Earthquake Reaearch in China  2018, Vol. 32 Issue (1): 89-99
The Crustal Velocity Structure of Western Inner Mongolia
Zhao Yanhong1, Jia Xiye2
1. Earthquake Agency of Inner Mongolia Autonomous Region, Hohhot 010051, China;
2. Baochang Seismic Station, Baochang 027099, Xilin Gol League, Inner Mongolia, China
Abstract: The terrain of Inner Mongolia is long and narrow, and the geological structure is complicated. The South China crustal velocity model and Inner Mongolia's optimal crustal velocity model 2015 cannot fully meet the earthquake location requirements of Inner Mongolia. Based on the seismological observations produced by Inner Mongolia Seismic Digital Network from 2009 to 2016, the initial model was obtained by using the linear fit of the seismic phases and the converted travel time curve. The Hyposat results of 225 earthquakes that occurred in western Inner Mongolia were scanned using this model, and the velocity model for western Inner Mongolia was determined as follows:V1=6.06km/s; VPb=6.61km/s; Vn=8.12km/s; H1=30m; and the Moho depth H=44km. Comparison of the test results of the new model and the reference model shows that the residual error of the new model and the mean deviation of the epicenter location have obviously decreased.
Key words: Inner Mongolia Region     South China crustal velocity model     Optimal crustal velocity model of 2015     Western Inner Mongolia crustal velocity model

INTRODUCTION

Obtaining accurate and reliable seismic positioning results is the key to earthquake prediction and seismic cataloging of the Regional Digital Seismic Network, and is also an important basis for earthquake prediction and seismology research. The crust velocity model is closely related to the location precision of earthquakes (Zhao Zhonghe, 2010), and a proper earth's crust velocity model can help seismologists accurately judge earthquake measuring accuracy (Zhu Yuanqing et al., 1997; Zhang Tianzhong et al., 2007), therefore, the role of the earth's crust velocity model becomes more prominent (Zhu Yuanqing et al., 1990). Accurately determining the longitude and latitude of the epicenter and focal depth of an earthquake has always been of concern to seismology research (Zhu Yuanqing et al., 1997).

In the complex geological western region of Inner Mongolia (36°-46°N, 96°-115°E), the east-west fold-fault structure and two major NE-and NW-trending fault zones are distributed. In the uplift zones, differential activity is not obvious, earthquake activity is weak, and there is no strong earthquake activity. The Hetao rift belt is located between the Yinshan uplift and Ordos uplift zone and borders the Langshan piedmont fault in the west, the Horinger fault in the east, the Seerteng Mountain, Wula Mountain Daqingshan piedmont faults in the north, and the northern marginal faults of Ordos in the south. Moderate strong earthquakes often occurred in the junctions between the major boundary faults and the secondary transverse faults.

Li Xiang et al. (1987) used the fitting method of the seismic waves of distant earthquakes to obtain the crustal thickness of Inner Mongolia, which is 46km to 54km. Using the seismic refraction sounding method, Liu Changquan et al. (1991) determined the velocity structure of the crust in the Ordos block, Hohhot-Baotou basin, Yinshan block and the fold belt in Inner Mongolia and other regions. The results show that the crust in the western Inner Mongolia has obvious layering characteristics, and the crustal thickness is 43km-48km. Zhang Hongdshuan et al. (2010) used the method of receiving functions to determine the crust thickness beneath 7 stations in the Hohhot-Baotou basin, and the results revealed a crust thickness of 42km to 46km.

The current general model used by the Inner Mongolia Digital Seismic Network is not consistent with the actual geological structure in western Inner Mongolia, the location residual is large, and the theoretical arrival is not consistent with the actual arrival time, which brings great confusion to earthquake monitoring, earthquake prediction and emergency rescue work. Therefore, the establishment of a crustal velocity model that conforms to the geological characteristics of the western region of Inner Mongolia is urgent. Using the observation data from the Digital Seismic Network of Inner Mongolia to determine and amend the existing velocity models (the South China model, and the 2015 optimal model of Inner Mongolia), the results can be applied to seismic network observation practice, thus improving the quality of earthquake location. The seismic stations in western Inner Mongolia are relatively dense, and a certain number of earthquakes have been recorded, which provides data for seismological research. In this paper, the two-layered velocity structure of the crust in the western region of Inner Mongolia is inverted by using the observed seismic travel time data from the Inner Mongolia Digital Seismic Network.

1 DATA SELECTION

Two types of data are selected for the establishment and inspection of the new model in western Inner Mongolia.

① The seismic data used to establish the western Inner Mongolia velocity model:

We select the records of 225 ML≥3.0 earthquake events that occurred in western Inner Mongolia recorded by the Digital Seismic Network of Inner Mongolia from 2009 to April 2016 and use the Pn and Pg combined focal depth determination method (Zhu Yuanqing et al., 1997) to re-measure the focal depths. Fig. 1 shows the radiographic map of the selected earthquake and station.

 Fig. 1 Distribution of earthquakes, seismic ray paths and stations

② The seismic data used to test the eastern Inner Mongolia model.

The earthquakes occurring in eastern Inner Mongolia recorded by the Inner Mongolia Digital Seismic Network are selected by random to test the new model.

③ Blasting data used to test the eastern Inner Mongolia model.

Blasting data important for checking the location accuracy, and the blasting in eastern Inner Mongolia Region recorded by the Inner Mongolia Digital Seismic Network is randomly selected to test the new model.

2 PRINCIPLES AND METHODS

The continental crust is relatively thick, with an average of about 35km, and the mountain and plateau region can reach 70km. The continental crust can be divided into the granite layer and basalt layer in terms of chemical composition. It is reasonable to pick up the existence of the Conrad surface from a large number of Pb seismic phases in the seismic records collected in this study.

Assuming that the crustal structure has two layers, seismic waves within the layer have a uniform speed, the velocity on Conrad discontinuity is the same as that of the lower crust. The velocity model contains five parameters, the upper crust velocity V1(VPg), lower crust velocity V2(VPb), the velocity along the Moho surface Vn(VPn), upper crust thickness H1 and lower crust thickness H2. Fig. 2 shows the longitudinal wave propagation path.

 Fig. 2 Sketch of P waves propagation paths

The focal depth ranges from 0km to 21km, and the upper crust thickness of the study area is about 25km, so it can be assumed that the earthquakes all occur in the upper crust. In the case of limited data, in order to avoid the correlation effect between parameters, multiple parameters are not used in the joint inversion, and each parameter is determined step by step. The travel time of direct wave Pg can determine the upper crust P-wave velocity; the lower crust velocity is determined by the travel time of Conrad surface refraction wave Pb, and the upper crust thickness, the total crust thickness and the wave velocity along the Moho discontinuity are inverted by the travel time of first wave Pn.

Based on existing research on the Inner Mongolia regional crustal velocity model, the seismic phases velocity fitting curve and reduced travel time curve, and the results of Hyposat batch processing, reasonable adjustments have been made on the five parameters, namely, the upper crust velocity V1 (VPg), the lower crust velocity V2 (VPb), the velocity along the Moho V3 (VPn), the upper crust thickness H1 and the lower crust thickness H2, to establish the initial model, then, the final optimal crustal velocity model for the western region of Inner Mongolia was established.

The adjustment principle is: (1) seismic data dispersion adjustment: deleting the observed seismic phase data that has a 4s deviation from the theoretical line; (2) depth adjustment: The depth is adjusted so that the actual seismic data is located in the middle of the theoretical line; (3) velocity adjustment: The actual data should be parallel to the theoretical line.

3 THE ESTABLISHMENT OF THE INITIAL MODEL AND THE OPTIMAL MODEL 3.1 Initial Model

The initial model of the western region was determined by means of velocity fitting, sub-regional scanning, and the method of determining the thickness and velocity of the crust through reduced travel time.

In the western region of Inner Mongolia (36°-46°N, 96°-115°E), 225 ML≥3.0 earthquakes (station number: N≥8) are selected. From the selected seismic events, we extracted a total of 2, 467 items of Pg data, 810 of Pn data, 89 of Pb data, and 3516 of Sg data. By seismic phase velocity fitting of Pn, Pb, Pg and Sg, we get the fitting curves of VPg, VPb, VPn and VSg, as shown in Fig. 3.

 Fig. 3 Velocities fitting of 225 earthquakes in western Inner Mongolia
3.1.1 Subdivision Scanning

In order to investigate the stability of the seismic velocity wave, sub-regional scanning was performed over the western Inner Mongolia with the step length set according to the frequency of earthquakes, and then the velocity mean was obtained. The specific results are shown in Table 1.

Table 1 Sub-regional scanning results
3.1.2 The Thickness and Velocity of the Crust Determined by Reduced Travel Time

Based on the mean velocity of Pg, Pb and Pn, the reduced travel time is calculated as:

 ${{T}_{\text{Z}}}={{T}_{\text{L}}}-(\Delta /V)$ (1)

Where, TZ is the reduced travel time, TL is the theoretical travel time, Δ is the epicentral distance, and V is the wave velocity.

Based on the South China crustal velocity model, using formula (1), we get reduced the P-wave travel time curve of the 225(N≥8)ML≥3.0 earthquakes. As seen from Fig. 4(a) and 4(b), Pg and Pn phase values are above the theoretical line, indicating that the Moho depth determined by the South China and Inner Mongolia optimal 2015 model is shallower, the distribution of VPg and VPn seismic phase is close to the theoretical line. Accordingly, we made the adjustment as follows: The first adjustment: (1) Moho depth adjustment, increasing the Moho surface depth to 46km (H1=27km, H2=19km); (2) velocity fine tuning, increasing V1 from 6.12km/s to 6.15km/s, with an increase of 0.03km/s. The second adjustment: (1) Moho depth adjustment, increasing Moho surface depth to 46km (H1=32km, H2=14km); (2) velocity adjustment, decreasing V1 to 6.07km/s from 6.12km/s, with a decrease of 0.05km/s. Finally, the actual data is in the middle of the theoretical line, and the overall distribution trend of the data is parallel to the theoretical line (Fig. 4(c), (d)).

 Fig. 4 Reduced travel time curve of P-wave (a) Mean value from South China model; (b) mean value from the optimal model of Inner Mongolia, 2015; (c) reduced curve by the first adjustment; (d) reduced curve by the second adjustment

The initial model of the one-dimensional velocity in the western region of Inner Mongolia is obtained by the combined seismic velocity fitting curve and the reduced travel time curve.

3.2 The Optimal Model for Western Inner Mongolia

Selecting 225 earthquakes of ML≥3.0 recorded by the Digital Seismic Network of Inner Mongolia, using the initial model and setting up several models with different values within the scope of its disturbance of the initial model, and adopting the Hyposat batch processing method, two searches and positionings were conducted (Table 3). The minimum average residual error of the first search is RMS=0.476, the minimum average residual error of the second search is RMS=0.463. Accordingly, we can determine the optimal model of the western Inner Mongolia by the Hyposat batch processing method, that is: V1=6.06km/s, VPb=6.61km/s, Vn=8.12km/s, H1=30m, and the Moho depth H=44km. In the same way, the average residual is RMS=1.307 using the South China model batch processing and RMS=0.961 by the optimal model of Inner Mongolia, 2015.

Table 2 Initial model

Table 3 Two search results of Hyposat batch processing

Combining the previous research results, the seismic phase fitting curves of earthquakes in western Inner Mongolia and the reduced traveltime curve, we get the initial crustal velocity model of western Inner Mongolia and the optimal velocity model by Hyposat batch processing, and in combination with regional geological tectonic characteristics in the western area of Inner Mongolia, the western Inner Mongolia model is established as: V1=6.06km/s, VPb=6.61km/s, Vn=8.12km/s, H1=30km, H2=14km.

The comparison between the western model and the best model of Inner Mongolia of 2015 and the South China model shows that: (1) V1, VPb and Vn are not much different from each other in the three models; (2) the difference in crustal thickness is small between the western model and the optimal model, but both are larger than the South China model (Fig. 5).

 Fig. 5 Comparison of models
4 TEST OF THE WESTERN INNER MONGOLIA MODELS BY HYPOSAT BATCH PROCESSING

Choosing randomly 88 earthquakes recorded by the Digital Seismic Network in Inner Mongolia, using the seismic velocity model of South China (catalog), in 2015 the optimal model and the western Inner Mongolia model, we adopt Hyposat batch positioning software to do the batch processing on the seismic data and get the earthquake parameters such as epicenter location, magnitude, focal depth and the original time, as well as the residuals.

A comparative analysis of location residuals and epicentral positioning deviation (epicentral deviation) was made among the three models.

4.1 Analysis of Location Residuals

By using the western model, the South China (catalog) model and the optimal model in 2015 to the location, comparison chart of location residual was obtained. From Fig. 6, we can see catalog positioning residual of the 88 earthquakes ranges between 0.077-5.231 seconds, the average being 0.707 seconds. The location residual by the 2015 optimal model is in the range of 0.059-5.248 seconds, with a mean value of 0.634 seconds. The residual range by the western model is 0.055-1.424 seconds and the mean value is 0.382 seconds. The mean residual of the western model is significantly reduced.

 Fig. 6 Comparison of location residuals between the western, optimal (2015), and cataloging (South China) models
4.2 The Test and Analysis of Epicenter Deviation

Fig. 7 shows the epicenter deviation contrast of 88 earthquakes between the "Western Model-Catalog", the "Optimal Model-Catalog" and the "South China Model-Catalog". In this figure, the epicenter deviation in the "Western Model-Catalog" is between 0.165-18.204km, with an average of 2.902km, and there is one earthquake with epicenter deviation > 10km; The epicenter deviation in the "Optimal Model Catalog 2015" is between 0.139 and 18.204km, with a mean of 3.207km, and there are 4 earthquakes with epicenter deviation > 10km. The epicenter deviation in the "South China Model-Catalog" is between 0.209-79.395km, with a mean of 5.487km, and there are 10 earthquakes with an epicenter deviation > 10km.

 Fig. 7 Comparison of epicenter deviation between"Western Inner Mongolia Model-Catalog, "Optimal Model-Catalog" and "the South China Model-Catalog"
4.3 Random Earthquake Sampling to Check the Reduced Travel Time (Fig. 8)
 Fig. 8 Reduced travel time of P-wave

Fig. 8 shows that the actual data is in the middle of the theoretical line, and the overall distribution trend of the data is parallel to the theoretical line. The western model reasonably and effectively reflects the crust velocity structure of western Inner Mongolia.

4.4 Random Earthquake Sampling for Focal Depth Check (Fig. 9)
 Fig. 9 Comparison of focal depths determined by different models

We determined the focal depths of the randomly sampled earthquakes using the western model, the Hyposat and PTD method, respectively, and the results show that the depths determined by PTD method are all below 20km, suggesting the western model is reasonable and stable.

5 CONCLUSIONS

Through the analysis of the one-dimensional velocity model of western Inner Mongolia, the following conclusions are drawn:

(1) Through linear fitting of seismic phases and reduced travel time curve of 225 ML≥3.0 earthquakes occurring in the region, and using the methods of discrete degree adjustment, depth, and velocity adjustment the initial model was obtained. Then using the initial model to search for the secondary Hyposat batch location results of 225 earthquakes, the western Inner Mongolia velocity model was obtained, that is, V1=6.06km/s, VPb=6.61km/s, Vn=8.12km/s, H1=30m, and Moho depth H=44km.

(2) The Hyposat location algorithm was used to test the results, the travel time residuals were significantly reduced, the epicenter deviation in earthquake location was decreased to some extent, and the deviation range was reasonable and stable, so the western model improves the accuracy of earthquake location.

(3) The located focal depth is reasonable, and the new model improves the location quality.

(4) The Hyposat method is sensitive to the velocity model, and the location results of different velocity models are different.

(5) The two-layer uniform velocity model in western Inner Mongolia is in line with the actual crustal structure in the western region of Inner Mongolia, which is suitable for earthquake location in the Inner Mongolia Digital Seismic Network.

REFERENCES
 Chen Xiangjun, Shangguan Wenming, Song Xiuqing, Wang Jun, Liu Shuangqing, Miao Fajun, Zhu Yuanqing. Study on crustal velocity model in Xinjiang and its subareas[J]. Earthquake Research in China, 2014, 30(2): 178–187. Li Xiang, Bao Dongjian, Chen Yuanping. The layered structure of the curst and mantle medium in central Inner Mongolia[J]. North China Earthquake Sciences, 1987, 5(4): 58–65. Liu Changquan, Jia Shixu, Du Guanheng. Result of seismic refraction sounding along the transect from Xiangshui, Jiangsu, to Mondula, Nei Mongol[J]. Seismology and Geology, 1991, 13(3): 193–204. Liu Fang, Miao Chunlan, Gao Yanling. Study of code Q value in Midwest region of Inner Mongolia[J]. Earthquake, 2007, 27(2): 72–80. Liu Fang, Jiang Changsheng, Yang Yanming, Zhang Wentao, Wang Lei, Liang Ying. Material integrity analyses of Inner Mongolia Earthquake Networks[J]. Seismological and Geomagnetic Observation and Research, 2013, 34(1/2): 176–183. Liu Fang, Diao Guiling, Han Xiaoming, Zhang Fan, Jin Xiaoxi, Ji Baorong. Primarily discussion about the seismogenic structure of Horqin Left Back Banner M5.3 earthquake on April 22, 2013[J]. Seismological and Geomagnetic Observation and Research, 2014, 35(5/6): 63–67. Liu Fang, Jiang Changsheng, Zhang Fan, Yang Yanming, Liang Ying, Wang Lei, Miao Chunlan. A study on detection capability of the Inner Mongolia Regional Seismic Network[J]. Acta Seismologica Sinica, 2014, 36(5): 919–929. Ren Xiao, Xu Zhiguo, Yang Hui, Chen Hongfeng, Zou Liye. Moho depth distribution character beneath the Ordos block's southeastern margin areas[J]. Chinese Journal of Geophysics, 2012, 55(12): 4089–4096. Sun Anhui, Chen Qifu, Chen Yong, Li Gang, Zhang Zhiqiang. Relocation of earthquakes in the northeastern Tianshan Mountains Area and improvement of local 1-dimensional crustal velocity model[J]. Earthquake Research in China, 2011, 27(3): 235–246. Yang Feng, Huang Jinli. High precision 3D P-wave velocity model of the upper crust under the Chinese capital region based on oil seismic stack velocity and deep seismic sounding[J]. Chinese Journal of Geophysics, 2013, 56(5): 1487–1496. Zhang Fan, Liu Fang, Zhang Hui. Relocation of the earthquakes in Daihai fault zone[J]. Seismological and Geomagnetic Observation and Research, 2014, 35(1/2): 69–74. Zhang Fan, Liu Fang, Zhang Hui, Han Xiaoming, Zhao Xing. Research on 1D crust velocity model in western and central Inner Mongolia[J]. Seismological and Geomagnetic Observation and Research, 2015, 36(4): 1–5. Zhang Tianzhong, Wu Bater, Huang Yuan, Jiang Changsheng. Effect of the data recorded at nearby stations on earthquake relative location[J]. Chinese Journal of Geophysics, 2007, 50(4): 1123–1130. Zhao Zhonghe. Earthquake Location of Regional Seismic Network. Lecture Notes of Unified Cataloguing Training Course for National Practical Technical Training of Seismic Network Personnel[M]. Beijing: China Earthquake Networks Center, 2010 (in Chinese). Zhu Yuanqing, Fan Changqing, Pu Xiaofeng. Accurate determination and analysis on the temporal and spatial parameters of the South Yellow Sea earthquake sequence[J]. Earthquake Research in China, 1995, 11(1): 54–61. Zhu Yuanqing, Xia Congjun, Li Ping. A PTD method a new method for determinating focal depth and its application[J]. Seismological and Geomagnetic Observation and Research, 1997a, 18(3): 21–29. Zhu Yuanqing, Zhao Zhonghe. Research on the new method to raise earthquake location accuracy[J]. Seismological and Geomagnetic Observation and Research, 1997b, 18(5): 59–67. Zhu Yuanqing, Yu Haiying. Reverse station method of locating regional earthquake for Shanghai Seismic Array (Network)[J]. Seismological and Geomagnetic Observation and Research, 2002, 23(2): 97–106. Zhu Zhiping, Zhang Xiankang, Zhang Jianshi, Zhang Chengke, Zhao Jinren, Gai Yujie. Research of velocity structure of the crust and upper mantle along Beijing-Huailai-Fengzhen DSS profile[J]. Acta Seismologica Sinica, 1997, 19(5): 499–505.