فهرست مطالب taghi shirzad

Seismic interferometry is an efficient technique to extract the Empirical Green's Function (EGF) between station pairs when the source is considered at one of the stations. The geometry and energy flux of asymmetric noise sources have unavoidable impacts on the extracted EGFs, deduced from ambient seismic noise recorded in pairs of stations. In this study, to consider these effects, three methods of noise correlation functions stacking (linear, root mean square, root mean square ratio) are investigated using synthetic and real data processing. During synthetic data processing, effects of the noise sources geometry and energy flux inside and outside the Fresnel zone are examined. After separating stationary and non-stationary sources, the results have shown that the root mean square ratio contains the least effects of non-stationary signals compared to other methods of stacking. Moreover, comparison of the EGFs from the recorded data in Azerbaijan (NW Iran), indicates that the signal retrieved by root mean square ratio is more reliable than the other stacking methods' signals (e.g., linear, root mean square).

Analysis of earthquake events provides an efficient tool to extract the empirical Green's function (EGF) between pairs of earthquakes by interferometry approach. Because of sparse distribution of stations or low seismicity, many classical seismic studies (earthquake-receiver systems, ambient seismic noise and etc.), may yield a poor or noisy calculation of the tomographic result maps. However, inter-event EGFs between two earthquake locations can be retrieved by virtual stations, first outlined by Curtis et al. (2009). These EGFs are equivalent to the waveform produced as an impulse at one receiver location and that recorded by another receiver. Several researchers (e.g. Hong and Menke, 2006; Curtis et al., 2009) used a source-receiver reciprocity theorem, to indicate that inter-event EGFs could be retrieved when their waveforms are recorded by a set of receivers surrounding the events. According to this theorem, the stacked cross-correlations of event-pair (between a pair of earthquake event) waveforms recorded by these set of receivers, are equivalent to the estimated EGFs (Curtis et al., 2009). This technique can, therefore, provide new insight and useful tool to study fault planes and Earth's interior where real receivers can not be installed. However, the fault plane where earthquake ruptures occur at depth is often interpreted as being a transitional zone which is characterized by asperities and barriers (Aki, 1984). Thus the aftershock events interferometry approach could be applied to study fault plane by retrieving accurate, stable and reliable inter-event EGFs. After the Rigan earthquake occurred on 20 December 2010 (Mw 6.5) in Kerman province (southeastern Iran), aftershock events extended along the hidden part of the Kahurak Fault. In this paper, the cross-correlation of aftershock events was applied to retrieve the inter-event EGF on the hidden part of the Kahurak Fault plane in the Rigan area. This event-pair example was selected based on some criteria that the most important of these conditions is the similar (approximately) depth of events due to the ease of operation and processing. Aftershock event-pair projection and data processing is similar to that explained in detail by Bensen et al. (2007). The mean and trend were removed and the data were decimated to 10 sps. Time and frequency domain normalizations were then applied to suppress the influence of instrument irregularities and high energy events. After cross-correlation and stacking procedure, event-pair EGF signal was extracted. Then, 1D and 2D synthetic signals were generated using computer program in seismology (Herrmann and Ammon, 2013) and SPECFEM , respectively. Horizontal velocity result at depth of ~4 km, which is calculated by Shirzad et al. (2017), was applied for both 1D and 2D synthetic input modeling. Comparison between inter-event EGF and synthetic signals indicates that the inter-event EGF is in agreement with the synthetic models. Also, inter-event EGF signal propagates on the hidden part of the Kahurak fault plane. The correlation coefficient of 1D and 2D synthetic inter-event 031-163 EGF signals are of the order of ~75% and 80% within the signal window. In conclusion, these inter-event EGFs can be used for investigating the laterally varying the 2D mapping of surface wave group and/or phase velocities.

Various studies have shown that the cross-correlation (Wapenaar, 2004), cross-convolution (Slob and Wapenaar, 2007) and de-convolution (Wapenaar et al., 2008) can provide empirical Green’s functions (EGFs) between receiver pairs. These approaches, which are attributed to seismic interferometry, assume that one of the receiver acts as a source, whereas the other one is instated as a virtual receiver. The resulted EGFs allowed many studies to be applied in different regions even though (including) areas with low seismicity.
The main assumption of interferometry approach is based on completely diffuse signals which are generated by a closed surface of sources (Schuster et al. 2004). In other word, the distribution of sources and theirs energy in a medium are usually uniform. This condition ensured that inter-station EGF is extracted accurately. In general, the sources (left and right of the receivers) located on or near lines which is passed through both receivers are in the stationary region, and the sources above and below are in non-stationary regions. Also, Snieder (2004) indicated that the Fresnel zone of receivers surrounded all the sources which are located in stationary region. In this study, we referred to these sources as stationary sources. Consequently, all sources outside the Fresnel zone were referred to as non-stationary sources. It is generally accepted that the stationary sources play a major role/contribution to retrieve the inter-station EGF. Stationary sources and their energies are characterized by coherency and small wavenumber. In contrast, non-stationary sources and their energies are characterized by incoherency, larger wavenumber. We used this difference in order to separate stationary and nonstationary sources.
In the Earth, distribution of noise sources and theirs energy are strongly non-uniform, which contravenes the theoretical interferometry requirements (Stehly et al. 2006). In other words, cross correlations from non-stationary sources in stacking procedure do not cancel completely if the source coverage is incomplete. Consequently, incomplete source coverage leads to retrieve unreliable inter-station EGF.
In this study, we used 144 sources on circle environment (r=40 km) surrounded by two receivers which are located/installed in A(-4 , 0) and B(4 , 0) as shown in Figure 1. Moreover, synthetic time series were generated using Mexican-hat source time function (see left panel of Figure 2). All recorded waveform signals of these sources in station A and B are shown in middle and right panel of Figure 2, respectively. After the preprocessing procedure and cross-correlation performances, we constructed a cross-correlogram matrix, which is called CC, using a collection of cross-correlation function signals (see left panel of Figure 3). The dimensions of this matrix include time (number of point in signal data set, npts) and source counter/numerator. In brief, inter-station EGF is retrieved using stacking the cross-correlogram matrix signals along the source counter dimension. We followed the analysis and preprocessing of the cross-correlogram before stacking outlined in Poliannikov and Willis (2011). Thus, we decompose cross-correlogram matrix using singular value decomposition (SVD) technique to separate the stationary and non-stationary energies. This idea illustrates/explains that the cross-correlogram matrix could be calculated by its eigenvalues and eigenvectors. Poliannikov and Willis (2011) indicated that the large eigenvalues (singular values) are associated with events which is located in Fresnel zone. Afterward, we constructed lower-rank approximations of the cross-correlogram matrix using two larger eigenvalues, which is called CC2, and then stack CC2 along the source counter dimension to retrieve inter-station EGF (see Figures 4, 5 and 6).

There has been wide interest in ambient seismic noise studies for determining earth’ internal structures in the recent years. Ambient seismic noise contains waves with random amplitudes and phases which propagate in all directions (Van-Tighelen, 2003; Gorin et al., 2006). Therefore determining information of waves propagations is possible by extracting coherence signal. This information of propagation path is equal to Green’s function (Shapiro et al.,2005; Roux et al., 2005; Sabra et al., 2005). Ambient seismic noise method is applied in various researches such as acoustic, helioseismology, seismology, etc (Duvall et al., 1993;Rickett and Claerbout, 1999; Malcolm et al., 2004; Roux et al., 2004).
The isotropic and random noise source distribution is the basic assumption underlying retrieving empirical Green’s functions (hereafter EGFs) using this method (Weaver and Lobkis, 2001; Gouédard et al., 2008). Recent studies surrounding noise sources demonstrate the dominant presence of noise sources in oceanic regions (Stutzmann et al., 2009; Landes et al., 2010). Ambient seismic noise spectra contains two broad spectral peaks, one at the period of 17 s (the primary microseism), and the other at the period of 7 s (the secondary microseism) (e.g., Gutenberg, 1936; Berger et al., 2004).
Regarding the dominant presence of noise sources in oceanic regions and also sharp seasonal variations, noise sources distribution is non isotropic and directive (Stehly et al., 2008). Nevertheless, distribution of noise sources homogenizes when considered over long times (Snieder, 2004).
The randomization of the wavefield is enhanced by the scattering of the seismic waves on the small scale heterogeneity within the Earth (Shapiro and Campillo, 2004). Scattered coda waves, sampled randomly and repeatedly parts of wave propagations, similar to ambient seismic noise (Yao et al., 2006). Therefore scattered coda waves, contain valuable information about propagation properties of the media. Additionally these waves are also independent from distribution of noise sources (Stehly et al., 2008; Froment et al., 2011). Scattered coda waves energy flux, is equiparitioning of ambient seismic noise and are independence from distribution of noise sources (Shapiro et al., 2000; Margerin et al., 2009). Stehly et al. (2008) studies, illustrate that retrieving EGFs is possible from scattered coda waves part of noise correlation functions (hereafter NCFs), which was assigned as C3 method in brief. The C3 method is an efficient way, facing poorly oriented station pairs with directional energy flux of ambient seismic noise. Therefore the accuracy of estimating arrival times of the different parts of EGFs is improved by C3 method in the presence of inhomogeneous noise source distribution (Garnier and Papanicolaou, 2009; Froment at al.,2011).
The purpose of this study is retrieving EGFs by C3 method in the period bands of 1-3 and 3-10 s in Azerbaijan region. We processed vertical component recording of continuous data from 7 stations which are equipped with short period sensor (Kinemetrics SS-1) in Azerbaijan region (Figure 1). We use 1 year (Dec. 2011-Dec. 2012) of recording at these stations which are operated by the Iranian Seismological Center (IRSC) of the University of Tehran. NCFs were determined by preparation of raw data (i.e. removing the mean and trend, decimation, segmenting, time and frequency domain normalization). Rms-stacking method (see Shirzad and Shomali, 2013) was applied for all NCFs calculated for retrieving daily and total EGFs from ambient seismic noise method (C1). In this study, we investigate three types of NCFs including: (a) a coda wave signal window selected from NCFs which was calculated from raw data (b) a coda wave window identified from the subset of NCFs, which contributed to the rms-stacking method (c) a coda wave signal window selected from the subset of NCFs, which was subsequently used in daily EGFs from C1 method, in retrieving optimized EGFs by C3 method. We compared two parameters (including correlation coefficients and arrival time of Rayleigh waves fundamental mode) between extracted EGFs from C1 and C3 methods. Table 2 shows the results of this investigation. Analysis of this table shows that the standard deviation of the arrival time Rayleigh waves and correlation coefficients are 0.21, 0.98 in positive lag-time and 0.35, 0.96 in negative lag-time respectively. The results showed that all extracted EGFs using three types of coda wave signal windows were significantly similar in character. However, to save time and reduce the amount of calculations, we selected the first case i.e. using NCFs which was calculated from raw data for further processing (see table 1). In the similar way with C1 method, coda wave windows were stacked with rms-stacking method in monthly and yearly time intervals. Figure 8 shows, the monthly EGFs retrieved by C3 method which illustrate negligible (no) directionality in the region of study. Yearly (total) EGFs versus interstation distances in the period bands of 1-3 and 3-10 s, were depicted in Figure 9. Arrival time of Rayleigh waves fundamental mode is equal (to 2.09±0.04 (km/s) in the region of study.

Recent progress in seismology has demonstrated that empirical Green’s functions (EGFs) of inter-station distances can be extracted using cross correlation of ambient seismic noise recorded in the similar time at two stations (Weaver and Lobkis، 2002; Shapiro and Compillo، 2004; Wapenaar، 2004). Consequently، this method provides a great set of data even in low seismicity regions to apply in the tomographic studies. Thus، the resulted tomographic images using the ambient seismic noise method (hereafter ANT) can show interior earth structures with a higher resolution compared to classical tomography methods (Shapiro et al، 2005; Lin et al.، 2007; Shirzad et al، 2013). Diffused signals are the main assumption in the ANT method (Snider، 2004). Ambient seismic noise sources generate a coherent and transient noise wavefield with random amplitude and phase in a medium (Van-Tighelen، 2003; Gorin et al.، 2006). Reconstruction of the propagating path information using the amplitude of the recorded noise wavefield is impossible، but coherent information provided by propagating path can be extracted using cross correlation of long time ambient seismic noise recorded (Weaver and Lobkis، 2004; Gorin et al.، 2006). This coherent information is called elastic response of medium or empirical Green’s functions (Shapiro and Compillo، 2004; Roux et al.، 2005; Sabra et al.، 2005). Generally، the ambient seismic noise recorded for each station is composed of surface waves (Rayleigh and Love) with random amplitude and phase (Aki and Richards، 1980). Cross correlation function of these data will be symmetric if the ambient seismic noise wavefields generated by random sources are distributed uniformly (Snider، 2004). Earth structures can be studied using travel-time of extracted EGFs such as Rayleigh wave fundamental mode (Shapiro et al.، 2005). Some studies (e. g. Stehly et al.، 2006; Pedersen et al.، 2007) indicate that the inhomogeneous distribution of the signal energy in various azimuths، which results in directionality of ambient seismic noise، produces deviation in tomography results and causes incorrect interpretations. Consequently، optimization of extracted tomographic maps based on the ANT method needs comprehensive knowledge of spatial and seasonal distribution of the noise wavefield in study areas (Stehly et al.، 2006; Pedersen et al، 2007). Gutenberg (1936) suggested that the sources of primary and secondary oceanic microseisms observed throughout the Europe are located in the northeastern Atlantic Ocean. Primary and secondary microseisms dominate the noise wavefield in certain frequency ranges. The interaction between the swell and the sea bottom generates the primary microseisms which are dominated by periods of 12–25 s. Also، interfering water wavefield components travelling in opposite directions generate the secondary microseisms which are dominated by periods of 5-10 s (Gutenberg، 1936). In this study، we analyzed three-component recordings of continuous data from 30 stations in the Central Alborz region depicted in Figure 1. The Alborz Mountain range in the southern margin of the Caspian Sea is a part of the Alpine–Himalayan orogenic belt. The Alborz Mountain range resulted from a stress state derived from the horizontal compressive forces of the Central Iran Plateau has been induced by the collision of the Arabian plateau and the Asian continent (Berberian and King، 1981; Zanchi et al.، 2006). The dataset used in this study consisted of 10 digital accelerometers with CMG-5TD sensors operated by the Tehran Disaster Mitigation and Management Organization (TDMMO)، 18 digital narrow-band seismometers with SS1 seismometer sensors (corner frequency ≥1 Hz) operated by the Iranian Seismological Center (IRSC) at the University of Tehran، and two digital broadband instruments with a CMG-3T sensor operated by the International Institute of Earthquake Engineering and Seismology (IIEES). For the TDMMO acceleration network، the IRSC and the IIEES seismic networks continuous data from 2010 were analyzed. In the case of azimuthal distribution of the ambient noise، normalized amplitude of the cross-correlations versus azimuth (rose-diagram) constrained the direction to the sources of the ambient seismic noise، based on all available station-pairs. The average fractions (the number of Love/Rayleigh path with a SNR>10 in a given 20° azimuthal bin was normalized to the total number of Love/Rayleigh paths in that given bin) of the Love and Rayleigh yearly empirical Green’s functions with a SNR>10 were in the orders of 0. 78 and 0. 73، respectively، at the period band of 1–10 s. Our final results indicated that the average fractions per cent of Love and Rayleigh paths with SNR>10 were above 68% and 64% on a yearly scale، and never decreased to 45% and 50% on a monthly scale at the period band of 1-10 s، respectively.