Pure Appl. Geophys. 175 (2018), 193–205 Ó 2017 Springer International Publishing AG https://doi.org/10.1007/s00024-017-1691-y Pure and Applied Geophysics

The Crustal Structure of the North–South Earthquake Belt in Revealed from Deep Seismic Soundings and Gravity Data

1 1 2 2 YANG ZHAO, LIANGHUI GUO, LEI SHI, and YONGHUA LI

Abstract—The North–South earthquake belt (NSEB) is one of Liupan Mountains, and , along the major earthquake regions in China. The studies of crustal the southeastern edge of the to the structure play a great role in understanding tectonic evolution and in evaluating earthquake hazards in this region. However, some Red River fault zone of Province (Wang fundamental crustal parameters, especially crustal interface struc- et al. 1976, 2015a). The NSEB presents complicated ture, are not clear in this region. In this paper, we reconstructed the geological features and crustal structure. To the west crustal interface structure around the NSEB based on both the deep seismic sounding (DSS) data and the gravity data. We firstly of the NSEB is the relatively active Tibetan Plateau, reconstructed the crustal structure of crystalline basement (inter- and to the east are the stable Yangtze and Ordos face G), interface between upper and lower crusts (interface C) and blocks. Large variations of Moho depth and velocity Moho in the study area by compiling the results of 38 DSS profiles published previously. Then, we forwardly calculated the gravity structure are presented between the east and west anomalies caused by the interfaces G and C, and then subtracted areas (Liu et al. 1989; Song et al. 1991; Yin et al. them from the complete Bouguer gravity anomalies, yielding the 1998; Guo et al. 2009). The reconstruction of crustal regional gravity anomalies mainly due to the Moho interface. We structure in the NSEB would be beneficial in under- then utilized a lateral-variable density interface inversion technique with constraints of the DSS data to invert the regional anomalies standing tectonic evolution and geodynamics and in for the Moho depth model in the study area. The reliability of our evaluating earthquake hazards in this region. Moho depth model was evaluated by comparing with other Moho The crust of the NSEB is one of the most complex depth models derived from other gravity inversion technique and receiver function analysis. Based on our Moho depth model, we geological structures among the Chinese continent. mapped the crustal apparent density distribution in the study area Many geophysical studies were conducted around the for better understanding the geodynamics around the NSEB. NSEB in the past decades. However, so far the crustal

Key words: North–South earthquake belt, crustal structure, interface structure is still not clear in the NSEB area. Moho depth, gravity inversion, deep seismic soundings. Seismic refraction/reflection is the most direct and effective means to detect the crustal structure and interface information with good vertical resolution. Since 1980s, a number of deep seismic soundings 1. Introduction (DSS) profiles were accomplished in this area. Sev- eral crustal models in the Chinese continent were The North–South earthquake belt (NSEB) is the built based on the DSS data (Deng et al. 2011; Stolk main boundary between the east and west parts of et al. 2013). Since some new DSS profiles were China (see Fig. 1). It extends from the western edge conducted in recent years, a refined crustal model for of Ordos’s basin, through western Mountains, the NSEB should be constructed. Whereas, the reconstruction of crustal interface structure was restricted due to the sparseness of the DSS profiles in the study area. The complete Bouguer gravity 1 Key Laboratory of Geo-detection Ministry of Education anomalies are important data with a considerable and School of Geophysics and Information Technology, China lateral resolution for studying the crustal structure University of Geosciences (Beijing), Beijing 100083, China. E- and tectonics. They contain the gravity effects not mail: [email protected] 2 Institute of Geophysics, China Earthquake Administration, only from many of anomalous bodies and interface Beijing 100081, China. 194 Y. Zhao et al. Pure Appl. Geophys.

Figure 1 Left: Map of the NSEB and the adjacent areas with the major tectonic features. The black rectangle outlines the study area. Right: Distribution of the earthquake epicenters (1970–2016) in the NSEB and the adjacent areas, colored according to the magnitude of earthquakes. Blue: 5–7 magnitude earthquakes; green: 7-8 magnitude earthquakes. Red lines indicate block boundaries. KT Kunlun-Qaidam Terrane; YO Yanshan Orogen; OB Ordos Basin; QB Qaidam Basin; ST Songpan-Ganzi Terrane; TP Tibetan Plateau; SB Sichuan Basin; IP Indian Plate; YC Yangtze Craton structure within the crust, but also from construction interface inversion technique based on a lateral- of a Moho interface. Since high-resolution gravity variable density contrast model is required to seek a data cover the whole study area, it is necessary to reliable interface model. Prior information from the separate the gravity anomalies due to the density published literature is necessarily involved as con- interface and subsequently invert them for the inter- straints for better inversion. face depth. The aim of this study is to reconstruct the crustal Anomaly separation is one key technique in structure of the NSEB by combining the DSS and gravity data processing. Many frequency-domain fil- gravity data. We firstly constructed crustal interfaces tering techniques have been proposed for anomaly based on the published DSS data. Then, we separated separation, such as upward continuation (Jacobsen the gravity anomalies caused by the Moho interface 1987), matched filtering (Spector and Grant 1970), by removing the contribution of the basement inter- Wiener filtering (Pawlowski and Hansen 1990), face G and upper/lower crustal interface C. preferential continuation (Pawlowski 1995), and Subsequently, the reconstructed Moho interface was preferential filtering (Guo et al. 2012, 2013). But few applied to establish a variable density contrast model. of them could separate the target gravity anomalies We finally inverted the separated anomalies for Moho well, because of no theoretical relationship between depth by a lateral-variable density interface inversion the source depth of gravity field and anomaly spec- technique with constraints of the DSS data. trum. Hence, anomaly stripping would be an alternative technique for better separation. The con- 2. Data Source ventional technique of interface inversion from gravity data was based on a constant-density contrast 2.1. Deep Seismic Soundings Data model without any prior information (Oldenburg 1974). However, due to the complex crust structure Around the NSEB, many DSS profiles have been and the large scale of the NSEB, the density contrast completed for revealing the crustal structure in the value is not always a constant laterally. Thus, an past 30 years, with a cumulative length of over Vol. 175, (2018) The Crustal Structure of the North–South Earthquake Belt in China Revealed 195

Table 1 Information of 38 DSS profiles around the NSEB

Number Profile name Year References

1 Lijiang-Panzhihua 1984 Xiong et al. (1993) 2 Mouding-Xichuang 1984 Yin and Xiong (1992) 3 Zhefang-Binchuan 1982 Bai and Wang (2004) 4 Menglian-Malong 1982 Bai and Wang (2004) 5 Lijiang-Xinshizhen 1984 Cui et al. (1987) 6 Lazha-Changheba 1984 Lu et al. (1989) 7 Eryuan-Jiangchuan 1982 Zhang et al. (2007) 8 Simao-Zhongdian 1987 Lin et al. (1993) 9 Huashixia-Jianyang 1987 Cai et al. (2008) 10 Jinchuan-Tangke 1987 Cui and Chen (1994) 11 Batang-Zizhong 2000 Wang et al. (2003b) 12 Benzilan-Tangke 2000 Wang et al. (2003b) 13 Puer-Luxi 2011 Zhang et al. Figure 2 (2013a, b) The locations of the DSS profiles around the NSEB 14 Zhenkang-Luxi 2010 Pan et al. (2015) 15 Yushu-Gonghe 1998 Wang and Qian 18,000 km. Here, we assembled 38 DSS profiles (2000) around the NSEB for reconstructing the crustal 16 Altyn Tagh - 1987- Wang et al. (2005) Longmenshan 1989 model. Figure 2 shows locations of these profiles, 17 Aba-Shuangliu 1990s Zhu (2008) and Table 1 lists information of each profile. 18 Heishui-Beibei 1980s Song (1994) Although the geographical location of these profiles 19 Amuquhu-Lingtai 1986 Zhou (2005) 20 Maqin-Jingbian 1997 Li et al. (2002) is certainly uneven, the coverage is remarkably good 21 Menyuan-Weinan 1982 Zhang et al. (1985) along the NSEB, especially the regions of intense 22 Wendeng- Alxa left 1990s Wang et al. (2014a) tectonic deformation where earthquakes frequently banner 23 Xiji-Zhongwei 1999 Li et al. (2001) occur. 24 Chengxian-Xiji 1986 Li et al. (1991) 25 Tongwei-Huangling 2012 Li et al. (2014a, b) 26 Kangxian-Shiyan 2013 Li et al. (2015) 2.2. Gravity Data 27 Hezuo-Jingtai 2009 Zhang et al. (2013a, b) We assembled the free-air gravity anomaly 28 Aba-Suining 2010 Jia et al. (2014) around the NSEB from the Earth Gravitational Model 29 Jinchuan-Leshan 2013 Wang et al. (2015b) (EGM) 2008 (Pavlis et al. 2012). The EGM 2008 was 30 Chuxiong-Luoping 2005 Wang et al. (2009) 31 Zhongshan-Zizhi 1999 Wang et al. (2002) produced based on marine, land, air-borne gravity 32 Lincang-Yuxi 2011 Wang et al. (2014b) measurements and GRACE satellite altimetry with 33 Yunxian-Ninglang 2012 Chen (2015) resolution of 50 9 50. Compared to other previous 34 Maerkang-Gulang 2004 Zhang et al. (2008) 35 Aba-Wuqi 2014 Wang et al. (2017) models, the EGM 2008 model made great progress on 36 Maduo-Alxa right 2014 Wang et al. (2017) resolution and precision. After standard terrain and banner Bouguer corrections, we obtained the complete 37 Yanyuan-Mahu 2005 Yang et al. (2011) Bouguer gravity anomalies in the study area with 38 Tangke-Pujiang- 1983 Chen et al. (1988) Langzhong grid interval of 25 km. The ETOPO1 data (Amante and Eakins 2008) were utilized for terrain correc- noise in the anomaly data using the frequency- tions. The ETOPO1 effort integrates land topography domain low-pass filter with a cut-off wavelength of and ocean bathymetry to produce a global model of 100 km. The denoised complete Bouguer anomaly Earth’s surface. Then, we suppressed high-frequency values are shown in Fig. 3 with the values ranging 196 Y. Zhao et al. Pure Appl. Geophys.

the weight of each clear profile published in the past decade as 1, those roughly having some ambiguity in the interpretation were given a weight of 0.7, and those published over 30 years with lesser resolution were given a weight of 0.3. In contrast, we kept a weight of all profiles equal to 1 in the rest of the intersectional regions. We then calculated the depth distribution of the crustal interfaces G, C and Moho (Fig. 4a–c) after gridding with a working mesh of 0.25° 9 0.25°. In Fig. 4d, we also present the Moho depth of DSS data straight without gridding, except the Number 37 profile (Yanyuan-Mahu) and Number 38 profile (Tangke-Pujiang-Langzhong) which have no Moho data. According to Fig. 2, we focused on the regions inside the coverage of these DSS profiles and thus muted those outside the coverage in the subsequent figures. Figure 3 The map of the interface G shows the thickest The denoised complete Bouguer gravity anomalies around the sediment in the Sichuan Basin where the depth is NSEB 10 km. The eastern Qaidam Basin, the western Ordos Basin and the central Yunnan Province have an -5 2 from – 530 to – 25 9 10 m/s . One NNE-trending average sedimentary depth of 5 km. The average gravity anomaly gradient belt with dense contours depth of the interface C is around 20 km for all presented in the center of Fig. 3 is consistent with the regions except for the eastern Tibetan Plateau where striking of the NSEB. the interface depth was 22 km or more. In consider- ation of the supplements of new DSS data conducted in recent years, the reconstructed Moho model pre- 3. Reconstruction of Crustal Interface Models Based sents more details and higher resolution than the on the DSS Data previous Moho model (Deng et al. 2011) derived from DSS data as well. For instance, the recon- We reconstructed the top interface of the crys- structed Moho model has a 32 km depth in the talline basement (interface G), the interface between southern Yunnan Province, which is in agreement upper and lower crust (interface C), and the Moho with passive seismic work (Li et al. 2014b), while it interface around the NSEB based on the assembled is overestimated by the previous Moho model of 38 DSS profiles. Deng et al. (2011) where the depth is over 36 km. Firstly, we digitized the collected high-quality DSS profiles from the published literature. The pro- cedure of the digitization was that we interpreted the 4. Reconstruction of Moho Interface by Gravity- interface where P-wave velocity increases occurred: Constrained Inversion 5000–5600 to 5800–6000 m/s as the interface G, 6200–6400 to 6500–6600 m/s as the interface C, and Since the above Moho interface around the NSEB 6800–6900 to 7900–8000 m/s as the Moho (Duan (Fig. 4d) was reconstructed only from the sparse DSS et al. 2016). To address distinct discrepancies found data, inversion of high-resolution gravity data is around the intersection of profiles, a weight-average necessarily employed for infill. We first separated the algorithm was applied to ensure the continuity of the gravity anomalies caused by the Moho interface from interface models according to the quality of profiles. the denoised complete Bouguer anomalies, and then Around the intersectional regions of profiles, we set inverted the resulting anomalies for Moho depth by Vol. 175, (2018) The Crustal Structure of the North–South Earthquake Belt in China Revealed 197

Figure 4 The depth of crustal interfaces derived from the DSS results. The regions outside the coverage of DSS data were muted. a The interface G, b the interface C, c the Moho interface and d the digitized Moho from the DSS data without gridding employing a laterally variable density interface stripping the gravity anomalies caused by both the inversion technique constrained by the DSS data. interfaces G and C within the crust. First, we forwardly calculated the theoretical gravity anoma- lies due to the interfaces G and C (Fig. 4a, b) using 4.1. Separation for Regional Gravity Anomalies the Parker’s techniques (Parker 1973) with density We separated the regional gravity anomalies from contrasts of - 0.25 and - 0.1 9 103 kg/m3, respec- the denoised complete Bouguer anomalies (Fig. 3)by tively. Both the density contrast values are based on 198 Y. Zhao et al. Pure Appl. Geophys.

Figure 5 Figure 6 The regional gravity anomalies around the NSEB separated in this The Moho interface density contrast map paper the average density of the Crust 1.0 model (Laske mantle. After gridding with interval of 0.25° 9 0.25°, et al. 2013). Then we subtracted these theoretical we obtained the Moho density contrast map around gravity anomalies from the denoised complete the NSEB (Fig. 6). Then, we applied the gridded Bouguer anomaly values, yielding the regional grav- Moho density contrast model into the interface ity anomalies in the study area shown in Fig. 5, which inversion, and we softly constrained the inversion are mainly caused by the Moho interface. with a priori Moho depth values at 40 control points (red points in Fig. 7) picked from the DSS data, which simply penalized the inversion to ensure 4.2. Gravity-Constrained Inversion for Moho acceptable deviations between gravity model and Interface DSS data. Herein, 40 km was chosen as the reference We inverted the regional gravity anomalies datum depth which was the initial value of inversion. (Fig. 5) to improve the Moho interface using a Figure 7 showed the inversion result of Moho density interface inversion scheme (Oldenburg 1974; depth in the study area, which presents a large Salem et al. 2013). We employed a laterally variable variation from less than 38 km in the south to over density model for inversion along with several 70 km on the Tibetan Plateau. The Moho map shows control points of Moho depths from the DSS data extraordinary depth reaching down as far as 72 km utilized for constraints. beneath the eastern Tibetan Plateau. The crust thins First, we estimated the Moho density contrast gradually beneath the northern and eastern margin of values based on the Moho depth result (Fig. 4d) the Tibetan Plateau and the Qaidam Basin, where the compiled from the DSS data in this study. At each Moho depth shallows to 55 km. The Moho depth control point, we obtained the Moho density contrast varies gradually from the northeastern margin of values from the regional gravity anomaly field Tibetan Plateau to 43 km in the western part of the employing 1D slab formula Dq = Dg/2pGh, where Ordos Basin. The Sichuan Province has an average Dg is the regional gravity anomaly, G is the Moho depth of 66 km in the westernmost of the gravitational constant, and h is the DSS Moho depth. Sichuan Plateau but shallows sharply to 42 km in the The resultant density contrasts relate to density Sichuan Basin. The northwestern part of Yunnan variations between the crust and uppermost of the Province has a thick crust with a Moho depth of Vol. 175, (2018) The Crustal Structure of the North–South Earthquake Belt in China Revealed 199

depths beneath Ordos Basin (45 km) and northwest- ern Yunnan Province (50–57 km). These are also described in previous active and passive seismic studies (Li et al. 2006, 2014b). The distribution of features in both models agree over much of the study region. However, both models also show significant small-scale differences in some region. For instance, the average Moho depth throughout the eastern Tibet is slightly deeper (5–7 km) in our model, which seems more similar to previous studies (Wang and Qian 2000; Xu et al. 2007; Wang et al. 2015b) compared to the other models. In addition, our model displays a region of thin crust beneath the southern Yunnan Province, where Moho depth is 38–42 km, and which is close to the results of (Li et al. 2014b); however, the Moho depths for the same regions in the previous gravity model of Guo et al. (2012) are over 42 km. Figure 7 Of course, the inversion result largely depends on The Moho model derived from gravity-constrained inversion the accuracy of the anomaly separation and inversion around the NSEB. Red points give the locations of 40 constraint points from the DSS data techniques. The previous gravity inversion applied the frequency-domain filtering technique to separate the regional gravity anomalies, but there is no 50–57 km. Toward the southeastern tip of Yunnan theoretical relationship between the source depth Province, the Moho decreases to around 38 km. and gravity anomaly spectrum, which is the major Overall, the Moho depth variation pattern is consis- reason why anomaly separation by frequency-domain tent with the tectonic features around the NSEB. filtering has a potential negative influence on inver- sion results. Therefore, a lateral-variable density model was employed for better inversion in our 5. Comparisons of Different Moho Models technique, which is expected in reality. In conclusion, due to the careful anomaly separation and robust 5.1. Comparing with the Previous Gravity Result inversion technique, the effectiveness of the new We compared the new Moho model (Fig. 7) proposed technique was demonstrated in comparison obtained in this study with one of the previous results with the previous gravity results. by gravity inversion (Guo et al. 2012). Figure 8a shows the Moho depth around the NSEB from the 5.2. Comparing with Seismic Results gravity result of Guo et al. (2012). Wherein, the regional gravity anomalies were separated using the We then further compared our Moho model with preferential filtering technique and then were inverted another Moho model obtained from receiver function for the Moho depth using the non-linear regression (RF) (Fig. 9a; Li et al. 2014b). Figure 9b displays the technique. Figure 8b presented the difference differences between the RF Moho model at the between our new Moho model and the previous one estimated stations and those extracted from our new from Guo et al. (2012). To facilitate the comparison, Moho model using gravity-constrained inversion. The we plotted the Moho models using the same color uncertainty of the Moho depth from the RF data is scale. around ± 4 km (Li et al. 2014b). Furthermore, the Both the new Moho model in this study and the RMS value between the 40 control points derived one from Guo et al. (2012) indicate similar Moho from DSS data and those extracted from new Moho 200 Y. Zhao et al. Pure Appl. Geophys.

Figure 8 a The previous Moho model derived from the gravity inversion by Guo et al. (2012), b the difference between the new Moho model in this study and the previous one from Guo et al. (2012) model is less than ± 3 km. Thus, any discrepancy Fig. 9b, although there is obviously a strong agree- less than ± 7 km between the two models at an ment between our Moho model and the RF Moho arbitrary station is treated as statistically equivalent. model in most regions, the differences are unusually Figure 9b presents similar features around most large (more than 14 km) at several points, which are regions of Ordos and Sichuan Province, where the unexpected compared to surrounding points, and are differences are less than ± 6 km. The average Moho probably associated with multiple waves originated depth around the northeastern margin of Tibetan from the thick sediment. Whereas the gravity data Plateau of our Moho model has a strong resemblance mainly helps to construct a smooth, continuous and of the RF model, where the deviations are generally consistent Moho model. Overall, it is evident that within ± 3 km. The deviations are slightly large over 80% of the 353 RF stations and gravity results (6–9 km) in southwest of Yunnan Province where the agree within ± 7 km (Fig. 9c). This indicates that the Moho depth is underestimated by the RF Moho two models share similar features over large portions model. Actually, the Moho depth of southwestern of the NSEB. Yunnan Province is usually overestimated (over In addition, we also compared our new Moho 42 km) by the previous gravity model. The deviations model and DSS result along the Huashixia-Jianyang may seem like systematic errors in Moho depth profile (Number 9 in Fig. 2 and Table 1), which between the results derived from seismic and gravity crosses the NSEB with a length of over 800 km. data. But in our work, the result has apparently According to Fig. 10, it is certainly demonstrated improved (less than 42 km). To the best of our that the smooth curve of our Moho model possesses knowledge, these systematic differences are likely less depth detail than the DSS profile, but the small ascribed to the complicated geological structure of discrepancies (less than 5 km found everywhere) southwestern Yunnan Province whose crust is soft and the acceptable RMS error (2.44 km) simultane- and active relative to the Sichuan Basin. Therefore, ously demonstrate that our Moho model coincides the slight large deviations of southwestern Yunnan well with the DSS profiles over much of the NSEB Province may be insignificant. As shown from region. Vol. 175, (2018) The Crustal Structure of the North–South Earthquake Belt in China Revealed 201

Figure 9 The differences between our new Moho model and the previous RF Moho model. a The RF Moho model (Li et al. 2014b), b the differences between the RF Moho model and those extracted from our new Moho model, and c the deviations between our Moho model depth (lateral axis) and RF Moho model (vertical axis). Red lines indicate ± 7 km deviation

6. Apparent Density Distribution of the Crust structure. To add more insight of the crustal structure in the NSEB, we inverted the denoised complete Apparent density mapping is an effective way to Bouguer gravity anomalies (Fig. 3) for apparent provide density distribution for studies of geological density distribution of the crust using the space- 202 Y. Zhao et al. Pure Appl. Geophys.

Figure 10 The comparison between our new Moho model and the DSS data along the Number 9 profile (Huashixia-Jianyang). The red line indicates the Moho depth from the DSS data, and the blue line show the Moho depth derived from our new Moho model domain mapping technique similar to Guo et al. (2016). For simplicity sake, a sea level (0 km) was chosen as the top interface of the crustal layer, and the new Moho model (Fig. 7) was regarded as the bottom interface. The mapping result is shown in Fig. 11, which presents a distinct NNE-trending density gradient belt along the NSEB. The calculated average crustal density values show variations that range from the eastern Tibetan Plateau (2.53 9 103 kg/m3) or less, while increasing sharply to the value of 2.75 9 103 kg/m3 in the Sichuan Basin, and gradually to 2.61 9 103 kg/m3 beneath the northwest of Yunnan Province. The previous tomography studies (Li et al. 2009; Huang et al. 2013) suggested that low-velocity anomalies appeared below the eastern Tibetan Pla- teau, and that both the Sichuan Basin and the Ordos Basin were high-velocity blocks. As shown in Figure 11 Fig. 11, in view of the positive correlation between The apparent density distribution of the crust in the NSEB velocity and density, our apparent density distribution is consistent with these tomography results. In addi- between the two blocks on crustal structure. Collision tion, our apparent density distribution correlates well between the Indian Plate and the Eurasian plate with the crustal deformation and the tectonic activity resulted in crustal thickening around the Tibetan in the study area. It is clear that the density change Plateau and promoted the flow of crustal materials from the eastern Tibetan Plateau to the Sichuan Basin from the Tibetan Plateau towards the eastern regions. is very rapid, which implies large differences existed Nevertheless, the intact Sichuan Basin (Yangtze Vol. 175, (2018) The Crustal Structure of the North–South Earthquake Belt in China Revealed 203 block), a stable and high-density block, inhibited the might be more mafic in composition. In contrast, the materials trend of spreading eastward so that the flow relatively low-density values were presented in the is only to the soft southeastern and northeastern southeastern Tibetan Plateau where the crust was regions, where the low density and deep Moho likely to be partially melted. Large differences of the developed. This has also been described in the Pois- crustal density between two sides of the NSEB shown son’s ratio studies (Xu et al. 2007, 2013; Wang et al. in the apparent density map correlate with the motion 2010). In those studies, the intermediate values and collision of the plates. (0.26–0.28) around the Sichuan Basin denoted that the crust was likely to be mafic in composition, which usually are high density. The prominent high values Acknowledgements (t [ 0.30) found beneath the southeastern Tibetan Plateau implied that the crust was partially melted, We greatly thank Editor Hans-Ju¨rgen Go¨tze and two which caused low density. The density gradient belt anonymous reviewers for their helpful comments and along the NSEB displays a significant clockwise valuable suggestions. This work was financially rotation pattern from the eastern Tibetan Plateau to supported by the National Natural Science Founda- Sichuan-Yunnan block to southwest part of Yunnan tion of China (41774098, 41374093, 41430213), the Province, which basically agrees with the results of Fundamental Research Funds for the Central Univer- Pn velocity variations of Huang et al. (2003) and sities, and Key Laboratory of Geo-detection Ministry crustal stress field characteristics (Xu et al. 2008). of Education and School of Geophysics and Infor- mation Technology(China University of Geosciences, Beijing) open Project (GDL1507, GDL1509). 7. Conclusions REFERENCES We reconstructed the crustal interface models around the NSEB by compiling the results of avail- Amante, C., & Eakins, B. W. (2008). ETOPO1 1 arc-minute global able published DSS data, and then improved the relief model: Procedures, data sources and analysis. National Geophysical Data Center, NESDIS, NOAA, U.S. Department of Moho interface by gravity inversion with constraints Commerce, Boulder, CO. from the DSS data. The reconstructed interface Bai, Z. M., & Wang, C. Y. (2004). Tomography research of the models present more detailed depth distribution than Zhefang-Binchuan and Menglian-Malong wide-angle seismic profiles in Yunnan province. Chinese Journal of Geophysics (in previous work and contribute to bringing insight into Chinese), 47(2), 257–267. the crustal structure. Comparison of our Moho model Cai, X. L., Cao, J. M., Zhu, J. S., & Cheng, X. Q. (2008). A to previous Moho model from gravity inversion preliminary study on the 3-D crust structure for the Longmen lithosphere and the genesis of the huge Wenchuan earthquake, indicated that careful separation for the regional Sichuan, China. Journal of Chengdu University of Technology gravity anomalies is regarded as a crucial step for (Science & Technology Edition, in Chinese), 35(4), 357–365. inversion. Furthermore, a robust inversion technique Chen, S.W. (2015). 2D crustal velocity structure and structural is also considered to be a key for the result. The feature from Yunxian-Ninglang profile in western Yunnan, China. Master dissertation (in Chinese), Institute of Geophysics, reliability of the technique proposed in this work was China Earthquake Administration, Beijing, China. demonstrated by comparisons with another Moho Chen, X. B., Wu, Y. Q., Xu, W. M., et al. (1988). Developments in depth model derived from receiver function analysis the Research of Deep Structures of Chinese Continent (in Chi- nese). Department of Scientific Programming and Earthquake as well. Monitoring, China Earthquake Administration, Beijing: Geo- We also mapped the crustal apparent density logical Publishing House. distribution around the NSEB based on the Moho Cui, Z. Z., & Chen, J. P. (1994). Crustal and deep-seated structures of Longmen Mountains and west Sichuan Plateau. Bulletin of model reconstructed by gravity-constrained inver- The 562 Comprehensive Geological Brigade. Chinese Academy sion. The apparent density map provides some of Geological Sciences (in Chinese)., 11–12, 1–21. significant knowledge of the crustal composition of Cui, Z. Z., Lu, D. Y., Chen, J. P., Zhang, Z. Y., & Huang, L. Y. the NSEB, indicating that the high-density values (1987). The deep structure and tectonic features of the crust in Panxi area. Chinese Journal of Geophysics (in Chinese), 30(6), were presented in the Yangtze block where the crust 566–580. 204 Y. Zhao et al. Pure Appl. Geophys.

Deng, Y. F., Li, S. L., Fan, W. M., & Liu, J. (2011). Crustal the region of Xiji-Zhongwei. Seismology and Geology (in Chi- structure beneath south China revealed by deep seismic sound- nese), 23(1), 86–92. ings and its dynamics implications. Chinese Journal of Li, S. L., Zhang, X. K., Zhang, C. K., Zhao, J. R., & Cheng, S. X. Geophysics (in Chinese), 52(10), 2560–2574. (2002). A preliminary study on the crustal velocity structure of Duan, Y. H., Wang, F. Y., Zhang, X. K., Lin, J. Y., Liu, Z., Liu, B. Maqin-Lanzhou-Jingbian by means of deep seismic soundings F., et al. (2016). Three dimensional crustal velocity structure profile. Chinese Journal of Geophysics (in Chinese), 45(2), model of the middle-eastern Craton (HBCrust1.0). 210–217. Science China Earth Sciences (in Chinese), 46(6), 845–856. Lin, Z. Y., Hu, X. X., Zhang, W. B., Zhang, H. F., He, Z. Q., Lin, Guo, B., Liu, Q. Y., Chen, J. H., Liu, L. S., Li, S. C., Li, Y., et al. Z. M., et al. (1993). Study on velocity structure characters of (2009). Teleseismic P-wave tomography of the crust and upper crustal upper-mantle in Panxi region. Acta Seismologica Sinica mantle in Longmenshan area, west Sichuan. Chinese Journal of (in Chinese), 15(4), 427–441. Geophysics (in Chinese), 52(2), 346–355. Liu, J. H., Liu, F. T., Wu, H., Li, Q., & Hu, G. (1989). Three Guo, L., Meng, X., Chen, Z., Li, S., & Zheng, Y. (2013). Prefer- dimensional velocity images of the crust and upper mantle ential filtering for gravity anomaly separation. Computers & beneath North–South zone in China. Chinese Journal of Geo- Geosciences, 51, 247–254. physics (in Chinese), 32(2), 143–152. Guo, L., Meng, X., Shi, L., & Chen, Z. (2012). Preferential filtering Lu, D. Y., Cui, Z. Z., Chen, J. P., & Li, X. P. (1989). Application of method and its application to Bouguer gravity anomaly of Chi- explosion seismic sounding in the study of the crustal structure of nese continent. Chinese Journal of Geophysics (in Chinese), the Kangding-Dukou meridional structure belt. Geophysical 55(12), 4078–4088. Review (in Chinese), 35(1), 41–51. Guo, L., Shi, L., Meng, X., et al. (2016). Apparent magnetization Oldenburg, D. W. (1974). Inversion and interpretation of gravity mapping in the presence of strong remanent Magnetization: the anomalies. Geophysics, 39(4), 526–536. space-domain inversion approach. Geophysics, 81, J25–J38. Pan, S. Z., Wang, F. Y., Duan, Y. H., Deng, X. G., Song, X. H., Huang, Z. X., Li, H. Y., & Xu, Y. (2013). Lithospheric S-wave Duan, Y. L., et al. (2015). Basement structure of southern velocity structure of the North–South Seismic Belt of China from Yunnan and adjacent areas: The Zhenkang-Luxi deep seismic surface wave tomography. Chinese Journal of Geophysics (in sounding profile. Chinese Journal of Geophysics (in Chinese), Chinese), 56(4), 1121–1131. 58(11), 3917–3927. Huang, J. L., Song, X. D., & Wang, S. Y. (2003). Fine structure of Parker, R. L. (1973). The rapid calculation of potential anomalies. Pn velocity beneath Sichuan-Yunnan region. Science in China Geophysical Journal of the Royal Astronomical Society, 31(4), (Series D), 46, 201–209. 447–455. Jacobsen, B. H. (1987). A case for upward continuation as a Pavlis, N. K., Holmes, S. A., Kenyon, S. C., & Factor, J. K. (2012). standard separation filter for potential field maps. Geophysics, The development and evaluation of the earth gravitational model 52(8), 1138–1148. 2008 (EGM2008). Journal of Geophysical Research, 117, Jia, S. X., Liu, B. J., Xu, Z. F., Liu, Z., Feng, S. Y., Zhang, J. S., B04406. et al. (2014). The crustal structures of the central Longmenshan Pawlowski, R. S. (1995). Preferential continuation for potential- along and its margins as related to the seismotectonics of the field anomaly enhancement. Geophysics, 60(2), 390–398. 2008 Wenchuan Earthquake. Science China: Earth Sciences (in Pawlowski, R. S., & Hansen, R. O. (1990). Gravity anomaly sep- Chinese), 44(3), 497–509. aration by Wiener filtering. Geophysics, 59(1), 539–548. Laske, G., Masters, G., Ma, Z., & Pasyanos, M. (2013). Update on Salem, A., Green, C., Campbell, S., Fairhead, J. D., Cascone, L., & CRUST1.0—A 1-degree Global Model of Earth’s Crust. Geo- Moorhead, L. (2013). Moho depth and sediment thickness esti- physical Research Abstracts, 15, Abstract EGU2013-2658. mation beneath the Red Sea derived from satellite and terrestrial Li, Y. K., Gao, R., Gao, J. W., Mi, S. X., Yao, Y. T., Li, W. H., gravity data. Geophysics, 78(5), 89–101. et al. (2015). Characteristics of crustal velocity structure along Song, H. B. (1994). The comprehensive interpretation of geological Qingling orogenic belt. Process in Geophysics (in Chinese), and geophysical data in the orogeny belt of Longmen Mountains, 30(3), 1056–1069. China. Journal of Chengdu Institute of technology (in Chinese), Li, Y. K., Gao, R., Mi, S. X., Yao, Y. T., Gao, J. W., Li, W. H., 21(2), 79–88. et al. (2014a). The characteristics of crustal velocity structure for Song, Z. H., An, C. Q., Chen, G. Y., & Chen, L. H. (1991). Study Liupan Mountia-Ordos Basin in the northeastern margin of on 3D velocity structure and anisotropy beneath the west China Qinghai-Tibet Plateau. Geological Review (in Chinese), 60(5), from the Love wave dispersion. Chinese Journal of Geophysics 1147–1157. (in Chinese), 34(6), 694–707. Li, Y. H., Gao, M. T., & Wu, Q. J. (2014b). Crustal thickness map Spector, A., & Grant, F. S. (1970). Statistical models for inter- of the Chinese mainland from teleseismic receiver functions. preting aeromagnetic data. Geophysics, 35(2), 293–302. Tectonophysics, 611(2014), 51–60. Stolk, W., Kaban, M., Beekman, F., Tesauro, M., Mooney, W. D., Li, Q. H., Guo, J. K., Zhou, M. D., Wei, D. Q., Fan, B., & Hou, X. & Cloetingh, S. (2013). High resolution regional crustal models Y. (1991). The velocity structure of Chengxian-Xiji profile. from irregularly distributed data: application to Asia and adjacent Northwestern Seismological Journal (in Chinese), 33, 37–43. areas. Tectonophysics, 602(2013), 55–68. Li, Y. H., Wu, Q. J., Tian, X. B., Zhang, R. Q., Pan, J. T., & Zeng, Wang, F. Y., Duan, Y. H., Zhang, J. S., Jia, S. X., & Pan, S. Z. R. S. (2009). Crustal structure in the Yunnan region determined (2017). China seismic array waveform data of Himalaya project by modeling receiver functions. Chinese Journal of Geophysics (in Chinese). China Earthquake Administration (unpublished). (in Chinese), 52(1), 67–80. Wang, C. Y., Han, W. B., Wu, J. P., Lou, H., & Bai, Z. M. (2003a). Li, S. L., Zhang, X. K., Ren, Q. F., Zhang, C. K., Shi, J. H., Zhao, J. Crustal structure beneath the Songpan-Garze orogenic belt. Acta R., et al. (2001). Seismic sounding profile and its interpretation in Seismologica Sinica (in Chinese), 25(3), 229–241. Vol. 175, (2018) The Crustal Structure of the North–South Earthquake Belt in China Revealed 205

Wang, C. Y., Lou, H., Wang, X. L., Qin, J. Z., Yang, R. H., & Xu, L. L., Rondenay, S., & Hilst, R. D. (2007). Structure of the Zhao, J. M. (2009). Crustal structure in Xiaojiang fault zone and crust beneath the southeastern Tibetan Plateau from teleseismic its vicinity. Earth Science, 22(2009), 347–356. receiver functions. Physics of the Earth and Planetary Interiors, Wang, C. Y., Lou, H., Wu, J. P., Bai, Z. M., Huangfu, G., & Qin, J. 165(2007), 176–193. Z. (2002). Seismological study on the crustal structure of Xu, J. R., Zhao, Z. X., & Yuzo, I. (2008). Regional characteristics Tengchong volcano-geothermal area. Acta Seismologica Sinica of crustal stress filed and tectonic motions in and around Chinese (in Chinese), 24(3), 231–242. mainland. Chinese Journal of Geophysics (in Chinese), 51(3), Wang, Y. X., Mooney, W. D., Han, G. H., Yuan, X. C., & Jiang, M. 770–781. (2005). Crustal P-wave velocity structure from Altyn Tagh to Yang, Z. X., Wang, F. Y., Duan, Y. H., Zhang, C. K., Zhao, J. R., Longmen Mountains along the Taiwan-Altay geoscience tran- Zhang, J. S., et al. (2011). Basement structure of southeastern sect. Chinese Journal of Geophysics (in Chinese), 48(1), 98–106. boundary region of Sichuan-Yunnan active blocks; Analysis Wang, F. Y., Pan, S. Z., Liu, L., Liu, B. F., Zhang, J. S., Deng, X. result of Yanyuan-Xichang-Zhaojue-Mahu deep seismic sound- G., et al. (2014a). Wide angle seismic exploration of Yuxi-Lin- ing profile. Acta Seismologica Sinica (in Chinese), 33(4), cang profile-The research of crustal structure of the red river fault 431–442. zone and southern Yunnan. Chinese Journal of Geophysics (in Yin, X. H., Li, Y. S., & Liu, Z. P. (1998). Density and magnetic Chinese), 57(10), 3247–3258. structure of the crust and upper-mantle in the northern part of Wang, Y. X., & Qian, H. (2000). Study of crustal velocity structure South-North earthquake zone. Crustal Deformation and Earth- in east Qinghai. Earth Science Frontiers (in Chinese), 7(4), quake (in Chinese), 18(4), 11–17. 568–579. Yin, Z. X., & Xiong, S. B. (1992). Explosion seismic study for the Wang, Z. S., Wang, Z. Y., Gu, J. P., & Xiong, X. Y. (1976). A 2-D crustal structure in Xichang-Dukou-Mudian region. Chinese preliminary investigation of the limits and certain features of the Journal of Geophysics (in Chinese), 35(4), 451–458. North–South seismic zone of China. Chinese Journal of Geo- Zhang, Z. J., Bai, Z. M., Klemperer, S. L., Tian, X. B., Xu, T., physics (in Chinese), 19(2), 110–117. Chen, Y., et al. (2013a). Crustal structure across northeastern Wang, S. J., Wang, F. Y., Zhang, J. S., Jia, S. X., Zhang, C. K., Tibet from wide-angle seismic profiling: Constraints on the Zhao, J. R., et al. (2014b). The P-wave velocity structure of the Caledonian Qilian orogeny and its reactivation. Tectonophysics, lithosphere of the North China Craton-Results from the Wen- 606(2013), 140–159. deng-Alxa Left Banner deep seismic sounding profile. Science Zhang, X. K., Jia, S. X., Zhao, J. R., Zhang, C. K., Yang, J., Wang, China. Earth Sciences (in Chinese), 57, 2053–2063. F. Y., et al. (2008). Crustal structures beneath West Qinling-East Wang, S. J., Wang, F. Y., Zhang, J. S., Liu, B. F., Zhang, C. K., Kunlun orogen and its adjacent area-Results of wide-angle Zhao, J. R., et al. (2015a). The deep seismogenic environment of seismic reflection and refraction experiment. Chinese Journal of

Lushan MS 7.0 earthquake zone revealed by a wide-angle Geophysics (in Chinese), 51(2), 439–450. reflection/refraction seismic profile. Chinese Journal of Geo- Zhang, E. H., Lou, H., Jia, S. X., & Li, Y. H. (2013b). The deep physics (in Chinese), 58(9), 3193–3204. crust structure characteristics beneath western Yunan. Chinese Wang, C. Y., Wu, J. P., Lou, H., Zhou, M. D., & Bai, Z. M. Journal of Geophysics (in Chinese), 56(6), 1915–1927. (2003b). The P-wave velocity structure in western Sichuan pro- Zhang, S. Q., Wu, L. J., Guo, J. M., Chen, X. B., Zhao, J. X., Ding, vince and eastern Tibet region. Science in China (Series D in Y. Y., et al. (1985). An interpretation of the DSS data on Chinese), 33, 181–189. Menyuan-Pingliang-Weinan profile in west China. Acta Geo- Wang, C. Y., Yang, W. C., Wu, J. P., & Ding, Z. F. (2015b). Study physica Sinica (in Chinese), 28(5), 460–472. on the lithospheric structure and earthquakes in North–South Zhang, Z., Xu, C. M., Meng, B. Z., Liu, C., & Teng, J. W. (2007). tectonic belt. Chinese Journal of Geophysics (in Chinese), Crustal reflectivity characters from the Eryuan-Jiangchuan wide- 58(11), 3867–3901. angle seismic profile. Chinese Journal of Geophysics (in Chi- Wang, C. Y., Zhu, L. P., Lou, H., Huang, B. S., Yao, Z. X., & Luo, nese), 50(4), 1082–1088. X. H. (2010). Crustal thicknesses and Poisson’s ratios in the Zhou, M. D. (2005). The seismic tomography on the velocity eastern Tibetan Plateau and their tectonic implications. Journal structure of the crust and upper mantle in the Northeastern of Geophysical Research, 115, B11301. margin of the Qinghai-Tibetan Plateau. PhD Dissertation (in Xiong, S. B., Zheng, Y., Yin, Z. X., Zeng, X. X., Quan, Y. L., & Chinese), Institute of Geophysics, China Earthquake Adminis- Sun, K. Z. (1993). The 2D structure and its tectonic implications tration, Beijing, China. of the crust in the Lijiang-Panzhihua-Zhehai region. Chinese Zhu, J. S. (2008). The Wenchuan earthquake occurrence back- Journal of Geophysics (in Chinese), 36(4), 434–444. ground in deep structure and dynamics of lithosphere. Journal of Xu, X. M., Ding, Z. F., Shi, D. N., & Li, X. F. (2013). Receiver Chengdu University of Technology (Science & Technology Edi- function analysis of crustal structure beneath the eastern Tibetan tion, in Chinese), 35(4), 348–356. plateau. Journal of Asian Earth Sciences, 73(2013), 121–127.

(Received June 17, 2017, revised October 5, 2017, accepted October 9, 2017, Published online October 19, 2017)