Crustal heterogeneity and seismotectonics of the region around Beijing, China

Crustal heterogeneity and seismotectonics of the region around Beijing, China

Tectonophysics 385 (2004) 159 – 180 www.elsevier.com/locate/tecto Crustal heterogeneity and seismotectonics of the region around Beijing, China Jinli...

2MB Sizes 1 Downloads 67 Views

Tectonophysics 385 (2004) 159 – 180 www.elsevier.com/locate/tecto

Crustal heterogeneity and seismotectonics of the region around Beijing, China Jinli Huang a, Dapeng Zhao b,* a

Center for Analysis and Prediction, China Seismological Bureau, Beijing 100036, China b Geodynamics Research Center, Ehime University, Matsuyama 790-8577, Japan Received 20 January 2004; accepted 23 April 2004 Available online 2 July 2004

Abstract A detailed three-dimensional (3-D) P-wave velocity model of the crust and uppermost mantle under the Chinese capital (Beijing) region is determined with a spatial resolution of 25 km in the horizontal direction and 4 – 17 km in depth. We used 48,750 precise P-wave arrival times from 2973 events of local crustal earthquakes, controlled seismic explosions and quarry blasts. These events were recorded by a new digital seismic network consisting of 101 seismic stations equipped with high-sensitivity seismometers. The data are analyzed by using a 3-D seismic tomography method. Our tomographic model provides new insights into the geological structure and tectonics of the region, such as the lithological variations and large fault zones across the major geological terranes like the North China Basin, the Taihangshan and the Yanshan mountainous areas. The velocity images of the upper crust reflect well the surface geological and topographic features. In the North China Basin, the depression and uplift areas are imaged as slow and fast velocities, respectively. The Taihangshan and Yanshan mountainous regions are generally imaged as broad high-velocity zones, while the Quaternary intermountain basins show up as small low-velocity anomalies. Velocity changes are visible across some of the large fault zones. Large crustal earthquakes, such as the 1976 Tangshan earthquake (M = 7.8) and the 1679 Sanhe earthquake (M = 8.0), generally occurred in high-velocity areas in the upper to middle crust. In the lower crust to the uppermost mantle under the source zones of the large earthquakes, however, low-velocity and high-conductivity anomalies exist, which are considered to be associated with fluids. The fluids in the lower crust may cause the weakening of the seismogenic layer in the upper and middle crust and thus contribute to the initiation of the large crustal earthquakes. D 2004 Elsevier B.V. All rights reserved. Keywords: Seismic tomography; P-wave velocity; Crustal structure; Continental earthquakes; Seismotectonics

1. Introduction The Chinese capital, Beijing, is located in a seismically very active region in Northern China (Fig. 1). * Corresponding author. E-mail addresses: [email protected] (J. Huang), [email protected] (D. Zhao). 0040-1951/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.tecto.2004.04.024

According to Gu (1983), more than 100 earthquakes with magnitude (M) greater than 5.0 have occurred in this region during the last 1000 years, among them 35 events are larger than M = 6.0 and 7 events are larger than M = 7.0. In 1679, an earthquake of M = 8.0 occurred in the Sanhe county of Beijing (Fig. 1), which is the largest one among the known historical earthquakes in this region. The 1976 Tangshan earth-

160

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

Fig. 1. Map showing the major geographical features of the study area. The color shows the surface topography. White curved lines show major active faults, grey curved lines show the boundaries between provinces. White stars denote earthquakes with magnitudes (M) equal to or greater than 6.5 that occurred in the study area since BC 780. Blue dots show the earthquakes in the same period with M = 5.0 – 6.5. The earthquake magnitude scale is shown above the map. The insert map shows the location of the present study area. (1) Cangdong Fault, (2) Taihangshan Front Fault, (3) Zijinguan Fault, (4) Weixian – Yanqing Fault, (5) Xiadian Fault, (6) Tangshan Fault, (7) West Luanxian Fault and (8) Ninghe Fault.

quake (M = 7.8) occurred approximately 160 km southeast of Beijing, which totally destroyed the Tangshan city (then population 1 million) and killed about 240,000 people. It was perhaps the most destructive earthquake in the world in the human history.

In the present study region (Fig. 1), there are several large cities (e.g., Beijing, Tianjin, Tangshan, and Shijiazhuang) in addition to numerous towns and villages. The total population amounts to 80 millions. A detailed investigation of the crustal structure and

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

seismotectonics of this region is very important for the understanding of physics of continental earthquakes and for the assessment and mitigation of seismic hazard. From a geological point of view, the present study area is located in the intersection of the Yanshan and Taihangshan uplift regions (Fig. 1). The central portion of the study area is called North China Basin, which is a large epicontinental basin and is characterized by alternate uplift and depression zones (Li, 1981; Ye et al., 1985, 1987; Liu, 1987) (Fig. 2). The northeastern portion is the relatively stable Yanshan uplift region with its major structure and tectonic trend oriented in E – W direction. The western and northwestern portions are the Taihangshan uplift region with some small intermountain basins. Many active faults that are oriented in NE –SW direction exist in the North China Basin and the Taihangshan

161

uplift region. In the two regions, all the structure and mountain ranges have trends oriented in the NE – SW direction. In the present study area, earthquakes are concentrated in four seismic zones. The Zhangjiakou– Penglai seismic zone, oriented in NW –SE direction, is the most active one that contains a majority of large earthquakes in the study area (Fig. 1). The other three seismic zones (Tangshan –Xintai, Sanhe – Linshou, and Huailai –Weixian) are all oriented in NE – SW direction and are generally parallel to each other (Fig. 1). Many researchers have investigated the three-dimensional (3-D) seismic velocity structure of the crust and upper mantle under this region using arrival times from local and/or teleseismic events (Jin et al., 1980; Liu et al., 1986; Shedlock and Roecker, 1987; Zhu and Zeng, 1990; Sun and Liu, 1995; Yu et al., 2003). The data sets used by these previous studies were

Fig. 2. Tectonic map of the study area (after Research Group for the 1976 Tangshan Earthquake, 1982). The legend is shown on the right. (1) Major faults. Dashed lines show the deduced faults. (2) Boundary of Cenozoic basins, (3) Uplift areas in North China Basin, (4) Granitic areas, (5) Pre-Cambrian basement, (6) Coast line and (7) Cities.

162

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

recorded by a sparse seismic network with a small number of stations, few of which were digital seismic stations. Hence, the arrival times were not precise enough and the tomographic images they determined have a lower resolution. Some deep seismic soundings (DSS) were also conducted to investigate the crustal structure in this region (China Seismological Bureau, 1986; Zhang, 1998; Li et al., 2001). These studies revealed strong lateral heterogeneities in the velocity structure of the crust and uppermost mantle as well as significant lateral variations in the crustal thickness under the Chinese capital region. Other geophysical methods such as magnetotelluric, gravity and aeromagnetic soundings are also used (Feng and Zheng, 1987; Liu et al., 1989). During the last a few years, a new digital seismic network with 101 stations has been installed in the Chinese capital region (Fig. 3). It is the most advanced regional digital seismic network in main land China, which covers Beijing and the surrounding areas densely and uniformly. A large number of highquality arrival times have been recorded, which provides an unprecedented opportunity to determine a detailed 3-D crustal structure under the region. In the present work, we have applied a tomographic method to a large data set of local earthquake arrival times to determine a high-resolution 3-D P-wave velocity

structure of the crust and uppermost mantle under this region. Compared with the previous tomographic studies, we have used a larger quantity and higher quality data set and applied an updated tomographic method to take into account the effects of the complex Moho geometry in this region. Our results cast new light on the complex structure and seismotectonics of the Chinese capital region.

2. Data and method In this work, we have used arrival time data recorded by the digital Capital Seismic Network (CSN) which has been in operation since 2001. Over 150,000 arrival times from over 9000 events have been accumulated, which include the direct and head waves (Pg, Sg, Pn, Sn) from local crustal earthquakes and quarry blasts. We used four sets of data in this study (Table 1), which were selected with the following criteria: (1) The first data set contains arrival times from seven seismic explosions which were used to test the CSN operation performance. These seven events have precise hypocenter locations and origin times and were recorded by many CSN stations. The average number of P arrival times per event is 53.

Fig. 3. Distribution of seismic stations used in this study. The E’s denote new digital seismic stations. The 5’s denote stations of the Huabei Seismic Network.

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180 Table 1 Data sets used in this study Data set Data set Data set Data set Data set

Event type 1 2

controlled shots local earthquakes blasts

No. of events

No. of data

7

368

1854

30,164

relocated

864

15,606

248

2612

focal depth fixed to 0 km relocated

3 4

local earthquakes

Hypocenter parameters fixed

(2) For local earthquakes recorded by CSN, every event has at least 10 arrival times and a small formal uncertainty ( < 5 km) for the hypocenter locations. To collect a best set of the earthquake data, we adopted the following way. The study area is divided into cubic blocks with a spatial size of 5  5  1 km. Among the earthquakes within each block, we select only one event which has the greatest number of first P-wave arrivals and the smallest uncertainty for the hypocentral location. These are the second set of our data. Most of the events have absolute accuracy of hypocentral locations better than 5 km because they satisfy the earthquake location criteria with a local seismic network proposed by Bondar et al. (2004). (3) We also used quarry blasts which were located on the surface; every event has at least 12 P arrivals and a smaller uncertainty ( < 4 km) for the epicenter locations. We used the same approach as in (2) to select a best set of the quarry blast data. (4) In areas where few events are located, e.g., the North China Basin, we have added some local earthquakes recorded by the Huabei Seismic Network (HSN) during 1990 –1999. Every event has at least eight P arrivals and mislocation errors smaller than 5 km. In total, we have collected 48,750 P arrival times from 2973 events (Fig. 4). These events were recorded by 141 stations (Fig. 3), among which 101 stations belong to CSN. The remaining 40 stations belong to HSN. Note that 23 HSN stations are located in the same sites as the CSN stations (see Fig. 3). Most of the events have more than 20 P arrivals, some events have more

163

than 80 P arrivals. The picking accuracy of the P arrival times is 0.05 –0.15 S. To analyze the arrival time data, we have used the tomographic method of Zhao et al. (1992). This method is adaptable to a velocity structure which includes several complex-shaped velocity discontinuities and allows 3-D velocity variations everywhere in the model. The discontinuities represent known geological boundaries, like the Moho discontinuity and/or a subducting slab boundary, etc. A 3-D grid net is set up in the model to express the 3-D structure. Velocity perturbations at the grid nodes are taken as unknown parameters. The velocity perturbation at any point in the model is calculated by linearly interpolating the velocity perturbations at the eight grid nodes surrounding that point. To calculate travel times and ray paths accurately and rapidly, an efficient 3-D ray-tracing technique (Zhao et al., 1992) is employed that iteratively uses the pseudobending technique (Um and Thurber, 1987) and Snell’s law. Station elevations are taken into account in the ray tracing. The LSQR algorithm (Paige and Saunders, 1982) with a damping regularization is used to solve the large and sparse system of observation equations, allowing a great number of data to be used to solve a large tomographic problem. The nonlinear tomographic problem is solved by iteratively conducting linear inversions. In each iteration, perturbations to hypocentral parameters and velocity structure are determined simultaneously. A detailed description of the method is given by Zhao et al. (1992, 1994) and Zhao (2001).

3. Analysis We examined a number of 1-D velocity models for the present study area proposed by the previous studies (Teng et al., 1979; China Seismological Bureau, 1986; Li et al., 2001) using the high-quality arrival time data from the seven seismic explosions (controlled shots) whose locations and origin times are known (see Table 1). Finally, we selected the 1-D model as shown in Fig. 5 which gave the best fit to the explosion data. The Pwave velocity of the upper crust (0 –10 km depth) has a large vertical gradient, which is well constrained by the DSS studies (China Seismological Bureau, 1986). We used this 1-D model as the starting velocity model for our tomographic inversions.

164

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

Fig. 4. (a) Epicentral distribution of earthquakes used in this study. Crosses denote 1854 events recorded by the Capital Seismic Network (CSN). o’s denote 248 events recorded by the Huabei Seismic Network. (b) Distribution of 864 quarry blasts (crosses) and 7 controlled shots (stars) whose origin times and epicentral locations are known. These 871 events were recorded by CSN.

Previous studies have revealed the existence of the Conrad and Moho discontinuities in the study area as well as their significant lateral depth varia-

tions (China Seismological Bureau, 1986; Zhang, 1998; Li et al., 2001). Fig. 6 shows the geometry of the two discontinuities we compiled by referring

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

Fig. 5. The starting 1-D velocity model used for the tomographic inversion.

to the previous results. The Moho and Conrad have a similar pattern of depth variations. They are shallow under the Bohai Bay in the southeastern part of the study area and become deep toward mountainous areas to the northwest. The Moho depth ranges from 30 to 42 km, while the Conrad depth changes from 16 to 24 km (Fig. 6). After resolution analyses, we adopted a grid with a grid spacing of 25 km for the central portion where Beijing, Tianjin and Tangshan are located, and 50 km at the edges in the horizontal direction, and 4 – 17 km in depth (Fig. 7). We have conducted many inversions using different values of damping parameter (Fig. 8). We found the best value of the damping parameter to be 10.0 considering the balance between the reduction of travel time residuals and the smoothness of the 3-D velocity model obtained (EberhartPhillips, 1986). Convergent solutions were obtained for all the inversions. For the inversion with a damping parameter of 10.0, the root mean square (RMS) travel time residual was reduced from 0.681 to 0.402 s after three iterations. The variance reduction is 65%. Fig. 9 shows the change in the distribution of travel time residuals before and after the inversion.

165

Over 80% of the rays have residuals smaller than 0.4 s after the inversion (Fig. 9). The distribution of hit counts (number of rays passing through each grid node) for every layers is shown in Fig. 10. Most parts of the study area are well sampled by the rays. Fig. 11 shows the obtained 3-D P-wave velocity structure together with active faults and moderate to large earthquakes since BC 780. Fig. 13 shows four vertical cross sections of the velocity images along the profiles denoted in Fig. 12. The tomographic images are shown in areas with hit counts greater than 10. We also conducted inversions using only the Data Set 2 and Data Set 3 as shown in Table 1. For these inversions, we adopted a grid with a lateral grid spacing of 50 km. The focal depth is fixed to zero for all the quarry blasts. The inversion results show the near-surface geological features very well. The results are well consistent with that shown in Figs. 11 –13, although the images for the deep crust have a lower resolution. These results suggest that the identification of the quarry blasts is reliable and the arrival time data from the seismic explosions and blasts are very precise. We also conducted an inversion with a flat geometry of the Conrad and the Moho discontinuities (Fig. 14). The Conrad and the Moho depths are 20 and 35 km, which are the average depths of the two discontinuities in the study area (Fig. 6). Comparing Fig. 14 with Fig. 13, we can see that the overall patterns of the images are similar, but there are considerable changes in the amplitude of the velocity anomalies in the lower crust. The final RMS travel time residual is 0.416 s for the inversion with the flat Conrad and the Moho, which is about 4% higher than the RMS residual (0.402 s) of the inversion with the topography of the two discontinuities considered. We found this amount of residual reduction is statistically significant after statistic analyses. Hence, we prefer the results shown in Figs. 11– 13. The importance of taking into account the depth variations of the Moho and other discontinuities in the tomographic inversion has also been demonstrated in the earlier studies of the Japan and Tonga subduction zones (Zhao et al., 1992, 1997a), Southern Carpathians, Romania (Fan et al., 1998), and Southwest China (Huang et al., 2002). When the discontinuity topography is taken into account, ray paths and travel

166

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

Fig. 6. Depth distribution of (a) the Conrad and (b) the Moho discontinuities in the present study area which were constructed in this study by referring to the previous studies using gravity data and deep seismic soundings (Li et al., 2001; Zhang, 1998; China Seismological Bureau, 1986). The Conrad and Moho depths are shown in contours and grey scale. The darker areas denote deeper Conrad or Moho. The grey scales are shown at the bottom of each map.

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

167

Fig. 7. Three-dimensional configuration of the grid net adopted in the present study.

times can be computed more accurately, and the degree of nonlinearity of the tomographic problem is reduced. Thus, a better tomographic result is expected. Before describing the obtained tomographic results, we first evaluate the resolution of the tomo-

graphic image. A direct way to evaluate the resolution of a tomographic result is to calculate a set of travel time delays that result from tracing the corresponding rays through a synthetic structure as though they are data, and then to compare the inversion result with the initial synthetic structure. In this study we adopt the

Fig. 8. Trade-off curve for the variance of the velocity perturbations and root mean square travel time residuals. Numbers beside the gray circles denote the damping parameters adopted for the inversions. The black circle denotes the optimal damping parameter for the final tomographic model.

168

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

Fig. 9. Number of rays vs. travel time residuals. The dashed line denotes the result for the starting model (before the inversion). The solid line denotes the result after the inversion.

checkerboard resolution test (Zhao et al., 1992) to assess the adequacy of the ray coverage and to evaluate the resolution. To make a checkerboard, positive and negative 3% velocity perturbations are assigned to 3-D grid nodes that are arranged in the model space, the image of which is straightforward and easy to remember. Therefore, by just seeing the image of the synthetic inversion of the checkerboard, one can easily understand where the resolution is good and where it is poor. Random errors in a normal distribution with a standard deviation of 0.1 s are added to the theoretical travel times calculated for the synthetic models. Leveque et al. (1993) showed that in some cases small-sized structures in a checkerboard test can be retrieved effectively while larger structures are poorly retrieved. In order to know whether there is such a problem in our tomographic images, we conducted checkerboard tests with different grid spacings. Fig. 15 shows the result of a checkerboard resolution test with a grid separation of about 25 km. The resolution is generally high in Beijing, Tianjin and Tangshan areas in each of the depth slices. In the southern part of the study area, the resolution is lower in the middle to lower crust layers, as is expected from the distribution of hit counts (Fig. 10). Fig. 16 shows the results of another checkerboard resolution test with a grid separation of 50 km. We can see that the resolution is quite high and the checkerboard pattern

and anomaly amplitude are correctly reconstructed for most of the study area. From these resolution tests, we can say that the tomographic images obtained in this study have a spatial resolution of 25 –50 km in the horizontal direction and 4– 17 km in depth, and largescale structures are also resolved.

4. Results Strong P-wave velocity variations of up to 6% are found in the study area, indicating the existence of significant structural heterogeneities in the crust and uppermost mantle in this region. In the shallow depth slices (Fig. 11a and b), we can see that velocity images in North China Basin, Yanshan, and Taihangshan regions exhibit different patterns. In the North China Basin, two low-velocity (low-V) zones and one high-velocity (high-V) zone exist and are oriented in the NE –SW direction. The two low-V zones correspond to the Huanghua depression and Jizhong depression (see also Fig. 2), while the high-V zone sandwiched by the two low-V zones corresponds to the Cangxian uplift. A low-V zone is visible under the Jizhong area in our tomographic image at 10-km depth (Fig. 11c). This feature is consistent with an analysis of seismic converted waves which revealed that the sedimentary layer in the Jizhong depression is as thick as 12 km (Shao et al., 1980).

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

169

Fig. 10. Distribution of the number of the rays passing through each grid node (hit counts). The depth of the layer is shown at the lower right corner of each map. The hit count scale is shown at the bottom.

170

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

Fig. 11. P-wave velocity image at each depth slice (in percent from the average velocity). The depth of each layer is shown at the lower right corner of each map. Red and blue colors denote low and high velocities, respectively. White circles show earthquakes with magnitude equal to or greater than M = 6.0 that occurred from BC 780 to 1998 in the study area. The velocity perturbation and earthquake magnitude scales are shown at the bottom. Other labels are the same as those in Fig. 1.

In the Taihangshan and Yanshan mountainous regions, Paleozoic strata and Pre-Cambrian basement rocks outcrop widely on the surface (Fig. 2), which

exhibit strong and broad high-V anomalies in our tomographic images (Fig. 11a and b). The Quaternary intermountain basins such as Huailai basin and Weix-

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

171

Fig. 12. Locations of the vertical cross sections shown in Figs. 13 and 14.

ian basin (Fig. 1) show up as small low-V anomalies in the broad high-V zones for the uplift regions. Although the Tangshan area is located in the North China Plain, it is an uplifted block beside the Zhangjiakou – Penglai belt (Gao and Ma, 1993); hence, it shows up as high-V anomaly (Fig. 11a – c). Our tomographic images show that in the Yanshan uplift area, the trend of the velocity anomalies is oriented in E – W direction, while in North China Basin and Taihangshan uplift areas, the trend of the velocity anomalies is oriented in NE – SW direction. These results are well consistent with the trend of active fault zones and other surface geological features. As a whole, the velocity images of the shallow layers reflect well the surface geology, topography, and lithology. At 10 km depth, low-V zones are visible northeast of Beijing, north of Weixian (Yangyuan basin) and in the Huailai basin (Fig. 11c). Along the Taihangshan Front Fault, intermittent slow anomalies are visible. These results are well consistent with that obtained by a joint inversion of the DSS data from multiple profiles (Li et al., 2001). Yu et al. (2003) used only direct P wave data and obtained the 3-D velocity structure in the upper to middle crust. In this study we have obtained the

velocity images of the whole crust and the uppermost mantle (Fig. 11f) using Pn waves refracted from the Moho discontinuity in addition to direct P waves. The Pn velocity image has a high resolution in the whole study area (Figs. 15f and 16f). The Pn velocity is low under the Taihangshan mountainous region, but it is high under most parts of the Yanshan mountainous region and the North China Basin (Fig. 11f). This result is generally consistent with the recent Pn tomography (Wang et al., 2003) and a gravity investigation (Gao and Ma, 1993). Our tomographic image (Fig. 13a) also agrees with the DSS result along the Beijing –Huailai –Fengzhen profile (Zhu et al., 1997; Li et al., 2001). Velocity changes are visible across some of the fault zones such as the Taihangshan Front Fault. Such a feature is even visible in the uppermost mantle (Fig. 11f), suggesting that some of the faults may have cut through the crust and reached to the upper mantle. But no velocity contrast is visible across most of the faults, particularly in the middle to lower crust, suggesting that most of the faults may be just a shallow feature in the upper crust. It is also possible that our tomographic model has insufficient spatial resolution and so the fault zones are not imaged.

172

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

In Fig. 13 we also show the hypocenters of some earthquakes whose focal depths are determined reliably (Gu, 1983). Most of the events occurred in the upper and middle portions of the crust with focal depth shallower than about 25 km. All the large earthquakes with M z 6.5 are located in highV areas which are adjacent to or surrounded by low-V anomalies. Most of the smaller earthquakes are also located in high-V areas, although some are in low-V areas or at the boundary between the slow and fast areas.

5. Discussion Early tomographic studies for this region used only a small number of teleseismic data or regional and teleseismic data together (Jin et al., 1980; Shedlock and Roecker, 1987; Zhu and Zeng, 1990), and so their tomographic results have a lower resolution than the present model. Sun and Liu (1995) showed their velocity images only at 13and 15-km depths. Their image at 13-km depth reflects the general features of the surface geology and is in general agreement with the present result. Recently, Yu et al. (2003) applied the method of Thurber (1983) to the direct P arrival times and determined P-wave velocity images down to 25-km depth. They did not use the Moho head waves (Pn) as in this study. Their results show major velocity anomalies oriented in the E – W and N – S directions, which are very different from the present results and the major geological and tectonic tend in this region (Figs. 1 and 2). 5.1. Large earthquakes and crustal heterogeneity Fig. 11 shows our tomographic images together with epicentral locations of all the large historic earthquakes (M z 6.0) occurred in this region since BC 780. This region has a long history of civili-

173

zation and so has the most detailed earthquake records in China. Although we do not know the accurate focal depths of the historic strong earthquakes, a statistic analysis of the modern seismicity suggested that most of the earthquakes occur in the middle crust under this region, mainly in 10 – 15-km depth (Zang and Yang, 1984). In the tomographic images of 10- and 15-km depths, we can see that most of the large earthquakes are located in high-V zones and close to the boundary between the low-V and high-V anomalies. This feature has been pointed out by many previous researchers from the early tomographic results (Jin et al., 1980; Liu et al., 1986; Zhu and Zeng, 1990). Under the epicentral areas of the 1976 Tangshan and the 1679 Sanhe earthquakes, high-V zones exist in the middle crust, while very prominent and broad low-V zones exist in the lower crust (Fig. 13b – d). Zhao et al. (2002) found that large crustal earthquakes (M = 5.7– 8.0, depth 0 – 20 km) from 1885 to 1999 in Japan occurred in or around low-V zones or the boundary areas between low-V and high-V zones revealed by seismic tomography. They suggested that the low-V zones represent weak sections of the seismogenic crust, which are caused by magma chambers or fluids resulting from the dehydration of the subducting oceanic slab. Hauksson and Haase (1997) found that four M>5.9 earthquakes in the Los Angeles basin area occurred in or adjacent to high-velocity zones. They interpreted that the high-velocity zones form the upper block of a thrust fault or a thin-skinned structure; that is, the earthquake ruptures actually happened at the boundary between high- and lowvelocity zones as shown in Fig. 13 of Hauksson and Haase (1997). Recently, Huang et al. (2002) found that large fault zones and most of the large crustal earthquakes (in particular, those with M>5.0) in Southwest China are located in the boundary areas between high-V and lowV zones in the crust and some of them occurred above the low-V zones of the lower crust to uppermost

Fig. 13. Vertical cross sections of P-wave velocity perturbations (in percent from the average velocity at each depth) when the Conrad and Moho depth variations (Fig. 6) are taken into account. Red and blue colors denote low and high velocities, respectively. The surface topography along each profile is shown on the top of each cross section. White stars and open circles show the earthquakes that occurred within a 25-km width from each profile. The earthquake magnitude scale and the velocity perturbation scale are shown at the bottom and on the right, respectively. The thin and thick black curved lines denote the Conrad and the Moho discontinuities, respectively. Tianzhen (TZ), Huailai (HL), Changping (CP), Shunyi (SY), Beijing (BJ), Sanhe (SH), Ninghe (NH), Tangshan (TS), Hejian (HJ), Cangzhou (CZ), Luanxian (LX).

174

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

Fig. 14. The same as Fig. 13 but for the inversion result when flat Conrad and Moho are used. The Conrad and Moho depths are taken to be 20 and 35 km, respectively.

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

175

Fig. 15. Results of checkerboard resolution test when the grid net shown in Fig. 7 was adopted. The depth of the layer is shown at the lower right corner of each map. and o denote high and low velocities, respectively. The velocity perturbation scale is shown at the bottom.

.

mantle. Since the low-V zones also exhibit low electric resistivity, negative gravity anomaly, and high heat flow, they may represent high-temperature anomalies or magma chambers (under the Tengchong volcanic area), or fluid reservoirs, which may cause the weak-

ening of the seismogenic layer in the upper and middle crust (Zhao et al., 2002). The weak sections of the seismogenic crust are subject to the tectonic stress and hence prone to large crustal earthquakes (Zhao et al., 1996, 2002).

176

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

Fig. 16. The same as Fig. 15 but the grid spacing is 50 km.

5.2. The 1976 Tangshan and the 1679 Sanhe – Penggu earthquakes The 1976 Tangshan earthquake sequence consists of three large earthquakes: (1) the mainshock (M = 7.8)

occurred on July 28, 1976 right beneath the Tangshan city; (2) the largest aftershock (M = 7.1) (also known as Luanxian earthquake) occurred 15 h after the mainshock and was located about 25 km northeast of the main shock hypocenter under Luanxian; and (3)

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

the second largest aftershock (M = 6.9) (also known as Ninghe earthquake) occurred on November 15, 1976 west of Tangshan under Ninghe city (see Figs. 1, 13b and d). The focal depths of the three events are determined to be 11.0, 10.0 and 17.0 km, respectively, by the seismic network (Fig. 13b and d), which were confirmed by a waveform modeling study (Nabelek and Chen, 1987). All the three large earthquakes and the major aftershocks of the Tangshan earthquake sequence occurred in a high-V zone in the upper to middle crust (Fig. 13d). Right beneath the high-V zone there is a prominent low-V zone in the lower crust, and boundary between the slow and fast anomalies becomes shallower from southwest to northeast under the three big earthquakes (star symbols in Fig. 13d). Their hypocenters also become shallow from 17.0 km (Ninghe earthquake) in southwest to 11.0 km (Tangshan earthquake) and to 10.0 km (Luanxian earthquake) in northeast (Fig. 13d). Fig. 13b shows a cross section along the profile BB’ (Fig. 12) that passes through the hypocenter areas of the Tangshan and Ninghe earthquakes. The same feature shown in Fig. 13d is visible in Fig. 13b. Liu et al. (1989) determined the geoelectric structure in the Tangshan area using magnetotelluric soundings and revealed a high-conductivity anomaly in the lower crust deeper than 24 km under the epicenter of the Tangshan earthquake. Their result is in good agreement with the present tomographic result. We believe that the high-conductivity and low-velocity anomaly in the lower crust under Tangshan shows the existence of fluids in the earthquake source region. Similar lowvelocity and high-conductivity anomalies are also found in the source areas of the 1995 Kobe earthquake in Japan and the 2001 Bhuj earthquake in India (Zhao et al., 1996; Mishra and Zhao, 2003). On September 2, 1679, an M = 8.0 earthquake occurred in the Sanhe area, which is only 45 km east of Beijing. Its causative fault is Northeast Xiadian fault (Gu, 1983). The tomographic image at 15-km depth shows that the hypocenter of the Sanhe earthquake is located in a distinctive high-V zone (Fig. 11d). The cross sectional images show that the hypocenter is sandwiched by slow anomalies at 10- and 25-km depths (Fig. 13b and c). The low-V anomaly above the hypocenter in Fig. 13b corresponds to the Xiadian fault and Dachang depression along the fault. The lowV zone in the lower crust under the hypocenter

177

corresponds to a high-conductivity anomaly revealed by a magnetotelluric sounding (China Seismological Bureau, 1986), which may be caused by fluids in the crust. 5.3. Effect of fluids in seismotectonics Low seismic velocity and high-Poisson’s ratio anomalies are revealed in the source area of the 1995 Kobe earthquake (M = 7.2), which are interpreted as fluid-filled, fractured rock matrix that contributed to the initiation of the Kobe earthquake rupture (Zhao et al., 1996; Zhao and Negishi, 1998). Later it was found that the fluids at the Kobe hypocenter are associated with the dehydration of the subducting Philippine Sea slab under Southwest Japan (Zhao et al., 2002; Salah and Zhao, 2003). Similar low velocity and high Poisson’s ratio anomalies are also found in the source area of the 2001 Bhuj earthquake region, which are considered to be fluids that contributed to the rupture nucleation (Kayal et al., 2002; Mishra and Zhao, 2003). Our present results of seismic velocity anomalies in the source zones of the 1976 Tangshan and the 1679 Sanhe – Penggu earthquakes are consistent with the previous findings for the big earthquake source areas in Japan and India, suggesting the existence of fluids and their influence in the earthquake generation in the Beijing region. Fluids widely exist in the crust and uppermost mantle (Hickman et al., 1995; Zhao et al., 2002). The existence of fluids beneath the seismogenic layer affects the long-term structural and compositional evolution of the fault zone, change the strength of the fault zone, and alter the local stress regime (Sibson, 1992; Hickman et al., 1995). These influences enhance the stress concentration in the seismogenic layer leading to mechanical failure. Spatial and temporal variations in the crustal stress field have been reported for the source areas of 1976 Tangshan earthquake (Research Group for the 1976 Tangshan Earthquake, 1982), the 1994 Northridge earthquake (M = 6.7) in southern California (Zhao et al., 1997b), and also the 1995 Kobe earthquake in Southwest Japan (Katao et al., 1997), which were associated with fluids in the fault zones. These many pieces of evidence suggest that the generation of a large earthquake is not a pure me-

178

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

chanical process, but is closely related to the physical and chemical properties of materials in the crust and upper mantle, such as magma, fluids, etc. (Zhao et al., 2002). The rupture nucleation zone should have a three-dimensional spatial extent, not just limited to the two-dimensional surface of a fault, as suggested earlier by Tsuboi (1956) in the concept of earthquake volume. Complex physical and chemical reactions take place in the source zone of a future earthquake, causing heterogeneities in the material property and stress field, which can be detected with seismic tomography and other geophysical methods. The source zone of an M = 6 –8 earthquake extends from about 10 km to over 100 km (Kanamori and Anderson, 1975). The resolution of our tomographic imaging in the Beijing area is close to that scale of the earthquake sources, which may have enabled us to image the earthquake-related heterogeneities (i.e., earthquake volumes) in the crust and uppermost mantle in the Chinese capital region. These results indicate that large earthquakes do not strike anywhere, but only anomalous areas that may be detected with geophysical methods (Zhao et al., 2002). Higher resolution seismic imaging and combining seismological results with geological, geochemical and geophysical investigations would certainly provide us with a better understanding of the earthquake generating process and would also contribute to the mitigation of earthquake hazards.

6. Conclusions We have used a large number of high-quality arrival times to determine a detailed 3-D P-wave velocity model of the crust and uppermost mantle beneath the Chinese capital region. This model has a higher resolution than the previous results and provides new information on the complex structure and seismotectonics of the region. Main findings of the present work are summarized as follows: (1) The seismic velocity images are characterized by block structures bounded by large fault zones. This region consists of three geological terranes: the North China Basin, the Taihangshan and the Yanshan mountainous areas, which exhibit different patterns of velocity distribution

(2)

(3)

(4)

(5)

in the tomographic images. The trend of velocity anomalies is well consistent with the trend of regional tectonics. The velocity images of the shallow crust (0– 10 km depth) reflect well the surface geology and topographic features. The basin and depression areas show up as low-velocity anomalies, while the uplift and mountainous areas are imaged as high-velocity zones. The pattern of velocity distribution in the upper crust changes across the Zhangjiakou –Penglai seismic belt. In the crust and down to the uppermost mantle, velocity images change abruptly across the Taihangshan front fault. Our tomographic imaging has revealed significant velocity heterogeneities in the middle and lower crust, some of which are well consistent with those detected by deep seismic soundings and other geophysical investigations. Large crustal earthquakes [in particular, those with M>6.5, such as the 1976 Tangshan earthquake (M = 7.8) and the 1679 Sanhe earthquake (M = 8.0)], are generally located in highvelocity areas in the upper to middle crust. In the lower crust to the uppermost mantle under the source zones of the large earthquakes, however, low-velocity and high-conductivity anomalies exist, which are considered to be associated with fluids, similar to the 1995 Kobe earthquake (M = 7.2) in Japan and the 2001 Bhuj earthquake (M = 7.8) in India. The fluids in the lower crust may cause the weakening of the seismogenic layer in the upper and middle crust and thus contribute to the rupture nucleation of the large crustal earthquakes.

Acknowledgements We thank the Center for Capital Seismic Network in the Center for Analysis and Prediction, China Seismological Bureau for providing the data used in this study. This work was partially supported by grants from National Natural Science Foundations of China (No. 40374009) and the Chinese Earthquake Study Foundation (No. 103076) to J. Huang and grants from the Japan Society for the Promotion of Science (Kiban-B 11440134, S-12002006) to D.

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

Zhao. We thank reviewers T.B. Yanovskaya and E.R. Engdahl. All the figures in this work are made by using the GMT (Wessel and Smith, 1995). References Bondar, I., Myers, S., Engdahl, E., Bergman, E., 2004. Epicenter accuracy based on seismic network criteria. Geophys. J. Int. 156, 483 – 496. China Seismological Bureau, E., 1986. Findings of Exploring the Crust and Upper Mantle Structure of China. Seismological Press, Beijing, p. 407 (in Chinese). Eberhart-Phillips, D., 1986. Three dimensional velocity structure in Northern California Coast Ranges from inversion of local earthquake arrival times. Bull. Seismol. Soc. Am. 76, 1025 – 1052. Fan, G., Wallace, T., Zhao, D., 1998. Tomographic imaging of deep velocity structure beneath the eastern and southern Carpathians, Romania: implications for continental collision. J. Geophys. Res. 103, 2705 – 2723. Feng, R., Zheng, S., 1987. Synthetical study of lithosphere structure under North China. Chin. Sci. Bull. 32, 1723 – 1727. Gao, W., Ma, J., 1993. Seismo-Geological Background and Earthquake Hazard in Beijing Area. Seismological Press, Beijing, p. 211 (in Chinese). Gu, G., 1983. Catalogue of Chinese Earthquakes (1831 BC – 1980 AD). Science Press, Beijing (in Chinese). Hauksson, E., Haase, J., 1997. Three-dimensional Vp and Vp/Vs velocity models of the Los Angeles basin and central Transverse Ranges, California. J. Geophys. Res. 102, 5423 – 5453. Hickman, S., Sibson, R., Bruhn, R., 1995. Introduction to special section: mechanical involvement of fluids in faulting. J. Geophys. Res. 100, 12831 – 12840. Huang, J., Zhao, D., Zheng, S., 2002. Lithospheric structure and its relationship to seismic and volcanic activity in southwest China. J. Geophys. Res. 107 (B10). doi: 10.1029/2000JB000137. Jin, A., Liu, F., Sun, Y., 1980. Three-dimensional P velocity structure of the crust and upper mantle under Beijing region. Acta Geophys. Sin. 23, 172 – 182 (in Chinese). Kanamori, H., Anderson, D., 1975. Theoretical basis of some empirical relations in seismology. Bull. Seismol. Soc. Am. 65, 1073 – 1095. Katao, H., Maeda, N., Hiramatsu, Y., 1997. Detailed mapping of focal mechanisms in and around the 1995 Hyogo-Ken Nanbu earthquake rupture zone. J. Phys. Earth 45, 105 – 119. Kayal, J., Zhao, D., Mishra, O., 2002. The 2001 Bhuj earthquake: tomographic evidence for fluids at the hypocenter and its implications for rupture nucleation. Geophys. Res. Lett. 29 (12). doi: 10.1029/2002GL015177. Leveque, J., Rivera, L., Wittlinger, G., 1993. On the use of the checker board test to assess the resolution of tomographic inversions. Geophys. J. Int. 115, 313 – 318. Li, D., 1981. Geological structure and hydrocarbon generation under the Bohai Bay, China. In: Mason, J. (Ed.), Petroleum Geology in China. Penn Well, Tulsa, Okla, pp. 180 – 192. Li, S., Zhang, X., Song, Z., 2001. Three-dimensional crustal struc-

179

ture of the capital area obtained by a joint inversion of deep seismic sounding data from multiple profiles. Acta Geophys. Sin. 44, 360 – 368 (in Chinese). Liu, G., 1987. The Cenozoic rift system of the North China Plain and the deep internal process. Tectonophysics 133, 277 – 285. Liu, F., Qu, K., Wu, H., 1986. Seismic tomography of the North China region. Acta Geophys. Sin. 29, 442 – 449 (in Chinese). Liu, G., Sun, J., Liu, J., 1989. The electrical structure of the crust and upper mantle in North China. In: Ma, X. (Ed.), Lithospheric Dynamics ATLAS of China. China Cartographic Press, Beijing, pp. 59 – 62 (in Chinese). Mishra, O., Zhao, D., 2003. Crack density, saturation rate and porosity at the 2001 Bhuj, India, earthquake hypocenter: a fluid-driven earthquake? Earth Planet. Sci. Lett. 212, 305 – 393. Nabelek, J., Chen, W., 1987. The Tangshan earthquake sequence and its implications for the evolution of the North China Basin. J. Geophys. Res. 92, 12615 – 12628. Paige, C., Saunders, M., 1982. LSQR: an Algorithm for sparse linear equations and sparse least squares. ACM Trans. Math. Softw. 8, 43 – 71. Research Group for the 1976 Tangshan Earthquake, 1982. The 1976 Tangshan Earthquake. Seismological Press, Beijing (in Chinese). Salah, M., Zhao, D., 2003. Three-dimensional seismic structure of Kii Peninsula, Southwest Japan: evidence for slab dehydration in the forearc. Tectonophysics 364, 191 – 213. Shao, X., Zhang, J., Chen, X., 1980. Crustal structure of the Beijing, Tianjin and Tangshan region estimated from seismic converted waves. Seismol. Geol. 2, 5 – 13 (in Chinese). Shedlock, K., Roecker, S., 1987. Elastic wave velocity structure under the crust and upper mantle beneath North China. J. Geophys. Res. 92, 9327 – 9350. Sibson, R.H., 1992. Implications of fault-valve behavior for rupture nucleation and recurrence. Tectonophysics 211, 283 – 293. Sun, R., Liu, F., 1995. Crust structure and strong earthquakes in Beijing, Tianjin and Tangshan area: I. P wave velocity structure. Acta Geophys. Sin. 38, 599 – 607 (in Chinese). Teng, J., Yao, H., Zhou, H., 1979. The crust structure in the Beijing, Tianjin, Tangshan and Zhangjiakou area. Acta Geophys. Sin. 22, 218 – 235 (in Chinese). Thurber, C., 1983. Earthquake location and three-dimensional crustal structure in the Coyote Lake area, central California. J. Geophys. Res. 88, 8226 – 8236. Tsuboi, C., 1956. Earthquake energy, earthquake volume, aftershock area, and strength of the Earth’s crust. J. Phys. Earth 4, 63 – 66. Um, J., Thurber, C., 1987. A fast algorithm for two-point seismic ray tracing. Bull. Seismol. Soc. Am. 77, 972 – 986. Wang, S., Xu, Z., Pei, S., 2003. Velocity structure of the uppermost mantle beneath North China from Pn tomography and its implications. Sci. China, Ser. D 46, 130 – 140 (Suppl.). Wessel, P., Smith, W., 1995. New version of the generic mapping tools released. EOS Trans. AGU 76, 329. Ye, H., Shedlock, K., Hellinger, S., 1985. The North China basin: an example of a Cenozoic rifted intraplate basin. Tectonics 4, 153 – 169. Ye, H., Zhang, M., Mao, F., 1987. The Cenozoic tectonic evolution

180

J. Huang, D. Zhao / Tectonophysics 385 (2004) 159–180

of the great North China: two types of rifting and crustal necking in the Great North China and their tectonic implications. Tectonophysics 133, 217 – 227. Yu, X., Chen, Y., Wang, P., 2003. Three-dimensional P wave velocity structure in Beijing – Tianjin – Tangshan area. Acta Seismol. Sin. 25, 1 – 14. Zang, S., Yang, J., 1984. The depth distribution of North China intraplate earthquakes and their physical background. Seismol. Geol. 6, 67 – 75. Zhang, X., 1998. Study of the crust and upper mantle structure in and around the Zhangjiakou – Bohai seismic belt. Active Fault Research, vol. 6. Seismological Press, Beijing, pp. 62 – 75. Zhao, D., 2001. New advances of seismic tomography and its applications to subduction zones and earthquake fault zones: a review. The Island Arc 10, 68 – 84. Zhao, D., Negishi, H., 1998. The 1995 Kobe earthquake: seismic image of the source zone and its implications for the rupture nucleation. J. Geophys. Res. 103, 9967 – 9986. Zhao, D., Hasegawa, A., Horiuchi, S., 1992. Tomographic imaging of P and S wave velocity structure beneath northeastern Japan. J. Geophys. Res. 97, 19909 – 19928.

Zhao, D., Hasegawa, A., Kanamori, H., 1994. Deep structure of Japan subduction zone as derived from local, regional and teleseismic events. J. Geophys. Res. 99, 22313 – 22329. Zhao, D., Kanamori, H., Negishi, H., 1996. Tomography of the source zone of the 1995 Kobe earthquake: evidence for fluids at the hypocenter? Science 274, 1891 – 1894. Zhao, D., Xu, Y., Wiens, D., Dorman, L., Hilderbrand, J., Webb, S., 1997a. Depth extent of the Lau back-arc spreading center and its relation to subduction processes. Science 278, 254 – 257. Zhao, D., Kanamori, H., Wiens, D., 1997b. State of stress before and after the 1994 Northridge earthquakes. Geophys. Res. Lett. 24, 519 – 522. Zhao, D., Mishra, O., Sanda, R., 2002. Influence of fluids and magma on earthquakes: seismological evidence. Phys. Earth Planet. Inter. 132, 249 – 267. Zhu, L., Zeng, R., 1990. Three-dimensional P-wave velocity structure under the Beijing network area. Acta Geophys. Sin. 33, 267 – 277 (in Chinese). Zhu, Z., Zhang, X., Zhang, J., 1997. The crust and upper mantle velocity structure along the Beijing – Huailai – Fengzhen profile. Acta Seismol. Sin. 19, 499 – 505.