Discrimination between earthquakes and quarry blasts in the Aswan region, southern Egypt, using P-wave source spectra

In order to distinguish between shallow earthquakes and quarrying activity, we evaluated 3069 seismic signals from 413 occurrences (112 explosions and 301 shallow earthquakes, depths 5 km), collected by the Aswan seismic sub-network in southern Egypt from 2010 to 2021. The spectral properties and related source parameters (e.g., seismic moment and corner frequency) using the estimated P-wave source spectra for both earthquakes and quarry blasts were investigated. The analysis showed that the P-wave source spectra of shallow earthquakes have higher corner frequencies (8.8–23 Hz) than quarry blasts (1.0–2.6 Hz) within the same magnitude range. The source spectra of quarry blasts exhibited significant misfits with the omega-square model and had steeper falloffs at high frequencies. The selected quarry blasts have a narrow seismic moment range, from 2.03 × 1011 to 1.35 × 1012 Nm. Our results demonstrate that the evaluation of misfit of P-wave spectra from the omega-square source model, based on spectral amplitude characteristics of high- and low-frequency bands, is the most reliable discriminant method in the routine data analysis of the target area.

craton. Most of the active faults in western Aswan that are related to the Western Desert fault system are oriented east-west and few faults are north-south oriented (Fig. 1a). The east-west faults are located adjacent to the Sinn El Kaddab plateau and are relatively long, extending to the Nubia plain; some extend more than 200 km (Issawi 1973;Issawi 1978;Woodward-Clyde Consultants 1985;Hamimi et al. 2018). The most active faults are distributed in the Kalabsha fault zone (Hamimi et al. 2018), and the largest was the magnitude 5.6 earthquake of November 14, 1981. Former studies in Aswan have shown that seismic activity is concentrated along the active faults and increases near the intersection of north-south and east-west faults (Fig. 1b). Most earthquakes have a strike-slip focal mechanism, while normal fault motion is rarely observed (e.g., El-Amin 2003;Abou-Elenean 2003;Hassib et.al 2010;Hussein et al. 2013;Hosny et al. 2014;Badreldin et al. 2019;Saadalla et al. 2020;Mostafa and Mohamed 2021).
The cement industry in Egypt is growing rapidly in Aswan since limestone, the primary raw material for cement, is mined in the Sin el-Khadab plateau. The blasting of many quarries in the area has yielded 70% of the limestone used in the cement industry. The quarrying activity has been recorded by the local Aswan seismic sub-network, operated by the National Egyptian Seismic Network (ENSN). The anthropogenic activity decreases the reliability of the earthquake catalog in the target region, which leads to incorrect earthquake hazard and risk analysis. Additionally, these quarry blasts are carried out in an area with high seismic activity and complex tectonic setting. The explosions may trigger earthquakes in the nearby faults and disturb the stress pattern of existing tectonic structures (Fat-Helbary et al. 2012). Quarry blasting near the seismically active areas has fostered the interest of researchers to identify the characteristics of seismic and quarry activity in the Aswan region.
This study aims to differentiate the characteristics of quarry blasts from earthquakes in Aswan. We investigated the spectral properties of ripple-firing quarry blasts and earthquakes in the southwestern part of Aswan, Southern Egypt, to quantify the differences in spectral characteristics between natural earthquakes and blasts. This discrimination helps to provide a reliable seismic catalog containing only earthquakes in the study area, which can evaluate the seismicity of seismic zones in and around Aswan.

Data
The Aswan seismic sub-network, operated by the Aswan Regional Seismological Research Centre of the National Research Institute of Astronomy and Geophysics (NRIAG), consists of 15 broadband seismic stations. The seismometers installed in the network are Trillium 40 and Trillium-120QA (Nanometrics Inc.), and the data were recorded with a sampling frequency of 100 Hz. We processed 3069 signals from 301 shallow earthquakes (< 5 km depth) and 112 quarry blasts recorded by the sub-network in southern Egypt from January 2010 to December 2021. Since the objective of this study is to investigate the differences between blasting and earthquakes, we obtained information regarding the date and time of blasting in the target period from the cement company in advance. Figure 2a shows the distribution of epicenters of the selected dataset. The selected earthquakes and quarry blasts have the same magnitude range (1.5 < Ml ≤ 3) (Fig. 2b). Figure 2c shows the number of earthquakes and quarry blasts recorded at each seismic station. Table S1 of the supplementary materials lists the original parameters of the selected events. We selected waveform data based on the conditions that (a) the seismic event was recorded at more than four stations (Fig. 2c), and (b) waveforms were unclipped with high signal-to-noise ratios (S/N). Figure 3a and b show an example of the recorded data for an earthquake on October 22, 2017, and for quarry blasts on February 11, 2018, respectively. The waveforms of the observed quarry blast show trailing phases in addition to the initial P-waves, which makes the discrimination difficult. Waveform data selected are pre-processed using seismic analysis code software (Goldstein and Snoke 2005).

Spectral decomposition method
The waveforms of shallow earthquakes and quarry blasts to derive the P-wave source spectra were extracted and the displacement spectra over a 1.28-s window for noise and signal before and after the P-wave arrivals and ensured acceptable signal to noise ratio (S/N) for all spectra used in the analysis was computed (see an example in Fig. 4). The signal to noise amplitude ratio exceeds 10 in the frequency range between 0.8 and 50 Hz for all events are used in this study. A Parzen window of 5.0-Hz duration was applied to smooth the obtained spectra. The obtained P-wave displacement spectra are a superposition of the different effects and responses (source, path, and site effects), where O ij (f) is the observed spectra from source i at receiver j, S i (f) is the source effect, G j (f) is the site response, and Pij (f) represents seismic wave attenuation along the path. The next step is to process the spectra to remove the path and site effects from the observed spectra. Equation (1) is a linear system of equations with an overdetermined problem, resulting in a trade-off between the source and site results. This shortage is solved numerically using a robust iterative least square analysis using a reference site.
The reference site is a rigid rock site with no amplification compared to other receiver sites. However, many studies in other regions pointed out that finding such a site is difficult (Steidl et al. 1996;Archuleta 1998). Other researchers assumed that the amplification at the possible reference station site is around 2.0 due to the free surface effect (e.g., Iwata and Irikura 1988;Takemura et al. 1991;Kato et al. 1992). The significant disadvantage of referencing a rock site is that the number of processed events is limited to the number of events recorded by the selected reference site. Saadalla et al. (2020Saadalla et al. ( , 2019 pointed out that station New Gabal Marwa (NGMR), which belongs to the Aswan seismic subnetwork, has a uniform amplification curve around 2.0. The station can be used as a suitable reference site in the inversion analysis.
The data selection procedure, data preparation, and spectral inversion analysis assume reference sites are applied with the same weights, conditions, and constraints for both families of earthquakes and quarry blasts of the dataset. The spectral analysis follows Andrews (1986) and Iwata and Irikura (1988). The inversion finally yields the site terms except for the reference site and anelastic attenuation factor of the area. These factors are introduced into Eq. (1) to obtain the source spectra for all events at all observation sites. We regard the average source spectra of all available stations derived for each event as the source spectrum for the individual event.

Source parameter estimations
We obtained theoretical P-wave source spectra for earthquakes and quarry blast based on the omegasquare source formula proposed by Brune (1970).
Brune's model relates seismic moment to the displacement source spectra as The seismic moment M 0 is calculated from the flatlevel P-wave displacement spectra Ω 0 . Here, ρ is the density, v is the P-wave velocity at the source, F is the free surface effect, R θφ is the P-wave radiation pattern, and G(r,h) is the geometrical spreading effect. Radiation patterns were used for correction using 0.52 for the P-wave (Aki and Richards 2002). We obtained values of ρ and v from the 1D seismic velocity model of Khalil et al. (2004) (Table 1). The corner frequency f c of each event is estimated by finding the intersection between the low-and high-frequency asymptotes of the obtained displacement source spectra ( ( where A 0 is the flat level of the acceleration spectra. The obtained seismic moment for each event is used to calculate the moment magnitude using the formula of Kanamori (1977), Figure 5a shows an example of the average source spectra and standard deviation of the earthquake that occurred on October 22, 2017, at 11:20 (GMT), and Fig. 5b shows an example of the average source spectra of the explosion on February 11, 2018, at 06:32 (GMT) with the standard deviations. The obtained spectra slightly vary among observation points, implying that the inversion errors and scattering effects caused by different azimuths of the receivers are small. We calculated corner frequencies

Results and discussions
and seismic moment values for each event based on Eqs. 3-6 to discriminate between natural earthquakes and artificial events. Figure 6a and b show that the obtained displacement source spectra for quarry blasts show steeper falloffs at high frequencies in comparison with earthquakes. We found that the displacement P-wave source spectra of the earthquakes show suitable fittings with the omega-square models, where the quarry blasts indicate large misfits. Figure 7a and b show the fitting results with the omegasquare models for different-size earthquakes and quarry blasts, respectively. Additionally, the quarry blasts exhibited steeper falloffs at high frequencies than earthquakes, which led to lower stress drops and corner frequencies. The deficiency of quarry blasts from high-frequency seismic energy was attributed to the ripple firing mechanism of the quarry explosions (e.g., Allmann et al. 2008;Shearer and Allmann 2007;Su et al. 1991;Bennett and Murphy 1986;Taylor et al. 1988). Figure 8 shows the corner frequencies of the earthquakes and quarrying of the same magnitude ranged between 8.8-23 Hz and 1.0-2.6 Hz, respectively. We obtained the seismic moment values are in the range 2.03 × 10 11 -1.35 × 10 12 Nm for quarry blasts and 1.02 × 10 10 to 5.83 × 10 12 for natural earthquakes, respectively. Quarry blasting has a narrower seismic moment range than shallow earthquakes in the Aswan region (Table S2, supplementary material).
We finally discuss the discrimination method between earthquakes and quarry blasts. The classification between explosion events and earthquakes is usually done in a simple manner based on the epicenter location, epicenter depth, or the complexity of the waveform. However, the locations and depths between quarry blasts and earthquakes are similar, and the waveforms-based classification is not easy because of the presence of later phases in both waveforms (Fig. 3). We attempted to identify earthquakes and explosions using different magnitude scales; Ml was obtained from the earthquake catalog, and M w was derived from this study. Figure 9 shows the relationship between Ml and M w for both earthquakes and  Comparisons between the derived source spectra (colored solid lines) and the omega-square source model (black dashed lines) for different magnitude bins of A earthquakes and B quarry blasts quarry blasts. The relationship derived from earthquakes is M w = 0.7 Ml + 0.4, while M w = 0.6 Ml + 0.9 for quarry blasts, indicating that the discrimination using the M w -Ml relationship is not realistic. Our results suggest that only the spectral properties and source parameters could be used as reliable and robust discriminant factors. The most applicable method is deriving the source spectrum from the observed P-wave by removing the path and site characteristics obtained in this study and evaluate the agreement with the omega-square model. Alternatively, using the ratio of the amplitudes in the low-frequency Fig. 8 A Calculated moment magnitude and estimated corner frequencies for earthquakes (blue circles), and quarry blasts (red stars). B Histogram distribution for earthquakes (blue column), and C for quarry blasts (red column) Fig. 9 Estimated relationship between local magnitudes from the catalogues (Ml) and obtained moment magnitudes (M w ) for A earthquakes and B quarry blasts component (< 10 Hz) to those in the high-frequency component (> 20 Hz) from the P-wave spectra of the observed waveforms as an indicator may enable the identification.

Conclusions
A total of 3069 seismic records were analyzed from 112 quarry blasts and 301 earthquakes with local magnitudes (Ml) between 1.7 and 3.1 from 2010 to 2021. The obtained P-wave displacement spectra from local shallow earthquakes and quarry blasts revealed various spectral properties and parameters (corner frequency and seismic moment) between both types of events. The Fourier spectra of quarry blasts poorly fit the omega-square source model and have lower fc than shallow earthquakes. The P-spectra for the quarry blasts have steeper falloffs and attenuated rapidly at a lower frequency in comparison with shallow earthquakes in the same tectonic region and within the same magnitude range. The investigated quarry blasts had longer source time durations of blasts than natural earthquakes, leading to lower corner frequencies and low-stress drop values than the studied natural seismic events. Our study focused on Aswan shallow earthquakes and quarry blasts, but we expect the obtained discriminant parameters and insights found in this analysis will be applicable to other regions of more direct interest to mining and quarrying. The spectral properties obtained in this study are helpful in event discrimination processing. Making a catalog of quarry blasting will be essential to evaluate the hazards and environmental impacts of the mining development in Egypt.
Author contribution HS processed the earthquake and quarry blast data and performed spectral inversions. The computer program for spectral decomposition analysis was prepared by TH, and HM contributed to interpretations of the results. HS wrote the initial draft of the manuscript, and TH edited it. All the authors have read and approved the final manuscript.
Funding Open access funding provided by The Science, Technology & Innovation Funding Authority (STDF) in cooperation with The Egyptian Knowledge Bank (EKB). This study was supported by the Quarries Services project working under the Seismology Department of the National Research Institute of Astronomy and Geophysics (NRIAG).

Data availability
The data that support the findings of this study are available from NRIAG but restrictions apply to the availability of these data, which were used under license for the current study, and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of NRIAG.

Declarations
Ethics approval and consent to participate Not applicable.

Competing interests The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.