Estimation of Coda Q Wave Attenuation Factors for Northern Iraq

Abstract


Introduction
Attenuation is one of the fundamental properties of seismic waves from which the material and physical conditions in the Earth's interior can be inferred (Aki, 1980).The decay of seismic wave amplitude with distance defines the attenuation of the medium.The energy of seismic waves decays due to geometrical spreading, intrinsic and scattering attenuation (Padhy et al., 2011;Sedaghati and Pezeshk, 2016).Seismic attenuation is usually considered to be a combination of two mechanisms: intrinsic absorption and scattering loss.As seismic waves propagate through the Earth's interior and finally arrive at seismic stations on the surface, their energy decays due to the conversion of elastic energy to heat (intrinsic attenuation) as well as energy redistribution in the heterogeneities present in the lithosphere called scattering (Brahma, 2012).
The attenuation property of the medium is usually measured by a dimensionless quantity called the quality factor Q, which is defined as the ratio of stored energy to dispersed energy.It measures the relative energy loss per oscillation cycle (Aki and Chouet, 1975).The attenuation of the medium is inversely proportional to the quality factor (Q), which means that seismic waves are highly attenuated in regions having lower Q values.
There are two models for interpreting coda waves (Aki and Chouet, 1975).The first is the single scattering model, according to which waves are backscattered S waves and generated when an S wave encounters different heterogeneities present in the propagating media in which the scattering process is weak.The second model is the multiple scattering model, and this model is mainly used for long >100 km lapse times (Kumar et al., 2005).There are several ways to calculate the quality factor of a seismic wave at each station.The widely used coda normalization method (Aki, 1980;Sato et al., 2012) calculates the Q of S waves using the spectral ratio of S and coda waves.The other is the coda wave decay (CWD) method.This method is not appropriate for long lapse times due to a lack of high-frequency content.The CWD method used in this study tests the effect of processing parameters and obtains a reliable coda estimate Q.So, the results can be easily compared among different stations.To determine seismic wave attenuation, numerous studies have been conducted in various places in the world, and the single backscattering model, proposed by Aki and Chouet (1975), is used for describing the behavior of the coda waves from local earthquakes (Kumar et al., 2005;Ma'hood and Hamzehloo, 2009;Rahimi et al., 2010;Sertçelik, 2012;Brahma, 2012;Birch et al., 2015;Sedaghati and Pezeshk, 2016;Bora et al., 2018).Al-Sinawi and Al-Tikriti (2000) presented the first study on seismic wave attenuation in Iraq using P-wave data from earthquakes that originated in Turkey and Iran and were recorded at observatories in Baghdad (BHD) and Mosul (MSL).The attenuation was determined by using 91 events with an epicentral distance of 30-100 km that were recorded by four stations: Baghdad, Mosul, Rutba, and Sulaymaniyah, with the aim of identifying the structure of the Earth crust in Iraq.Four different methods were used to estimate the cortex.The overall mean attenuation value obtained was α = 0.0035 km-1 (Alsinawi, 2002).Where the attenuation values were calculated in the Taktaq area in north central Iraq using the field seismic intensity map (Fahmi et al., 1989).In this research, seismic wave attenuation was studied using the single back-scattering model of Aki and Chouet (1975), extended by Sato (1977).The tail or end of a seismogram recorded at a short distance from an earthquake after the arrival of major wave types was analyzed (Aki and Chouet, 1975).Consequently, it calculated quality factor (QC) values as a function of frequency for epicentral distances < 100 km.The QC values are calculated for a time window of 30 seconds, and the coda window usually starts with twice the travel time S from the origin time (Havskov and Ottemöller, 2010).
A study area is located in northern Iraq.Numan (1997) described a set of listric normal faults in the basement rocks beneath the sedimentary cover in northern Iraq.These faults were tensional normal faults that resulted from the opening of the Neo-Tethys Ocean in the Early Triassic.The fault planes (dipping mostly to the north and northeast) are steep at shallow depths and curve to a subhorizontal attitude deep into the crystalline basement.The closure of the Neo-Tethys Ocean and subsequent continental plate collision created a compressional environment, which led to a reversal of movement on the originally normal listric faults, sedimentary basin inversion, and the shaping and tightening of folds in the Foreland Folds Belt.According to Fouad (2010), Iraq is divided into two main tectonic units: an inner and an outer platform.The outer platform is unstable and covers the Mesopotamian foredeep and a Zagros fold-thrust belt.An outer platform has surface folds at a Mesopotamian foredeep and a Zagros fold-thrust belt.The Zagros fold-thrust belt consists of four subdivisions (from northeast to southwest): the Suture Zone, Imbrication Zone, High Folded Zone, and Low Folded Zone.The study stations are located within the low-folded zone and high-folded zone.
This study aims to calculate the attenuation factor for northern stations and to comprehend the details of the structure of the earth's crust in that region.Then, to test if they are homogeneous or non homogeneous, through quality factor(QC) values with frequency.

Materials and Methods
The research was carried out at stations located in northern Iraq (Table 1).Based on the epicentral distances between the stations and the seismic events of approximately 100 km, we started analyzing 77 seismic waveform data, as shown in Table 2. Two networks are used to take the events.The first is the Mesopotamian Network (MP), which records data from September 2007 to September 2021 using stations SLY1 and DHK1.These are broadband stations that record 100 samples per second and have magnitudes ranging from 4 to 6.1 ML.The second network is the Iraqi Meteorological Organization and Seismology (IMOS), which recorded data from May 2012 to February 2021 from the station IKRK , and of magnitudes ranging from 4.2 to 5.1 ML The data that is available and has been analyzed is 10, 20, and 100 samples per second readings.
Local events and the vertical component of the Coda Q estimate are used in this research.Data can be in SAC or Seisan formats.The header file for all files is required before carrying out the analysis software.The locations of these stations are shown in Table 1.The stations and epicenters that were used for each, were scaled by a magnitude according to The International Seismological Center Web (ISC) catalog (Fig. 1).The seismic analysis Seisan software was used, developed at the University of Bergen, Norway (Ottemöller et al. 2011).It includes a full suite of programs and a simple database for analyzing digital recordings.It can be used, among other things, for phase picking, spectral analysis, plotting seismograms and azimuth determination.Seisan is supported by DOS, Windows 95, SunOS, Solaris, and Linux, and contains conversion programs for the most common data formats.The program can be downloaded from the manual's front page via the link "Download programs and files" (Ottermöller et al., 2011).
The CWD method is the most widely used method to estimate coda Q.It is used the single backscattering model developed by Aki and Chouet (1975) and Sato (1977), and is mostly used to describe the behavior of coda waves from local earthquakes.According to Havskov and Ottemöller (2010), coda waves are made up of incoherent waves that are scattered by inhomogeneities, and their amplitude only appears to decrease as a result of attenuation and geometrical spreading.The coda amplitude decay can be expressed as: (1) In which A0 is the initial amplitude, t lapse time, f frequency, β is the geometrical spreading parameter (body waves are 1.0 and surface waves are 0.5), and QC is the quality factor.Taking natural logarithms and rearranging them as: (2) When plotting the envelope of Ln [A (f, t) + β Ln (t)] as a function of t for a specific center frequency by band pass filter, the signal around the center frequency gives a straight line with a slope πf /Qc (f), and QC (f) may be determined using least squares.The envelope is calculated by the root mean square values in a running window of a specified length starting at a given time, the lapse time, and after the origin time.
All studies of coda Q show that attenuation increases with frequency (Havskov et al., 2016) by following the law: (3) Q0 is Q at the reference frequency f0, and α is a constant.All studies assume f0 =1Hz.As a result, the equation becomes: QC = Q0 f α (4) Q0 is QC (quality factor) at 1 Hz, and α represents the degree of frequency dependence.

Data Analysis
The SEISAN software has been used to analyze seismic data using the vertical components (Z) of different local events.The software must determine the event's origin time and P arrival time.Before beginning the analysis software, each seismic event must have the required headers, as illustrated in Fig. 2. It is considered input data to run the CODA program.The coda program requires data to be recorded at 100 samples per second, except for the data from the Kirkuk station, which some recorded in 10, 20, and 100 samples per second.
The seismogram is filtered using the band-pass of eight center frequencies (1,2,3.5,6,8,10,12,and 16 Hz) as shown in Table 3.The coda window used for seismic data analysis is 30 seconds.Rautian and Khalturin (1978) observed the amplitudes of band-pass-filtered seismograms and found that coda amplitudes decay with a common shape at all the stations for windows, starting mostly after about two times and always after three times the S-wave travel time.Therefore, most Coda Q studies use a lapse time of at least twice the S-wave travel time.The central frequency determined for each coda Q determination is used to filter the signals.To prevent ringing, the bandwidth increases in proportion to the frequency, resulting in a constant relative bandwidth.The signal-to-noise ratio for each filtered seismogram is calculated using the RMS amplitude of the last 5 s of data in the lapse time window and the noise data of the same window length before the P-wave arrival.Any positively sloped coda amplitude decay was ruled out of the event.Seventy-seven local earthquake coda values have been calculated using the equation with the QC values obtained as QC = Q0 f α , in which Q0 is Qc at 1 Hz and α is the degree of frequency dependence.QC would not be calculated for one seismogram and a single station, but rather as an average of events for the same station to get a single value of the

Results and Discussion
The QC values are calculated as a frequency function for each station (SLY1, DHK1, and IKRK).The QC values were obtained from the central frequencies using the SEISAN program, as shown in Table 4, as one event was mentioned for each station.These events (2, 4, and 11) are given for the stations SLY1, DHK1, and IKRK, respectively, in Table 2. Total Qc measurements for each station (SLY1 293,DHK1 206,and IKRK 57), for 30 sec coda window length estimated from the linear regression of Ln [A (f, t) + β Ln (t)] vs. t plot, for a specific center frequency by band pass filter.The obtained QC values at 1 Hz differed from those at 16 Hz.The distributions of quality factor (QC) values with frequency are shown in Fig. 5.The general trend in Fig. 5 shows that QC values follow a power law of QC = Q0 f α .In Table 5, a summary appears showing the general attenuation of each station for all the events worked on, with the estimated values Q0 and standard deviation .1.04 0.09 In all three stations (SLY1, DHK1, and IKRK), the QC values increase as the frequency increases, as shown in Fig. 5.The frequency-dependent parameter α is not constant, other studies have discovered that values rise in tectonically active regions (Bora et al., 2018).In Table 5, by displaying the work results for the three stations, the values α (1.02, 1.04, and 0.97) of SLY1, IKRK, and DHK1 were found.Suppose α had considered as a tectonic activity indicator.It was observed that the three stations, Kirkuk, Sulaymaniyah, and Dohuk, are close in terms of tectonic activity.The obtained values, in summary, indicate that the stations are very heterogeneous and tectonically active.The non-homogeneity could be attributed to a change in the geometry of the Arabian plate edge collisions with the Iranian and Turkish plates (Al-Sinawi and Al-Heety, 1997).
The attenuation parameter Q0 varied between 88 and 91.A lower Q0 corresponds to high attenuation.It is well known that the low value of Q0 implies higher seismic activity.These lower Q0 values reflect scattering effects either due to the dense faulting or because of the complex structure.Observations of the exponent α, describing the frequency dependence of 1/QC in regions of the world, seem to indicate that higher values of α are consistent with frequency-dependent exponents found in tectonically active areas (Singh and Herrmann, 1983).A correlation between 1/Q0 and α seems exist worldwide, α appears to decrease as 1/Q0 decreases.Fig. 6 shows the results for the stations in Kirkuk, Sulaymaniyah, and Dohuk.The figure shows an easier comparison of the results of the Coda Q as a function of frequency to those of three stations.Through the values of Coda, are notice the highest attenuation rate in the Sulaymaniyah station, followed by the Dahok and Kirkuk stations.According to the geology of Iraq, the study stations are located on the unstable shelf according to Jassim and Goff (2006), where northern Iraq is characterized by high seismic activity.The tectonically similar stations also have similar coda Q (Havskov et al., 2016).NE Iraq has higher attenuation values than N Iraq (Jassim and Goff, 2006).The QC value increases with depth, indicating that attenuation decreases because homogeneity rises with depth (Aki, 1980).

Conclusions
A Coda-Q (QC) has been determined for the northern stations of Iraq using a single back-scattering model of S-coda envelopes.Based on the analysis, the frequency-dependent attenuation relationship for the 30 second window length of the station SLY1 is obtained as QC = 88 f 1.02, QC = 90 f 0.97 for the DHK1 station, and QC = 91 f 1.04 for the IKRK station.Comparing the QC for these three stations, Sulaymaniyah has the highest attenuation, followed by Dohuk and Kirkuk.The average quality factor (QC) values, estimated for northern stations, and their frequency-dependent relationships show that the medium is heterogeneous.Since the attenuation decreases with depth, it may indicate that the deep crust is more homogeneous than the upper crust.Also, the measurement of QC is important for understanding tectonic areas.

Fig. 1 .
Fig. 1.Location map of the study area depicts the seismic stations of the Mesopotamian Network (MP) and Iraqi Meteorological Organization and Seismology (IMOS)

Fig. 2 .
Fig. 2. Raw seismogram recorded at station SLY1 showing the origin time, the headers, and the P arrival time of the event in the SEISAN program.
in Fig.3is automatically selected at twice the S-travel time from the origin time and the P arrival time.The figure shows the original and band-pass filtered seismogram of a local earthquake recorded by the SLY1 station, as well as displays the envelope of Ln [A (f, t) + β Ln (t)] vs t, with a window length of 30 seconds.

Fig. 4
an example mentioned, showing the idea of the work of the Seisan program to extract the value of QC.It is a statistical method for calculating amplitude and time using the Seisan program to calculate the central frequencies and Excel to clarify the relationship of amplitude with time, for a central frequency = 1Hz.

Fig. 4 .
Fig. 4. Example shows the idea of the work of the Seisan program to extract the value of Qc seismic event recorded by the SLY1 station to the event number 37 in Table 2. (a) The original seismogram shows the time arrival P and S, origin time, and length of the coda window 30 sec.(b) Filtered seismogram, for the central frequency 1 Hz.(c) The amplitude of the seismogram shows the envelope of ln amplitude versus t.

Fig. 5 .
Fig. 5.The distributions of quality factor values with frequency for coda window length 30 sec in the (a) SLY1, (b) DHK1, and (c) IKRK Stations, with the final equation for each station.

Fig. 6 .
Fig. 6.Equation of QC relationship with frequency Between the three stations for Comparison.

Table 1 .
The Network names and locations of the seismic stations used in this study.

Table 2 .
The seismic events used in this study that recorded at the stations.

Table 3 .
Displays the band-pass filter's low and high cutoff frequencies for the central frequencies.

Table 4 .
The QC Values at each frequency band were taken from local Earthquakes in the different stations and coda window length of 30 sec.

Table 5 .
Shows the attenuation parameter variations (Q0, α) with standard deviation to the three stations.