This paper presents a tomographic study on the S wave velocity structure of China and adjacent regions. Group velocity dispersions of fundamental Rayleigh waves along more than 4000 paths were determined with frequency-time analysis. The study region was divided into a 1° × 1° grid, and velocities in between grid nodes were calculated by bilinear interpolation. The Occam's inversion scheme was adopted to invert for group velocity distributions. This method is robust and allows us to use a fine grid in model parameterization and thus helps to restore a more realistic velocity pattern. Checkerboard tests were carried out, and the lateral resolution was estimated to be 4°–6° in China and its eastern continental shelves. The resulting group velocity maps from 10 to 184 s showed good correlation with known geological and tectonic features. The pure path dispersion curves at each node were inverted for shear wave velocity structures. The three-dimensional velocity model indicates thick lithospheres in the Yangtze and Tarim platforms and hot upper mantles in Baikal and western Mongolia, coastal area and continental shelves of eastern China, and Indochina and South China Sea regions. The Tibetan Plateau has a very thick crust with a low-velocity zone in its middle. Beneath the crust a north dipping high-velocity zone, mimicking a subducting plate, reaches to 200 km in depth and reaches to the Kunlun Mountains northward. In northern Tibet a low-velocity zone immediately below the Moho extends eastward then turns southward along the eastern edge of the plateau until it connects to the vast low-velocity area in Indochina and the South China Sea.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 China and vicinal areas are located in the southeastern part of Eurasia, cornered at the junction of three major plates in the world. To China's east the Pacific and Philippine plates are subducting under the Eurasia plate. To the south, collision and subduction are vigorously occurring between the Indo-Australia and Eurasia plates. Cenozoic and contemporary tectonic activities are intense in this region as manifested by strong topographic contrasts, extensive crustal movement, and active seismicity in most parts of the region. This makes it an ideal place for studying continental dynamics; consequently, its crust and upper mantle structures are of great interest to geoscientists in the world.
 Surface waves provide an important data source for the structural study in view of the sparse distribution of seismograph stations, especially in the western part of this region. In the 1960s, Zeng and Song  and Song and Tan  used phase and group velocities, respectively, to study the crustal structure in China. Later, Feng et al.  carried out a more detailed study on the crustal structure of China with surface wave data. In these studies the limited passband of instrument response restricted the attainable depth of structural studying. In the early 1990s, Song et al. [1993a, 1993b] studied the three-dimensional (3-D) S wave structure of China and adjacent regions using surface waves recorded by a network of twenty-seven 763-type long-period seismographs in China and three World-Wide Standardized Seismograph Network (WWSSN) stations. They divided the region into a 4° × 4° grid, assuming a constant velocity in each grid cell. The Rayleigh group velocity distributions were determined, and then the S wave velocity structure from 0 to 240 km was inverted for each cell. Xu et al.  used the same kind of data to study the phase velocity and structure in eastern China.
 The establishment of broadband digital seismograph networks in China and surrounding regions greatly improved the path coverage and data quality; surface wave tomography became a more feasible task. Wu et al.  conducted a tomographic study on Siberia, China, and neighboring areas with Rayleigh waves between 30 and 70 s. Curtis et al.  studied the phase velocity distribution of surface waves between 20 and 170 s in Eurasia. Ritzwoller and Levshin  obtained the group velocity maps of Rayleigh waves from 20 to 200 s and of Love waves from 20 to 125 s in Eurasia. Villasenor et al.  studied the S wave velocity structure of central Eurasia with surface wave data, assuming transverse isotropy from Moho down to 220 km. A large number of surface wave records were used in these studies, and the results showed generally good correlation with known geologic and tectonic features in the study region. In recent years, waveform modeling also has been used in regional structure studies. Cao et al.  studied the S wave velocity structure in the South China Sea and adjacent regions by surface waveform inversion. Lebedev and Nolet  studied the upper mantle structure in the Southeast Asia and western Pacific regions by inversion of both S and Rayleigh waveforms, which enabled them to resolve structures in the entire upper mantle.
 In this paper we present the result of a surface wave tomographic study of China and neighboring regions. The conventional two-step method was used in the study. That is, group velocity distributions of fundamental Rayleigh waves between 10 and 184 s were first obtained through a 2-D inversion; then the pure path dispersion curves at each grid node were inverted to get a 3-D S wave velocity model of the study region. Our purpose was to improve the structure model relying mainly on two measures. First, we utilized more data, particularly the records from small events at short epicentral distances, to improve the path coverage. Consequently, the group velocity measurements at shorter periods were reasonably good. Second, in the 2-D inversion for group velocity distributions we developed the Occam's inversion method [Constable et al., 1987; deGroot-Hedlin and Constable, 1990; Huang and Zheng, 1998] and adopted a fine grid in model parameterization. Synthetic data tests indicated that this method is advantageous for suppressing the adverse effect of measurement errors and inadequate path coverage, as well as for more accurately restoring the shape of tectonic blocks.
2. Data and Method
2.1. Study Region
 The study region is confined within latitudes 5°–55°N and longitudes 68°–150°E (Figure 1). This region contains oceanic and continental areas. The oceanic area includes a small part of the old northwest Pacific Ocean and young sea basins formed by back arc spreading, such as the Japan Sea and the Philippine Sea, as well as the Okinawa Trough, which is currently undergoing strong spreading. The subduction of the Pacific plate and the Philippine plate beneath Eurasia plays an important role in the modern geodynamic process of east Asia. In the land area, there are the world-famous Tibetan Plateau and other strongly uplifting areas such as Pamir, Tianshan, and Altai. There are also ancient platforms including Yangtze, Tarim, and Siberia, as well as reactivated platform areas like Shanxi graben and north China plain which show strong contemporary tectonic movement. On the whole, this region is tectonically very complicated and is one of the regions with strong tectonic movement and intensive seismicity in the world. The study of crust and upper mantle structure is of great importance for understanding the lithospheric dynamics.
 Rayleigh wave data recorded by digital seismograph stations in the study region were utilized in the tomographic study. Within China, there are 11 China Digital Seismograph Network (CDSN) stations; CDSN's earliest station, BJI, started providing data in late 1987, and the latest station, XAN, started in 1994. Data also come from 22 digital stations in the surrounding areas (Figure 1), including some stations no longer in operation, such as KAAO and SHIO. Some of the stations (SRO stations) started in the 1970s, but most of them were built in the 1990s. To improve path coverage, we also used some records from the FREESIA network of Japan and the long-period data from portable stations in Tibet deployed by a Sino-U.S. joint research project during 1992–1993. These stations are marked as open triangles in Figure 1.
 For earthquake locations and origin times we used International Seismological Centre reports for events before June 1998 and preliminary determination of epicenter reports for events between 1998.6 and 2000.1. To improve the dispersion measurement at short periods, we included recordings of moderate earthquakes, some below magnitude 5, in the data set. The average epicentral distance over all periods is ∼3000 km. It is 2400 km at 10 s and increases with period to 3900 km at 184 s.
2.3. Dispersion Measurement
 Rayleigh wave group velocity dispersions between 10 and 184 s periods were determined with our frequency-time analysis software. It is an interactive graphic program based mainly on the multiple filter technique [Dziewonski et al., 1969; Dziewonski and Hales, 1972]. The program is designed to extract dispersion curves as efficiently and objectively as possible. It automatically reads the data following a work list, makes instrument response correction, makes high and low filtering, and sets an appropriate time window for Rayleigh waves. It then calculates the energy distribution on the time-frequency plane, picks out the energy peak and next five local maxima at each period, and displays them with different symbols. At shorter periods, there is generally more than one branch of incoming energy. The analyst needs to set the high and low period for usable dispersion range and to assign a quality flag to the dispersion curve. Sometimes the analyst may manually select a proper dispersion branch at short periods and smooth the dispersion curve if he/she deems it appropriate.
 After the dispersions along all paths were measured, dispersion curves along similar paths were grouped into composite paths. That is, we take the earthquakes within 15 km from a group center as a cluster and calculate average dispersion curves to be used in the inversion. After a preliminary inversion we discarded a small number of paths that have very large travel time residuals. The final path numbers in the period range 10–100 s are >2500 (counting a composite path as a single one), and above 125 s they decrease fast. For example, the path number is 2563 at 10 s, 3785 at 34.1 s, and 897 at 184 s.
Figure 2 shows the path coverage at 34.1 and 158 s. Although the general situation is good, the path distribution is not even in the study region. The coverage is better in the eastern part of the study region. In the margins the path coverage is the worst, where the paths are generally in one direction without crossing each other. The northeast, southeast, and southwest corners are not covered with any path; these areas will be unshaded in the resulting velocity maps.
 During the inversion for group velocity distributions the computer automatically leaves those marginal nodes out of the inversion that do not have a single path passing by. However, inside the region, all nodes participate in the inversion even if some nodes have no paths passing by, so that there will be no holes inside the region. The velocities at these nodes will be the average value of surrounding nodes owing to the smoothness constraints used in the inversion.
2.4. Inversion for Group Velocity Maps
 Assuming surface waves are propagated along great circle paths, the total travel time along a path is the sum of travel times spent on all elementary path segments. We divide the study region into a 1° × 1° grid. The velocities at the grid nodes represent the discrete group velocity model; velocities in between the nodes are calculated by bilinear interpolation. Thus the inversion problem is reduced to solving a linear equations system.
 In order to suppress the adverse effect of dispersion measurement errors and improper path coverage we adopted the Occam's inversion method proposed by Constable et al.  and deGroot-Hedlin and Constable . This method seeks a smooth model that reasonably satisfies the observational data. Thereby we try to extract the major tectonic features of the study region while suppressing as much as possible artifacts caused by measurement errors, path coverage defects, and inadequacy of basic assumptions such as great circle propagation.
 This Occam's inversion method was elucidated by Constable et al.  and deGroot-Hedlin and Constable . For completeness, we give a brief account of the method here. Denote the velocity model with N unknowns as m and the observation data of M paths as d, and put the observational errors in a diagonal matrix W = diag(1/σ1, 1/σ2, … 1/σM). There is a definite functional relation between the data and model parameters, denoted as d = F(m), and the corresponding partial derivative matrix is J. The smoothness of the model is described by roughness R defined as
where δXm and δYm are vectors composed of first-order spatial partial derivatives in x and y directions, respectively. Parallels denote the modulus of a vector, and δX and δY are N × N matrices in the following form:
Δ is the distance between neighboring nodes in x or y direction. If the node corresponding to a diagonal element is on the rim of the region, all the elements in that row are zeros. Otherwise there are two nonzero elements, and the position of the off-diagonal element depends on the numbering of nodes.
 The objective of inversion is to minimize the following functional:
The first term represents the fitting to observational data, the second term is the smoothness constraint, and μ is a smoothing factor. As μ becomes larger, the resulting model becomes smoother. In theoretical tests with synthetic data it was found that the best result was achieved when the decrease of residual started to slow down with decreasing μ. After that the residual reduced slightly, whereas the roughness of model increased rapidly [Huang and Zheng, 1998].
 In the present inversion, wave propagation is fixed along the great circle path; therefore the inversion problem is essentially linear. At each period we assume a uniform initial model with a velocity averaged over all paths at that period. If the initial model is m0, the resulting model is
 The equation system is solved by the singular value decomposition method. This inversion method is robust, and a stable solution can be obtained even if the number of unknowns is much greater than the path number. As a consequence, we can use a much finer grid and basically eliminate the objectiveness in model discretization. Furthermore, the inversion result is improved in the margins and areas of poor paths coverage. On the other hand, the introduction of smoothness constraints into the inversion smears sharp velocity boundaries that may exist between tectonic blocks and reduces the power to resolve small-scale heterogeneities. The selection of μ value in the inversion of actual data depends largely on subjective judgment. The final model is the result of a compromise between having more details and maintaining a reasonable smoothness for the velocity model.
 Checkerboard tests were conducted to estimate the achievable resolution. Figure 3a shows a theoretical velocity model that is discretized into a 1° × 1° grid as the real model. The size of alternatively high- and low-velocity cells is 3°. Each cell has a constant velocity 5% above or below an average velocity of 3.5 km/s. Synthetic data of group velocities were calculated according to the actual paths at period 34.1 s, and then a random error in between +0.05 and −0.05 km/s was added to each path. The average absolute error was 0.025 km/s. Then the data were used to reconstruct the velocity model by the same inversion method and smoothing factor as used for actual data. The resulting model is shown in Figure 3b.
 It can be seen from Figure 3b that the resolving power is generally not bad in the Chinese territory and neighboring seas but it deteriorates toward the periphery. Figure 3c shows the result of a similar test with the path coverage at 158 s. In that case the theoretical velocity model has a 5° × 5° checkerboard pattern and is not shown here. The reconstructed velocity image is basically correct in most of China and neighboring seas.
 It should be pointed out that in these tests the theoretical travel times were calculated along great circle path, exactly like the calculations in the inversion. In reality, such is not the case, and the errors caused by off great circle propagation or multipathing are most likely systematic rather than random. Therefore the actual resolution should be worse than that in the tests. In view of this we would like to put, as a rough estimation, the resolution as 4° for shorter periods and 6° for longer periods within Chinese territory and neighboring seas.
3. Rayleigh Wave Group Velocity Distribution
 We obtained Rayleigh group velocity maps at 39 periods from 10 to 184 s. As the Rayleigh wave energy penetrates deeper with increasing period, the group velocity maps demonstrate the lateral variations of shear wave structure in increasingly greater depth ranges. We divide these maps into four groups of increasing period, discuss the relation of velocity pattern with surface geology and tectonics of the region (Figure 4), and show two typical maps for each group in Figure 5. These group velocity maps as well as the shear wave velocity profiles in sections 4.1–4.4 were all produced using the Generic Mapping Tools [Wessel and Smith, 1995]. They display the absolute velocities rather than velocity anomalies, and the nonlinear color scale is area equalized, meant to achieve maximum clarity in the whole region.
3.1. Group Velocity Maps Between 10 and 20 s
 In the period range between 10 and 20 s the group velocities are much influenced by the shallow part of the crust. In the 10 s map (Figure 5a) the major sedimentary basins of China, such as Tarim, Junggar, Qaidam, Sichuan, and Ordos, are clearly manifested. Mountain areas generally show higher velocities than plains and basins. For example, the Shanxi graben in between Ordos basin and north China plain and the Qinling-Dabie mountain ranges in between the north and south China blocks appear as relatively high velocity zones. In the continental shelf area, prominent low velocities are apparently correlated with river mouths of major water systems such as the Yellow River, the Changjiang (Yangtze River), the Zhujiang (Pearl River), the Ganges, and the Irrawady River [Metivier et al., 1999].
 The 19.9 s velocity map (Figure 5b) still shows the influence of sedimentary layers. At this period the Rayleigh wave energy is mostly concentrated in the crust in the land areas, but it penetrates into the upper mantle of high velocity in oceanic areas. Therefore the map displays high group velocities in the Japan Sea, Pacific Ocean, Philippine Sea, and South China Sea, which form a sharp contrast with the low velocities in the land area and continental shelves.
3.2. Group Velocity Maps Between 20 and 60 s
 In the period range between 20 and 60 s the group velocity maps are strongly influenced by the thickness of the crust. The 34.1 s map (Figure 5c) is typical, showing the crustal feature of the land area of China, which can be divided approximately along 105°E longitude into west and east parts. In the western part the crust is thick as compared with the eastern part. The Tibetan Plateau stands out as having a conspicuous low velocity owing to its very thick crust. Dense contour lines along its periphery indicate rapidly varying crustal thickness. In this part the contour lines trend dominantly in an east-west direction.
 In the eastern part the contour lines are generally in a NNE or NE direction, and the crust thins toward the east or southeast. There is a belt of high-velocity gradient passing near stations HIA, BJI, XAN, ENH, and farther to the southwest. This belt coincides with a gradient belt of Bouguer gravity anomalies [Ma, 1989], which is also confirmed to be a belt of rapidly varying crustal thickness. The general pattern of group velocity contour lines on the land is quite similar to that of Moho depth variations [Ma, 1989]. In the sea areas a linear feature of relatively low velocities correctly delineates the position of island arcs.
 From 40 to 60 s the group velocity distribution evolves from a pattern dominated by the crust to one controlled by lithospheric features. The Okinawa Trough, especially its northern segment, shows up as a prominent low-velocity zone in the east of the study region (Figure 5d). This low velocity is attributed to the active back arc spreading caused by the subduction of the Philippine plate.
3.3. Group Velocity Maps Between 60 and 125 s
 In the period range between 60 and 125 s the pattern of group velocity distribution varies slowly and shows some stable features. Figures 5e and 5f are the 73.5 and 100 s velocity distribution maps, respectively. They show the characteristics of first-order lithospheric blocks and have a clear correlation with the tectonic regionalization of the study region (Figure 4). To the west of ∼103°E the north India platform, the Tibetan Plateau, the eastern Tianshan and Tarim platforms, and west Mongolia show alternatively high and low velocities. From the maps above 50 s it is easily seen that the low velocity of the Tibetan Plateau diminishes gradually with increasing period, while the low velocity of western Mongolia becomes more prominent as the period increases. These phenomena indicate that the low velocity of the Tibetan Plateau can be attributed to its thick crust, whereas the low velocity of western Mongolia has a deep origin in the upper mantle. By analyzing phase velocity anomalies, Griot et al.  pointed out that much of the low-velocity zone in Tibet at short periods could be attributed to its unusually thick crust.
 To the east of 103°E we can also see four zones of alternatively low and high velocities but in opposite sense to their western counterparts. In the south the large area in between Indo-Burma arc and Philippines is a low-velocity zone, indicating a thin lithosphere and shallow asthenosphere there. The high velocity of Yangtze platform and the low velocity in north China platform (mainly including the Shanxi graben, the north China plain, the Bohai Sea, and the northern Yellow Sea) form a sharp contrast. The boundary between them, as manifested by dense contour lines, follows the Qinling-Dabie mountain ranges quite well. North China and Yangtze are both Precambrian platforms with very old basement rock. The clear boundary and distinctive lithospheric characteristics indicate that they underwent different tectonic histories. The northward subduction and subsequent collision along Qinling-Dabie during the suturing process [Mattauer et al., 1985] might have caused profound change to the north China platform, resulting in lithosphere thinning and asthenosphere rising. In later time the subduction of the Kula-Pacific plate under east Asia did not affect the western part of Yangtze platform (west of ∼113°E) and left the boundary unchanged. However, it reworked the eastern part of the Yangtze platform, resulting in a fundamental difference in the structures of the western and eastern Yangtze platforms.
 It is interesting to notice that the low-velocity zones in the land area of Figures 5e and 5f are seismically active. Almost all M ≥ 7 inland earthquakes took place inside the low-velocity zones or along the boundaries between high and low velocities. Intraplate earthquakes can be viewed as indicators of plate deformation. Therefore, although earthquakes are a phenomenon occurring in the crust, they are closely related to the deformation of whole lithosphere. The low velocities in this period range mean that the surficial 100 km or so of the Earth has a lower average velocity and hence is weaker and easier to be deformed. This is in accord with the notion of Bourne et al.  that the motion of crustal blocks is driven by flow of the lower lithosphere.
 Above the 50 s period the young sea basins, such as the Japan Sea, Philippine Sea, and South China Sea, started to change from high to low velocities, indicating that the asthenosphere in these areas is shallow. The Philippine Sea was formed by back arc spreading in the wake of Pacific subduction. The west Philippine Sea basin was formed about 60–35 Ma,while the east Philippine Sea is much younger, about 30–15 Ma. Although the resolution of our inversion is poor in this area, the difference between the east and west Philippine Sea is still visible. The low velocities behind the Burma arc are also evident. On the other hand, the northwest Pacific Ocean, which is the oldest sea basin in the world, remains a high-velocity area.
3.4. Group Velocity Maps Between 125 and 184 s
 With the further increase of period the group velocity distribution pattern shows more features of the asthenosphere and gradually loses its correlation with surface geology. In the 146 and 185 s velocity maps (Figures 5g and 5h) we see a triangle-shaped high-velocity core in the land area consisting of parts of the Yangtze and Tarim platforms and the Inner Mongolia-Great Xinan fold system. For periods above 60 s the ancient Yangtze craton remains as the most prominent high velocity. It has the deepest continental root of China. The Tarim platform has a relatively thick lithosphere. On the other hand, the high-velocity zone along the Inner Mongolia-Great Xinan fold system is not obvious around 100 s. As seen from the later shear velocity profiles, there is no continental root beneath this fold system, but its lithosphere has relatively higher velocity and the asthenosphere is thin.
 In Figures 5g and 5h we see three large-scale low-velocity zones. One is centered at the southwest end of Lake Baikal and extends south and westward. Because of the poor ray coverage in the north margin of the study region, the scope and shape of the low-velocity zone will not be accurate; nevertheless, the existence of this low velocity should not be in doubt. Lake Baikal is a well-known continental rift region. There are controversies in regard to its origin. Zorin et al.  found that the lithosphere beneath Lake Baikal is only 40–50 km thick, with a wide asthenosphere upwelling reaching to the bottom of its crust. After reviewing various kinds of geophysical data, Logatchev and Zorin  pointed out that the Baikal rift system is mainly caused by asthenospheric diapirism, although the intraplate stress related to India-Eurasia collision might have played some role. The Mongolia plateau is a vast area southwest of Lake Baikal with high topography, high heat flow, widespread Cenozoic rifts, and intense basaltic volcanism since the Miocene, all of which are thought to be associated with a mantle plume [Windley and Allen, 1993]. In this study we are not able to discern structures below 300 km and therefore cannot confirm whether or not the plume exists. However, our result shows evidence for low velocities at the top of mantle in a vast area southwest of Baikal.
 Another large-scale low-velocity zone is centered at the north end of the Okinawa Trough and extends northwestward into eastern China and southeastward into the eastern Philippine Sea. The subduction of the Philippine plate beneath Eurasia reaches to a shallow depth of <300 km and to a small lateral extent. Therefore the low velocities in eastern China cannot be explained by the small-scale convection above subducting slab. The large-scale low velocities in eastern China and the Philippine Sea could be the remaining effect of previous tectonic processes such as the subduction of the Pacific plate and the opening of the Philippine Sea. Nevertheless, the subduction of the Philippine Sea plate underneath Eurasia is still an important contemporary event affecting the tectonics of this region. The Okinawa Trough is a strongly back arc spreading zone in the present world. It is 60–100 km wide in the south and 230 km wide in the north. The northern segment (north of Tokara channel) has a subducting angle different from that of the southern segment. The tear of the slab may allow the asthenosphere beneath the slab to rise. Recent volcanism is more extensive in the northern Okinawa Trough and Kyushu [Letouzey and Kimura, 1986; Sibuet et al., 1998]. From body wave tomography, Sadeghi et al.  also found low-velocity anomalies in the upper mantle at the northeast edge of the Okinawa Trough and west off Kyushu.
 The third low-velocity zone covers the entire Indochina and South China Sea regions. Unfortunately, the path coverage is poor and resolution is low for this region. From the velocity maps above period 70 s we see a prominent low-velocity zone along the Ailao Shan-Red River shear zone. Its center at shorter periods is at the China-Thailand border, and then it moves with increasing period to Hainan Island. This large-scale low-velocity zone includes extensive Cenozoic extensional rift systems in both Indochina and the South China Sea attributed to collision-induced extrusion of lithosphere or to asthenospheric mantle flow [Lee and Lawver, 1994; Morley et al., 2001; Flower et al., 2001]. By waveform inversion, Lebedev and Nolet  found a deep mantle plume beneath Hainan Island.
 Thus, at the normal sublithospheric depth we see a deep continental root in the center of China surrounded on the north, east, and south by three low-velocity anomalies characterizing uplifted asthenosphere of great thickness. These three low-velocity zones belong to three different tectonic domains associated with the doming and rifting of Baikal and the Mongolian plateau, the subduction of the Philippine and Pacific plates, and the extrusion of Indochina (or mantle plume), and are all featured by intensive Cenozoic extension. In the southwest corner of the study region a prominent high-velocity zone is associated with the collision of the Indian and Eurasian plates. This overall pattern is also observed in some global surface wave velocity maps [e.g., van Heijst and Woodhouse, 1999].
4. The 3-D Shear Wave Velocity Structure
 The group velocity maps shown in sections 3.1–3.4 were obtained from a linear inversion based on great circle propagation assumption and neglecting anisotropy and source effect. From these we get the pure path dispersion curves for each grid node in the study region. They constitute a useful data set for constraining the shear wave velocity structure and, at the same time, give some idea for the lateral variation of shear wave velocities in different depth ranges. Although it is generally true that Rayleigh waves of longer period are more affected by deeper structures, the sensitivity kernel of group velocity has a rather complicated relation with the structure. There is no simple correspondence between the group velocity maps and the 3-D shear wave structure.
 From the pure path dispersion curves we carried out inversions for the shear wave velocity structure at each node of the 1° × 1° grid, using a program developed at Saint Louis University. We adopted an isotropic layered Earth model composed of 18–20 layers (0–420 km) overlying a half-space. The S wave velocity in each layer is constant and taken as the inversion parameter, and P velocity and density are calculated from S velocity using empirical formulas. Initial models were constructed by referring to available knowledge of crustal thickness and tectonic type [Chen et al., 2001; Ma, 1989], and the inversion result at a node was often used as the initial model for neighboring nodes. The same initial structure was used below 300 km for all nodes. The resolving power below this depth is very limited, and the resulting structure below 300 km is of little significance.
 It is well known that structure inversion from surface wave dispersion data is plagued with poor resolution and nonuniqueness due to strong nonlinearity, especially when a single surface wave mode is used. In this study we did not put much effort into seeking possible ranges of the true model. Our goal is to get a preliminary 3-D structure, which satisfies the present data set of fundamental group velocity dispersions and shows the essential features of the lithosphere and asthenosphere of various tectonic units in the study region. Figures 6a–6h show the S wave velocity structure along eight latitudes, and Figures 7a–7d show four longitudinal profiles. The contour line intervals are as follows: 0.4 km/s between velocities 2.0 and 3.7 km/s, 0.1 km/s between 3.7 and 4.3 km/s, and 0.07 km/s between 4.3 and 5.0 km/s. In this way the crust will not be crammed by many contour lines. A thick line represents the Moho discontinuity. Notice that the color scale is nonlinear; it is meant to clearly show the structure in the upper mantle. In sections 4.1–4.4 we give a brief account of the structural characteristics of some tectonic units in China.
4.1. Tarim and North China Platform
 Tarim is a stable Proterozoic platform, and the Tarim basin is the largest basin in China with very thick sediments and mild deformation. In the 39°N profile (Figure 6f) the Tarim platform (between ∼76° and 92°E) shows a thick sedimentary layer and relatively thin crust. The lithosphere thickness (∼150 km) and lid velocity are basically normal as for a stable continental platform, and the underlying asthenosphere is not very remarkable in terms of velocity and thickness. These features are also seen on the 85°E longitudinal profile (Figure 7a).
 The north China platform, also called the Sino-Korean platform which in some literatures refers to a larger area including the Tarim platform, has a very different lithosphere-asthenosphere structure as compared with that of the Tarim platform or, especially, the Yangtze platform. On the 39°N profile (Figure 6f) we see a systematic change of the structure: From west to east the lithosphere thins, lid velocity decreases, and the asthenosphere becomes thicker and its velocity decreases. The Alashan and Ordos blocks are between 102° and 111°E. The lithosphere there is ∼120 km thick, and the velocity in lid is relatively high; the asthenosphere, although thick, has relatively higher velocity. East of 111°E the thickness of lithosphere reduces to ∼80 km, and the asthenosphere becomes more conspicuous eastward.
 On the 35°N profile (Figure 6e), to the east of 108°E in Shanxi graben and the north China plain, the lithosphere is 70–80 km thick, and the asthenospheric velocity is very low. On all latitudinal profiles from 31° to 35°N we can see east dipping high velocity beneath the Qinling-Dabie mountain range, probably indicating the remaining effect of northward subduction during the suturing of north and south China.
4.2. Tibet and Surrounding Area
 Our result indicates that a low-velocity zone exists in the middle crust of Tibet. It extends in a vast area approximately between 30°–35°N and 82°–100°E. It is more prominent in the Lhasa and Qiangtang blocks as seen on 30° and 35°N profiles (Figures 6d and 6e) but is also visible near the east edge of the plateau (Figure 7b). A number of researchers also found a crustal low-velocity zone in Tibet but in different depth ranges (for example, in lower crust [Cotte et al., 1999], in middle crust [Yao et al., 1981; Romanowicz, 1982], in midlower crust [Bourjot and Romanowicz, 1992], and in upper crust [Yuan et al., 1997]).
 Several ancient suture zones separate the Tibetan Plateau into structurally different blocks. It is seen on the 85°E profile (Figure 7a) that in the north of the Banggong-Nujiang suture and south of the Kunlun Mountains (approximately between 32° and 36°N) the top part of the mantle shows abnormally low velocities. The contrast between the Lhasa and Qiangtang blocks is obvious by comparing profiles 30° and 35°N (Figures 6d and 6e). This difference between north and south of Banggong-Nujiang suture has been found by a number of research groups with different data and methods. Brandon and Romanowicz  found a “no-lid” zone in the Qiangtang block from pure path phase velocities of Rayleigh wave. From Project International Deep Profiling of Tibet and the Himalaya (INDEPTH) data, Owens and Zandt  pointed out that north of 32°N the upper mantle is anomalous due to high temperature. Rodgers and Schwartz  found evidence for partial melting in the lithosphere of Qiangtang by waveform modeling. Barazangi and Ni  first found inefficient propagation of Sn waves beneath Qiangtang block. McNamara et al.  redefined this Sn-inefficient region as approximately within 32°–36°N and 82°–100°E. In this study we found that the sub-Moho low-velocity zone extends eastward to ∼102°E. It then turns southward along the eastern edge of the Tibetan Plateau and deviates slightly to west, until it connects with a prominent low-velocity zone at the Burma-China border.
 In the south of Banggong-Nujiang suture the thickness of lithosphere exceeds 200 km. The lithosphere plunges northward to 36–37°N (Figure 7a). Beneath the low-velocity anomaly of Qiangtang block a high-velocity zone mimics the delaminated lithosphere [Meisner and Mooney, 1998]. With P-to-S converted waves from INDEPTH data, Kosarev et al.  constructed an Earth model that is similar to our structure profile along 90°E (not shown here). In a recent surface wave study, Shapiro and Ritzwoller  constructed a velocity profile crossing Tibet in N-S direction, which also contains a north dipping high-velocity zone similar to our result. This high-velocity zone could be interpreted as the Indian lid subducting at a shallow angle into the relatively low-velocity upper mantle of Tibet, as indicated by Tapponnier et al.'s [2001, Figure 3] conceptual model for the Cenozoic evolution of Tibet. On the other hand, the structures north of the Banggong-Nujiang suture are quite different from Tapponnier et al.'s model. South dipping subducting slabs, as proposed by them, were not detected in this study; instead, a thick low-velocity zone was found on top of the upper mantle in a vast area of the northern Tibetan Plateau.
 Besides the obvious difference between the north and south of the Banggong-Nujiang suture, there are also visible differences in the structures of western and eastern Tibet. Chung et al.  found that the magmas emplaced in eastern Tibet (east of 92°E) are 40–30 Myr, while those in western Tibet are <20 Myr, implying that the rapid uplift of the western part was 20 Myr later than that of the eastern part. Our result indicates that the velocity structure in eastern Tibet (approximately between 94° and 98°E in Figure 6e), while still showing a low sub-Moho velocity and a detached lithosphere block underneath, has higher velocities than the corresponding part in western Tibet and near the eastern edge of the Tibetan Plateau. The phenomenon can also be deduced from the group velocity maps around 100 s. At the first glance, the high velocity in eastern Tibet seems to tally with its older magmatic age [Chung et al., 1998]. However, the phenomenon calls for detailed geodynamic studies to explain many questions, such as why the older detached slab did not sink to a greater depth. The more detailed 3-D velocity structure of Tibet is presented by Su et al. .
4.3. Yangtze Platform and South China Fold Belt
 The western part of the Yangtze platform, approximately west of 113°E and centered at the Sichuan basin, is a stable continental block. On the 30°N velocity profile (Figure 6d) it is situated between longitudes 104° and 112°E, where the lithosphere is 200 km thick and the asthenosphere is thin. East of 113°E the profile enters the lower Yangtze platform, the lithosphere is only 70–80 km thick, and the asthenosphere is of large scale and low velocity, which presents a sharp contrast to the western Yangtze platform. The velocity structure of the south China fold belt is seen on the 25°N profile (Figure 6c) between 112° and 120°E. Its lithosphere is also 70–80 km thick, with a very thick asthenosphere underneath.
 Comparing the structures of lower Yangtze and south China fold belt to that of the north China plain, we found that except for some minor differences, such as a thicker crust and a lower lid velocity in north China, the overall characteristics of their lithosphere-asthenosphere structures are quite similar. The lithosphere is thin (70–80 km), the asthenosphere is thick, and the transition between lithosphere and asthenosphere is generally marked by a large velocity gradient. Thus, along the east and south coast of China we see a wide belt characterized by eroded lithosphere and lifted asthenosphere. This belt crosses several different tectonic units; therefore the similar structures may imply that they were involved in similar recent tectonic processes, probably the subduction of the Pacific plate.
4.4. Sea Areas East of China
 The Yellow Sea and East China Sea are situated in the east margin of Asia, and they are included in the region of better resolution in this study (Figure 3). In the longitudinal profile of 125°E (Figure 7d) the southern end of Ryukyu Islands and Okinawa Trough is at 25°–26°N, where the velocity beneath Moho is very low. North of that in the East China Sea and Yellow Sea the structure varies slightly along N-S direction until ∼44°N, and farther north the lithosphere becomes thicker. In E-W direction the structure varies more rapidly in the East China Sea owing to the subduction of the Philippine plate. In Figure 6d we see that the lithosphere of the continental shelf thins eastward, the lid is missing under the Ryukyu Islands and Okinawa Trough, and the subducted slab is generally discernible.
5.1. Resolution of the Velocity Image
 In section 2.5 we estimated that the lateral resolution is ∼4°–6° in the area of good path coverage, that is, between 20° and 45°N, west of Japan and the Ryukyu Islands, and east of Pamir and the Himalayas. By scrutinizing the resulting velocity images we believe that this estimation is basically correct. In shallower depths some small-scale geological features, such as sedimentary basins, island arcs, the Okinawa Trough, and some mountain ranges, are clearly manifested. In greater depths, not only do the first-order tectonic provinces all bear distinctive signature in the lithosphere-asthenosphere structures, but the boundaries of some secondary tectonic blocks are also clearly seen in the 3-D structure. For example, the boundary between the Lhasa block and the Qiangtang block in Tibet and the boundary between Ordos and Shanxi graben in north China show deep structural differences. These tectonic boundaries can be defined with small ambiguities generally <2° and tally with the surface geological lines quite well. In the velocity profiles 5° apart from each other (e.g., Figure 6), significant change in the structures can be seen from one profile to another.
5.2. A Comparison With Previous Studies
 Several surface wave tomographic studies have been carried out which covered similar or larger areas as compared with our study region [Curtis et al., 1998; Ritzwoller and Levshin, 1998; Song et al., 1993a, 1993b; Wu et al., 1997]. Some earlier works used much smaller data sets and studied a narrower period range. It makes the comparison with our result difficult, although some consensus exists concerning the general features of large tectonic blocks, such as Tarim, Tibet, north China, and the Yangtze craton. Here we compare our result to that of Ritzwoller and Levshin  because they studied surface wave group velocities between 20 and 200 s period and their study region included ours.
Ritzwoller and Levshin  provided group velocity maps of fundamental Rayleigh waves from 20 to 200 s. By visual comparison we see a good resemblance between the two sets of velocity maps up to 100 s. For example, in their 100 s map the region from India to west Mongolia shows alternatively high and low velocities, whereas the eastern China region shows alternatively low and high velocities from the South China Sea northward and the East China Sea and Japan Sea show prominent low velocities. Our result agrees with this overall pattern quite well. There are, of course, differences regarding the scope, magnitude, and boundaries of these velocity anomalies, but we do not see a contradictory anomaly sense except in some less significant small areas. In the 150 and 200 s velocity maps we see increasingly more discrepancy between the two results. This may be expected because at long periods the path number reduces significantly and measurement error tends to be larger due to low signal-to-noise ratio.
 We also carried out a Love wave tomography for the same region with the same analysis and inversion methods but by a different person. The resulting lithosphere-asthenosphere structure is in agreement with that from Rayleigh wave data. The differences in velocity can be attributed either to measurement error or to polarization anisotropy. Only in a few small regions do the two results indicate significantly different structure; for example, beneath the Korean peninsula the Love wave data produced a peculiar upper mantle with a remarkable high velocity below 100 km which is not seen in the Rayleigh wave result.
5.3. Lithospheric Tectonic Boundaries
 China is composed of a number of continental blocks with fold belts or suture zones between them. Many important lithospheric boundaries seen in the 3-D structure are consistent with the ancient suture zones or fold belts bordered on different tectonic provinces. For example, the Kunlun Mountains in western China (at 36°–37°N in Figures 7a and 7b) and the Qinling-Dabie Mountains in eastern China (at 34°N in Figure 7c) mark the important boundary between north and south China. The former separates the main body of the Tibetan Plateau from the Tarim and Qaidam blocks, and the latter separates the north China platform from the Yangtze platform.
 On the other hand, there are some large-scale structural features which cross several tectonic provinces and seem to be related with more recent or contemporary tectonic processes. In the coastal area and continental shelves of eastern China the upper mantle is characterized by thinned lithosphere and large-scale low-velocity asthenosphere, likely caused by the subduction of the Pacific and Philippine plates. Along the eastern edge of the Tibetan Plateau (western Sichuan and Yunnan area), there is an N-S low-velocity channel in the top of upper mantle. Its north end is connected with the low-velocity zone in northern Tibet, and its south end joins with the conspicuous low-velocity zone in Burma and Indochina.
 The collision between India and Asia produced the highest plateau in the world and has a far-reaching effect on the recent tectonics of eastern Asia. It was proposed that the collision caused the extrusion of Indochina, the pull-apart extension in the north China platform, and even the rifting in the Baikal region. In our structure model these areas are all characterized with a thinned lithosphere and thick asthenosphere. It is hard to tell whether the rising of asthenosphere is a result of extension or a condition facilitating the extension because we only see a current picture. For a further understanding of the tectonic and structural evolution it is necessary to carry out geodynamic modeling incorporating available geological and geophysical information.
5.4. Interpretation of Structural Complexities
 In stable continental areas, such as the north India, Tarim, and western Yangtze platforms, the upper mantle structure is rather simple with a lithosphere overlying a low-velocity asthenosphere. In tectonically active areas the top of the asthenosphere is lifted up to much shallower depths and even reaches to the bottom of crust in some places. Some of such sub-Moho low-velocity zones are related to island arc and back arc spreading zones (Figures 6d, 6e, and 7d) or to continental rifting areas (Figures 6h and 7b). Some active volcanic areas, such as Tengchong near the China-Burma border and Changba Shan near the China-Korea border, are also associated with sub-Moho low-velocity anomalies.
 In these tectonically active areas the upper mantle structure is rather complicated. Embedded high-velocity bodies in asthenosphere, invaded low-velocity zones in lithosphere, and intercalated high- and low-velocity layers are commonly seen in the upper mantle of these areas. Some high-velocity bodies can be reasonably related to subducted or delaminated slabs, such as beneath the Okinawa Trough and in Tibet, but some are not so easily explained. Especially, the seemingly double lithosphere-asthenospheres seen on many velocity profiles are perplexing. They are generally more obvious in sea areas and higher latitudes where we do not have adequate path coverage, and the measured dispersion at long periods may have large errors. We shall leave this problem to future studies.
 The present study is founded on a ray theoretical basis valid only for laterally homogeneous or smoothly varying structures. The 3-D velocity structure resulted from the inversion, resolved the major features of various tectonic regions, and bore a wholesale reasonability. It may be used in future studies to construct time-variable or phase-matched filters for isolating modes and improving group or phase measurements. Within the same theoretical frame we feel that more work can be done to improve the present result. Besides expanding the present database, several things are in progress in an effort to enhance the resolution and credibility of the result, including introducing anisotropy into the model and utilizing more data, such as phase velocities and polarization anomalies, to better constrain the model.
The Occam's inversion method is robust and allows us to use a fine grid in model parameterization. This is favorable for more accurately defining the boundaries between different tectonic blocks and hence facilitates the interpretation of the resulting structure.
We have obtained group velocity maps of fundamental Rayleigh waves between 10 and 184 s in the study region. The lateral resolution is estimated to be 4°–6° within 20°–45°N west of Japan and the Ryukyu Islands and east of Pamir and the Himalayas. With increasing period the group velocity distribution pattern changes systematically and correlates well with the known geological and tectonic features in the study region.
A preliminary 3-D S wave velocity structure is obtained from the pure path group velocity dispersion curves. The velocity model indicates that a tripod of high-velocity lid constitutes the frame of lithosphere in the study region. The core of Yangtze craton, like a continental keel, reaches down to 200 km depth. Another leg of the tripod extends westward to Tarim basin, where the lithosphere is 150–200 km thick. The third leg extends northeasterly along the Inner Mongolia-Great Xinan fold belt, where the lithosphere is 100 km thick with a higher lid velocity. This tripod separates three different active tectonic domains. To its north, there are the west Mongolia and Baikal rifting areas. To the east in the coastal area and continental shelves the thin lithosphere (70–80 km) and remarkably thick asthenosphere are apparently related to the subduction of the Pacific and Philippine plates. In the southwest the collision between India and Asia resulted in a very thick crust in Tibet, and beneath that a north dipping high-velocity zone, mimicking a subducting plate, reached down to more than 200 km depth. In northern Tibet a low-velocity zone immediately below the Moho extends eastward then turns southward along the eastern edge of the plateau until it connects to the vast low-velocity area in Indochina and the South China Sea.
 We are grateful to the data management centers of CDSN in China, IRIS in the United States, Geoscope in France, and FREESIA in Japan for providing the data. We thank Zongjin Ma, Futian Liu, Jianhua Liu, and Hui Chen for their valuable help in the work. This work was supported by the Natural Science Foundation of China (49834020 and 40034010) and the Hi-tech Development Project of China (8200104). We thank M. Ritzwoller, an anonymous reviewer, and the editors for their constructive comments and suggestions.