Keywords

1 Introduction

The discrimination between seismic events and explosions is one of the most studied sections in seismology. Moreover, with the Comprehensive Nuclear Test Ban Treaty Organization (CTBTO), the research is increasing in order to obtain efficient and powerful packages to characterize the sources of seismic signals captured by the seismometers.

In order to improve the performance of our seismic data recorded on the stations of our network, some discrimination methods have been applied on the seismic signals recorded on Moroccan network, with the aim of developing more relevant and adapted methods to our data in the future. The objectives are the distinguish between artificial events and natural events and to allow the correction of trajectory effects on seismic wave propagation in the different geological parts of Morocco (Michard et al., 2008).

2 Materials and Methods

The data selected for this work was collected from the Picasso, Topo Iberia, and Morocco-Array networks (Fig. 1). The stations used are broadband seismographs (BBS) type with three components, equipped with seismometers Nanometrics Trillium 120P/Streckeisen STS-2/ and Guralp. Eighteen small events were selected for this study in the magnitude range between 2.0 and 4.3, of which seven are small natural earthquakes, and eight are explosions induced by the RIFSIS (Gil, et al., 2014) and SIMA (Ayarza et al., 2014) projects. All events were recorded between 2010 and 2012, located between latitude 30–36 and longitude from − 3 to − 9.

Fig. 1
figure 1

Deployment of temporary and permanent seismic stations used in this study

Full size image

A vast number of techniques have been used to discriminate between earthquakes and quarry explosions; in this work, we test some of these methods for the identification of chemical explosion signals from earthquakes. The main identification method previously used in discrimination is the duration of the signal, because for the majority of events, the duration is longer for earthquakes than for explosions (Li et al., 1995). The spectral analysis method is based on the distribution of the amplitude-frequency spectrogram in the time domain, and to study the spectral characteristics of signals and the frequency content of whole seismic waveforms (Plafcan et al., 1997). Finally, the moment magnitude MW is the most accurate and modern quantification for measuring the size of an earthquake; however, magnitude scales have long been used for this purpose (Hanks & Kanamori, 1979). The following relation defines the moment magnitude MW:

$$ M_{w} = \left( {\log_{10} M_{0} /1.5} \right) - 6.07 $$
(1)

M0 (N·m): is the seismic moment; it is a static parameter and is not related to the dynamic properties of the source.

In order to improve our study of discrimination, we calculate the ratio between the moment magnitude MW and the local magnitude ML using the maximum amplitudes on the Z channels (vertical) (Table 1). This method is based on a physical model of deformation of seismic events that allows the best calculation of the magnitude in both cases of large and small earthquakes.

Table 1 Moment magnitude and signal duration (SD) with corresponding ML and Mc, for the studied earthquakes and explosions indicated by EV and SP/SR, respectively
Full size table

3 Results and Discussion

Many papers have attempted to deal with the discrimination issue, including some similar to our study of small earthquakes and explosions (Allmann et al., 2008; Dahy & Hassib, 2009). This work indicates that one of the most effective methods for distinguishing artificial events from natural events is the use of relative amplitude and frequency.

Applying this ratio between the moment magnitude Mw and the local magnitude ML discussed in Sect. 2, we can see that the distinction between them is very clear (Fig. 2a); the discrimination is easier when the moment magnitude Mw is > 2.8. However, the local magnitude of the explosions is smaller than that of the earthquakes.

Fig. 2
figure 2

a The MW/ML ratios for the studied events b Relationship between time duration (second) and coda magnitude c Amplitude-Frequency average spectra from all events studied in this work. Explosions are in Red and earthquakes are in Blue colour

Full size image

We deduce from Fig. 2b that the correlation between the duration and coda magnitude (Mc) works as an important factor in discriminating between earthquakes and explosions, the two populations appear separated and there is no confusion between them. The results of the spectral distribution in the different stations (Table 1), represented in Fig. 2c show the general view of the average of the spectrograms of mining explosions and earthquakes in station. This demonstrates the relationship between the relative amplitude and frequency detected in the Moroccan seismic network. As the graph reveals, discrimination in the frequency domain is observed between 9 and 12 Hz for earthquakes and between 1 and 3 Hz for explosions.

4 Conclusions

In this paper, we have applied some methods for discriminating between earthquakes and explosions in Morocco. We work then to quantify the performance of these methods as sufficient discriminants in the case of small earthquake magnitudes and artificial events. The dataset used in this study was collected from 2010 to 2012 with more than 80 seismic stations distributed over the Moroccan territory.

The relative amplitude-frequency ratio used for the totality of the signals recorded at the Moroccan network shows that explosions have a relatively high amplitude-frequency ratio compared to earthquakes. We conclude that the best frequency band filter to apply for earthquakes is 1–3 and 9–12 Hz for explosions. It should also be noted that to complete our study of discrimination, we need to explore the effect of other methods. It is clear from the results obtained in this work that an additional research is needed for regional explosions recorded via teleseism in our seismic network to provide valuable information regarding discrimination for future studies.