantle. The location of the profiles is shown in Figure 11a. The perturbations are relative to the average values in each layer as shown in Figure 11. The black triangle indicates the location of Datong volcano. LM, FG, TM, YM, and NCB are the same as in Figure 1. FU ET AL. SHEAR VELOCITY AND RADIAL ANISOTROPY 2634

17 5. Conclusions The distinct seismic structure in the northern part of North China reveals how the lithosphere extension evolves from east to west. Rifting in the North China Basin has caused significant extensional deformation and strong radial anisotropy. High velocity from the lower crust to the depth of 14 km is probably due to the mafic intrusion and depleted mantle residual after the extraction of melt associated with the strong lithosphere extension. However, the rift in the Fenhe Graben and Taihang Mountain may be at the early stage of rifting, where partial melt is likely in the lithosphere and mantle upwelling and horizontal flow pull is possible, producing the low velocity and positive radial anisotropy. Our model in the crust and upper mantle from the integrated tomography of ambient noise and earthquake provides geophysical evidence for the first time about the systematic evolution of lithosphere extension from east to west in the northern part of North China. Acknowledgments Data are from the Institute of Geophysics, China Earthquake Administration and requested through We thank the participants in the North China Seismic Array for collecting the data. We thank M. Ritzwoller and Y. Yang for providing codes on ambient noise tomography of Rayleigh wave and thank J. Liu and L. Li for their assistance on Love wave phase velocity inversion. Most figures in this paper were made with General Mapping Tools [Wessel and Smith, 1998]. Two anonymous reviewers provide constructive suggestions that help to improve this paper. This work was supported by grant from the National Natural Science Foundation of China and 214IES12 from the science foundation of the Institute of Earthquake Science, China Earthquake Administration. References An, M., M. Feng, and Y. Zhao (29), Destruction of lithosphere within the north China craton inferred from surface wave tomography, Geochem. Geophys. Geosyst., 1, Q816, doi:1.129/29gc2562. An, W., and Z. Su (28), Landform of volcanoes in Datong of Shanxi, Earthquake Res. Shanxi, 133, 1 9. Barmin, M. P., M. H. Ritzwoller, and A. L. Levshin (21), A fast and reliable method for surface wave tomography, Pure Appl. Geophys., 158, Bensen, G. D., M. H. Ritzwoller, M. P. Barmin, A. L. Levshin, F. Lin, M. P. Moschetti, N. M. Shapiro, and Y. Yang (27), Processing seismic ambient noise data to obtain reliable broad band surface wave dispersion measurements, Geophys. J. Int., 169, Chang, C., L. Chen, H. Yao, M. Jiang, and B. Wang (213), Distinct variations of crustal shear wave velocity structure and radial anisotropy beneath the North China Craton and tectonic implications, Gondwana Res., 23, Chen, L. (29), Lithospheric structure variations between the eastern and central North China Craton from S- and P-receiver function migration, Phys. Earth Planet. Inter., 173, Deng, J. X. M., H. Zhao, Z. Wu, Z. Luo, and S. Su (24), A new model for the dynamic evolution of Chinese lithosphere: continental rootsplume tectonics, Earth-Science Reviews, 65, Fan, W. M., and M. A. Menzies (1992), Destruction of aged lower lithosphere and accretion of asthenosphere mantle beneath eastern China, Geotectonica et Metallogenia, 16, Forsyth, D. W., S. C. Webb, L. M. Dorman, and Y. Shen (1998), Phase velocities of Rayleigh waves in the MELT experiment on the east Pacific rise, Science, 28, Forsyth, D. W., and A. Li (25), Array-analysis of two-dimensional variations in surface wave phase velocity and azimuthal anisotropy in the presence of multi-pathing interference, in Seismic Earth: Array Analysis of Broadband Seismograms, Geophys. Monogr. Ser., vol. 157, edited by A. Levander and G. Nolet, pp , AGU, Washington, D. C. Foster, A., M. Nettles, and G. Ekstrom (214), Overtone interference in array-based Love-wave phase measurements, Bull. Seismol. Soc. Am., 14, Friederich, W. (23), The S-velocity structure of the East Asian mantle from inversion of shear and surface waveforms, Geophys. J. Int., 153, Fu, Y. V., and A. Li (215), Crustal shear wave velocity and radial anisotropy beneath the Rio Grande rift from ambient noise tomography, J. Geophys. Res. Solid Earth, 12, , doi:1.12/214jb1162. Gao, S., et al. (24), Recycling lower continental crust in the North China craton, Nature, 432, Gao, S., J. Zhang, W. Xu, and Y. Liu (29), Delamination and destruction of the North China Craton, Chinese Science Bulletin, 54, Huang, J., and D. Zhao (26), High-resolution mantle tomography of China and surrounding regions, J. Geophys. Res., 111, B935, doi: 1.129/25JB466. Huang, J., and D. Zhao (29), Seismic imaging of the crust and upper mantle under Beijing and surrounding regions, Phys. Earth Planet. Inter., 173(3-4), Huang, Z., W. Su, Y. Peng, Y. Zheng, and Y. Li (23), Rayleigh wave tomography of China and adjacent regions, J. Geophys. Res., 18(B2), 273, doi:1.129/21jb1696. Huang, Z., Y. Li, Y. Zheng, and Y. Peng (29), The lithosphere of North China Craton from surface wave tomography, Earth Planet. Sci. Lett., 288, Jia, S., and X. Zhang (25), Crustal structure and comparison of different tectonic blocks in north China, Chin. J. Geophys., 48(3), Kennett, B. L. N., E. R. Engdahl, and R. Buland (1995), Constraints on seismic velocities in the Earth from travel times, Geophys. J. Int., 122, Laske, G., G. Masters, Z. Ma, and M. Pasyanos (213), Update on CRUST1. A 1-degree global model of Earth s crust, Geophys. Res. Abstr., 15, EGU Levshin, A. L., and M. H. Ritzwoller (21), Automated detection, extraction, and measurement of regional surface wave, Pure Appl. Geophys., 158(8), Levshin, A. L., T. B. Yanovskaya, A. V. Lander, B. G. Bukchin, M. P. Barmin, L. I. Ratnikova, and E. N. Its (1989), Seismic Surface Waves in a Laterally Inhomogeneous Earth, edited by V. I. Keilis-Borok, Kluwer Acad., Norwell, Mass. Li, A., and L. Li (215), Love wave tomography in southern Africa from a two-plane wave inversion method, Geophys. J. Int., 22(2), 15 12, doi:1.193/gji/ggv23. Li, Y., M. Gao, and Q. Wu (214), Crustal thickness map of the Chinese mainland from teleseismic receiver functions, Tectonophysics, 611, Lin, F., M. P. Moschetti, and M. H. Ritzwoller (28), Surface wave tomography of the western United States from ambient seismic noise: Rayleigh and Love wave phase velocity maps, Geophys. J. Int., 173, Liu, M., X. Cui, and F. Liu (24), Cenozoic rifting and volcanism in eastern China: a mantle dynamic link to the Indo-Asian collision? Tectonophys., 393, FU ET AL. SHEAR VELOCITY AND RADIAL ANISOTROPY 2635

18 Menzies, M., Y. Xu, H. Zhang, and W. Fan (27), Integration of geology, geophysics and geochemistry: A key to understanding the North China Cration, Lithos, 96, Montagner, J. P., and D. L. Anderson (1989), Petrological constraints on seismic anisotropy, Phys. Earth Planet. Inter., 54, Moschetti, M. P., M. H. Ritzwoller, F. Lin, and Y. Yang (21), Seismic evidence for widespread western-us deep-crustal deformation caused by extension, Nature, 464, Nettles, M., and A. Dziewonski (211), Effect of higher-mode interference on measurements and models of fundamental-mode surfacewave dispersion, Bull. Seismol. Soc. Am., 11, Nishimura, C. E., and D. W. Forsyth (1989), The anisotropic structure of the upper mantle in the Pacific, Geophys. J., 96, Niu, F., and J. Li (211), Component azimuths of the CEArray stations estimated from P-wave particle motion, Earthquake Sci., 24, Obrebski, M., R. M. Allen, F. Zhang, J. Pan, Q. Wu, and S. Hung (212), Shear wave tomography of China using joint inversion of body and surface wave constraints, J. Geophys. Res., 117, B1311, doi:1.129/211jb8349. Priestley, K., E. Debayle, D. McKenzie, and S. Pilidou (26), Upper mantle structure of eastern Asia from multimode surface waveform tomography, J. Geophys. Res., 111, B134, doi:1.129/25jb482. Ritzwoller, M. H., and A. L. Levshin (1998), Surface wave tomography of Eurasia: Group velocities, J. Geophys. Res., 13, Saito, M. (1988), DISPER8: A subroutine package for the calculation of seismic normal-mode solutions, in Seismological Algorithms: Computational Methods and Computer Programs, pp , edited by D. J. Doornbos, Elsevier, N. Y. Takeuchi, H., and M. Saito (1972), Seismic surface waves, in Methods in Computational Physics, pp , Academic, N. Y. Tarantola, A., and B. Valette (1982), Generalized non-linear problems solved using the least-squares criterion, Rev. Geophys., 2, Tian, Z., P. Han, and K. Xu (1992), The Mesozoic-Cenozoic east China rift system, Tectonophys., 28, Tian, Y., and D. Zhao (211), Destruction mechanism of the North China Craton: Insight from P and S wave mantle tomography, J. Asian Earth Sci., 42, Tian, Y., and D. Zhao (213), Reactivation and mantle dynamics of North China Cration: Insight from P-wave anisotropy tomography, Geophys. J. Int., 195, Tian, Y., D. Zhao, R. Sun, and J. Teng (29), Seismic imaging of the crust and upper mantle beneath the North China Craton, Phys. Earth Planet. Inter., 172(3-4), Wang, J., D. Zhao, and Z. Yao (213), Crustal and uppermost mantle structure and seismotectonics of North China Craton, Tectonophysics, 582, Wang, W., J. Wu, and L. Fang (212), High resolution Rayleigh wave phase velocity tomography in northern North China, Geophys. J. Int., 189, Wang, X., Z. Ding, and L. Zhu (213), Lithospheric structure beneath Tanghai-Shangdu seismic profile [in Chinese], Chin. J. Geophys., 56(11), Wessel P., and W. H. F. Smith (1998), New, improved version of the Generic Mapping Tools released, EOS Transactions AGU, 79, 579. Wu, Y., Z. Ding, and L. Zhu (211), Crustal structure of the North China Craton from teleseismic receiver function by the common conversion point stacking method [in Chinese], Chin. J. Geophys., 54(1), Xu, Y. (21), Thermo-tectonic destruction of the Archaean lithospheric keel beneath the Sino-Korean craton in China: Evidence, timing and mechanism, Phys. Chem. Earth, 26, Xu, P., and D. Zhao (29), Upper-mantle velocity structure beneath the North China Craton: Implications for lithospheric thinning, Geophys. J. Int., 177(3), Xu, Y., S. Chung, J. Ma, and L. Shi (24), Contrasting Cenozoic lithospheric evolution and architecture in western and eastern Sino-Korean craton: Constraints from geochemistry of basalts and mantle xenoliths, J. Geol., 112, Yang, T., et al. (212), 3-D S-wave velocity structure of crust and upper mantle beneath North China [in Chinese], Prog. Geophys., 27(2), Zhang, X. (1996), Fine crustal structure in Yanqing-Huailai region by deep seismic reflcetion profiling, Chin. J. Geophys., 39(3), Zhao, L., and M. Xue (21), Mantle flow pattern and geodynamic cause of the North China Craton reactivation: Evidence from seismic anisotropy, Geochem. Geophys. Geosyst., 11, Q71, doi:1.129/21gc368. Zhao, L., R. M. Allen, T. Zheng, and S. Hung (29), Reactivation of an Archean craton: Constraints from P- and S-wave tomography in North China, Geophys. Res. Lett., 36, L1736, doi:1.129/29gl Zheng, J., S. Y. Oreilly, W. L. Griffin, F. Lu, and M. Zhang (1998), Nature and evolution of Cenozoic lithospheric mantle beneath Shandong Peninsula, Sino-Korean Craton, eastern China, International Geology Review, 4, Zheng, Y., W. Shen, L. Zhou, Y. Yang, Z. Xie, M. H. Ritzwoller (211), Crust and uppermost mantle beneath the North China Craton, northeastern China, and the Sea of Japan from ambient noise tomography, J. Geophys. Res., 116, B12312, doi:1.129/211jb8637. Zheng, Y., and F. Wu (29), Growth and reworking of cratonic lithosphere, Chinese Science Bulletin, 54, Zhou, X., and R. L. Armstrong (1982), Cenozoic volcanic rocks of eastern China-secular and geographic trends in chemistry and strontium isotopic composition, Earth Planet. Sci. Lett., 59, FU ET AL. SHEAR VELOCITY AND RADIAL ANISOTROPY 2636