Earthq Sci (2018)31: 234–241 234 doi: 10.29382/eqs-2018-0234-3 Phase velocity maps of Rayleigh waves in the Ordos block and adjacent regions* Shaoxing Hui1,* Wenhua Yan1 Yifei Xu1 Liping Fan2 Hongwu Feng1 1 Shaanxi Earthquake Agency, Xi'an 710068, China 2 Institute of Geophysics, China Earthquake Administration, Beijing 100081, China Abstract We analyze continuous waveform data 1 Introduction from 257 broadband stations of the portable seismic array deployed under the "China Seismic Array-northern part of NS In recent years, one of the important methods for seismic belt" project as well as data from a permanent seismic network from January 2014 to December 2015. The phase obtaining the S-wave velocity structure of the crust and velocity dispersion curve of 7,185 Rayleigh waves is obtained upper mantle is ambient noise tomography. Its main with a method based on the image analysis of phase velocity advantages are free of uncertainties of earthquake source extraction, and the inversion is obtained. The period of and a better ray coverage, more dispersion curves with Rayleigh wave phase velocity distribution has a range of 5– moderate and short periods are obtained, and finally the 40 s, and minimum resolution close to 20 km. The results distribution of the observation stations is an important show that the phase velocity structure image well reflects the factor in its resolution. With a broad and dense distribution geological structural characteristics of the crust and of observation stations, the deep tectonic environment can uppermost mantle, and that the phase velocity distribution has be detected, and the crustal structure with better resolution obvious lateral heterogeneity. The phase velocity of the 5– can be achieved in the region with weak seismicity. This 15 s period is closely linked to the surface layer and sedim- method has been widely used in many areas (Fang et al., entary layer, the low-velocity anomalies correspond to loose 2013; Fan et al., 2015; Yan et al., 2016; Guo et al., 2017). sedimentary cover, and the high-velocity anomalies corres- The Ordos block and adjacent regions are located in pond to orogenic belts and uplifts and the boundary between the transitional area between the northeastern margin of high and low velocity anomalies is consistent with the block boundary. The phase velocity of the 5–15 s period is strongly the Tibetan Plateau and the North China block. The main affected by the crust layer thickness, the northeastern Tibetan tectonic units in the area include the Alxa block, the Ordos plateau has low-velocity anomalies in the middle to lower block, the Fenwei graben, and the Qinling mountain, and crust, the west side of the Ordos block is consistent with the the area has been affected by the collision of Indian and northeastern Tibetan plateau, which may imply the material Eurasian plates. However, the relatively stable Alxa and exchange and fusion in this area. The velocity variation is Ordos blocks have considerably blocked the push of the inversely related to the Moho depth in the 40 s period of S- Qinghai block (Chao, 2002; Chang et al., 2011). The wave, and the lateral velocity heterogeneity represents the seismogenic structure in this area is intricate and complex, lateral variation of the Moho depth. The Ordos block and the and there is a strong neotectonic movement. Historically, northern margin of Sichuan basin are located in the uppermost strong earthquakes have been very active, and there have mantle at this depth, and the depth in the transition zone is been many devastating medium and strong magnitude still located in the lower crust. earthquakes (Xu et al., 2000). In recent years, different researchers have carried out a lot of study in the region and Keywords: Ordos block; ambient noise tomography; Rayleigh obtained some important results about the deep structure of wave; phase velocity the region (Pan et al., 2017; Ding et al., 2017; Zheng et al., 2018). As a result, the lower crust flow, crustal growth, and low-speed anomaly in the crust were discussed in * Received 28 August 2018; accepted in revised form 22 November detail. However, resolution has been limited because there 2018; published 11 April 2019. is no combination of permanent and portable stations data * Corresponding author. e-mail: [email protected] © The Seismological Society of China and Institute of Geophysics, in previous study, the fineness of existing research results China Earthquake Administration 2018 needs to be further improved. Earthq Sci (2018)31: 234–241 235 Since September 2013, the Shaanxi Earthquake Admini- each processing step. stration and other departments have deployed 160 broad- 2.1 Data preprocessing and cross-correlation band portable seismographs in central and western Shaa- function calculations nxi, and we collected a total of 257 seismic stations data, including 160 portable broadband stations of the Himal- We first converted the continuous waveform data from ayan array (Phase II) (ChinArray-Himalaya, 2011) and 97 Seed and Pascal files to SAC data files, merged the seismic permanent stations of China Seismic Networks, with an data for each station, split it into hourly segments, and average spacing of about 30 km (see Figure 1). We obta- resampled from 100 Hz to 1 Hz. Next, moved the instru- ined 5–40 s phase velocity maps of Rayleigh wave based mental response, removed the mean and trend of the on ambient noise tomography in Ordos and its adjacent seismic data, then band-pass filtered the seismic data in the regions, our research utilized the observation data of the 5–50 s period. To eliminate the interference from the highest density and the largest number of seismic stations single-frequency signal and make the ambient noise of the in this area. The new research results provide an important signal band more continuous (Bensen et al., 2007), we reference for understanding important issues such as the performed temporal normalization and spectral whitening crust and upper mantle structure, tectonic plate boundaries, for the data to remove the unsteady noise signal around and crust-mantle coupling and its deformation mechanism. each station and signal distortions caused by instrument malfunction. Finally, we calculated cross-correlations and 106°E 108°110° 112° 114° stacked hourly cross-correlation functions to obtain the 38°N time domain cross-correlation function for each station pair (Figure 2). It can be seen from the figure that the symmetrical component has a higher signal-to-noise ratio than the asymmetric component at each period. The more 36° Ordos block cross-correlation functions are stacked, the stronger the surface wave signal, and the lag time of the surface wave North China block signal is positively correlated with the interstation Fenwei graben 34° distance. When the interstation distance is relatively small, Qinling fold belt the body wave precursor signal in the surface wave signal is relatively obvious (Wang et al., 2011). Northeast margin of Tibetan plateau Tibetan of margin Northeast 700 32° South China block 600 Portable station Paleoearthquake (M>7.5) 500 Permanent station Plate boundary Figure 1 A sketch map of the boundaries and seismic 400 stations in the Ordos block and its adjacent regions 300 2 Seismic data and methods Distance (km) 200 We collected continuous broadband seismic data of 100 the vertical components from 257 seismic stations from January 2014 to December 2015. The continuous seis- 0 mic data was recorded in the portable broadband stations −500 0 500 using Güralp CMG-3T seismometers (0.02–120 s) and Time (s) CMG-3ESP seismometers (0.02–60 s) equipped with Figure 2 Cross-correlation waveforms between station 61012 Reftek-130 seismic recorders. The average spacing bet- and others ween stations was 30 km. All the stations used GPS timing 2.2 Phase velocity measurement to ensure consistency. The station distribution is shown in Figure 1. The data processing flow consists primarily of The phase velocity dispersion calculation formula data preprocessing, calculations of the cross-correlation (Dahlen and Tromp, 1998) is functions, Rayleigh wave phase velocity measurements, ∆ 2D phase velocity tomography, and analysis of the CAB (T) = (1) t − T inversion results. The following is a brief description of 8 236 Earthq Sci (2018)31: 234–241 8 CAB(T) is the average phase velocity, Δ is the inter- station distance, T is the period, and t is the travel time of 7 ) 3 the peak after filtering the empirical Green’s function near 6 period T. Interstation phase velocity dispersion are measured using rapid analysis software based on an image 5 transformation technique proposed by Yao et al. (2004). 4 The efficiency of this software is far superior to that of 3 manual extraction. Moreover, the software quickly impro- 2 ves the measurement accuracy of the phase velocity Number of measurements (×10 dispersion while quickly tracking the entire phase velocity 1 dispersion curve and is reliable and fast for batch proce- 0 ssing ambient noise. In order to obtain a reliable and 510 15 20 25 30 35 40 Period (s) accurate extraction phase velocity dispersion curve, the following three conditions must be satisfied simultaneou- Figure 3 Number of rays for inversion at different periods sly: (1) the ambient noise must be continuous waveforms 2.3 Inversion method of more than 1 year for the cross-correlation calculation, (2) cross-correlation functions with low signal-to-noise In this study, 2D Rayleigh wave phase velocity maps ratios (SNR) must be removed, so we only use the cross- for different periods are inverted using the method correlation functions with signal-to-noise ratios greater proposed by Ditmar and Yanovskaya (1987), Yanovskaya than 10 (Bensen et al., 2007; Fang et al., 2010) for the and Ditmar (1990)).