Spectral classification methods in monitoring small local events by the Israel seismic network

We use the dense Israel Seismic Network (ISN) to discriminate between low magnitude earthquakes and explosions in the Middle East region. This issue is important for CTBT monitoring, especially when considering small nuclear tests which may be conducted under evasive conditions. We explore the performance of efficient discriminants based on spectral features of seismograms using waveforms of 50 earthquakes and 114 quarry and underwater blasts with magnitudes 1.0–2.8, recorded by ISN short-period stations at distances up to 200 km. The single-station spectral ratio of the low and high-frequency seismic energy shows an overlap between explosions and earthquakes. After averaging over a subnet of stations, the resolving power is enhanced and the two classes of events are separated. Different frequency bands were tested; the (1–3 Hz)/(6–8 Hz) ratio provided the best discriminant performance. We also estimated normalized r.m.s. spectral amplitudes in several sequential equal frequency windows within the 1–12 Hz band and applied multiparametric automatic classification procedures (Linear Discrimination Function and Artificial Neural Network) to the amplitudes averaged over a subnetwork. A leave-one-out test showed a low rate of error for the multiparametric procedures. An innovative multi-station discriminant is proposed, based on spectral modulation associated with ripple-firing in quarry blasts and with the bubbling effect in underwater explosions. It utilizes a distinct azimuth-invariant coherency of spectral shapes for different stations in the frequency range (1–12 Hz). The coherency is measured by semblance statistics commonly used in seismic prospecting for phase correlation in the time domain. After modification, the statistics applied to the network spectra provided event separation. A new feature of all the above mentioned procedures is that they are based on smoothed (0.5 Hz window), instrument-corrected FFT spectra of the whole signal; they are robust to the accuracy of onset time estimation and, thus well suited to automatic event identification.     

Empirical site response evaluations: case studies in Israel

A closed-loop, time-efficient technique for site effect evaluation for seismic microzonation in urban areas is developed and implemented. It includes the following successive steps: microtremor measurements using a triangular array of three-component seismometers; estimation of the Nakamura horizontal-vertical spectral ratio, calculation of Rayleigh wave dispersion using the Aki spatial correlation method; inversion of the dispersion curve to determine a shear wave velocity model of the shallow subsurface; analytical computation of Rayleigh wave spectral ellipticity and 1-D SH transfer function for this model; and a comparison of empirical and analytical ratios and correction of the model if required. The technique has been tested at several locations in a number of towns near the active Jordan Rift valley. A reasonable agreement is obtained when comparing empirical spectral ratios and analytical transfer functions.     

Signal Processing for Indian and Pakistan Nuclear Tests Recorded at IMS Stations Located in Israel

v—vIn compliance with the Comprehensive Nuclear-Test-Ban-Treaty (CTBT) the International Monitoring System (IMS) was designed for detection and location of the clandestine Nuclear Tests (NT). Two auxiliary IMS seismic stations MRNI and EIL, deployed recently, were subjected to detectability, travel-time calibration and discrimination analysis. The study is based on the three recent 1998 underground nuclear explosions: one of India and two of Pakistan, which provided a ground-truth test of the existing IMS. These events, attaining magnitudes of 5.2, 4.8 and 4.6 correspondingly, were registered by many IMS and other seismic stations.¶The MRNI and EIL broadband (BB) stations are located in Israel at teleseismic distances (from the explosions) of 3600, 2800 and 2700ukm, respectively, where the signals from the tests are already weak. The Indian and the second Pakistan NT were not detected by the short-period Israel Seismic Network (ISN), using standard STA/LTA triggering. Therefore, for the chosen IMS stations we compare the STA/LTA response to the results of the more sensitive Murdock-Hutt (MH) and the Adaptive Statistically Optimal Detector (OD) that showed triggering for these three events. The second Pakistan NT signal arrived at the ISN and the IMS stations in the coda of a strong Afghanistan earthquake and was further disturbed by a preceding signal from a local earthquake. However, the NT signal was successfully extracted at EIL and MRNI stations using MH and OD procedures. For comparison we provide the signal analysis of the cooperating BB station JER, with considerably worse noise conditions than EIL and MRNI, and show that OD can detect events when the other algorithms fail. Using the most quiet EIL station, the most sensitive OD and different bandpass filters we tried in addition to detect the small Kazakh chemical 100-ton calibration explosion of 1998, with magnitude 3.7 at a distance approaching 4000ukm. The detector response curve showed uprising in the expected signal time interval, but yet was low for a reliable decision.¶After an NT is detected it should be recognized. Spectra were calculated in a 15-sec window including P and P-coda waves. The spectra for the first Pakistan NT showed a pronounced spectral null at 1.7uHz for all three components of the EIL station. The effect was confirmed by observation of the same spectral null at the vertical component of the ISN stations. For this ground-truth explosion with a reported shallow source depth, the phenomenon can be explained in terms of the interference of P and pP phases. However, the spectral null feature, considered separately, cannot serve as a reliable identification characteristic of nuclear explosions, because not all the tests provide the nulls, whereas some earthquakes show this feature. Therefore, the multi-channel spectral discrimination analysis, based on a spectral ratio of low-to-high frequency energy (in the 0.6–1uHz and 1–3uHz bands), and a semblance of spectral curves (in the 0.6–2uHz band), was conducted. Both statistics were calculated for the vertical component of the ISN stations as well for the three components of the EIL station. The statistics provided a reliable discrimination between the recent NT and several nearby earthquakes, and showed compliance with the former analysis of Soviet and Chinese NT, where nuclear tests demonstrated lower values of energy ratio and spectral semblance than earthquakes. ¶Accurate location of NT requires calibration of travel time for IMS stations. Using known source locations, IASPEI91 travel-time tables and NEIC origin times we calculated expected arrival time for the P waves to the EIL and MRNI stations and showed that the measured arrival time has a delay of about 4 sec. Similar results were obtained for the nearby Pakistan earthquakes. The analysis was complimented by the P travel-time measurements for the set of Semipalatinsk NT, which showed delays of about 3.7usec to the short-period MBH station which is a surrogate station for EIL. Similar delays at different stations evidence a path- rather than site-effect. The results can be used for calibration of the IMS stations EIL and MRNI regarding Asian seismic events.     

Underwater Explosion (UWE) Analysis of the ROKS Cheonan Incident

The underwater explosion (UWE) resulting in the sinking of the South Korean warship, ROKS Cheonan occurred on March 26 2010. Raw data was analyzed from several 3-component stations—Baengyeong-do Korea Meteorological Administration (KMA) station (BAR), Ganghwa KMA station (GAHB), Incheon Incorporated Research Institutions for Seismology (IRIS) station (INCN), the short-period station—Deokjeok-do KMA station (DEI), as well as from the seismo-acoustic array Baengyeong-do Korea Institute of Geoscience and Mineral Resources (KIGAM) station (BRDAR). The ROKS Cheonan incident has been investigated by both the Multinational Civilian-Military Joint Investigation Group (Ministry of National Defense, 2010) and Hong (Bull Seism Soc Am 101:1554–1562, 2011). Their respective methods and conclusions are also presented in this study. One of the main differences between their findings and ours is that we deducted that the fundamental bubble frequency was 1.01 Hz with a subsequent oscillation of 1.72 Hz. Also, in contrast to findings by the MCMJIG and Hong, our analysis shows the first reverberation frequency to be 8.5 Hz and the subsequent one to be ≈25 Hz. The TNT-equivalent charge weight (seismic yield) and seismic magnitude were estimated from an observed bubble frequency of 1.01 Hz and the analytical model of a bubble pulse. From the data analyzed, we deducted that the seismic yield would be about 136 kg of TNT, which is equivalent to the individual yield of a large number of land control mines (LCM) which were abandoned in the vicinity of the ROKS Cheonan incident by the Republic of Korea (ROK) Navy in the 1970s (Ministry of National Defense 2010). Also, whereas both the MCMJIG and HONG estimated the local magnitude at 1.5, our findings came to the conclusion of a local magnitude of approximately 2.04 based on the bubble frequency of 1.01 Hz measured on the vertical component of BAR station data considering the empirical relationship between charge weight (TNT yield) and underwater explosion magnitude. Strong high-frequency signals collected at the 3-component BAR station approximately 30 s after P-wave arrivals and infrasound records at BRDAR clearly indicate powerful acoustic phases and N-waves caused by a relatively shallow UWE. T-phases are also observed on seismograms and spectra at 15–17 Hz on the DEI, GAHB, and INCN stations.     

GT0 Explosion Sources for IMS Infrasound Calibration: Charge Design and Yield Estimation from Near-source Observations

Three large-scale on-surface explosions were conducted by the Geophysical Institute of Israel (GII) at the Sayarim Military Range, Negev desert, Israel: about 82 tons of strong high explosives in August 2009, and two explosions of about 10 and 100 tons of ANFO explosives in January 2011. It was a collaborative effort between Israel, CTBTO, USA and several European countries, with the main goal to provide fully controlled ground truth (GT0) infrasound sources, monitored by extensive observations, for calibration of International Monitoring System (IMS) infrasound stations in Europe, Middle East and Asia. In all shots, the explosives were assembled like a pyramid/hemisphere on dry desert alluvium, with a complicated explosion design, different from the ideal homogenous hemisphere used in similar experiments in the past. Strong boosters and an upward charge detonation scheme were applied to provide more energy radiated to the atmosphere. Under these conditions the evaluation of the actual explosion yield, an important source parameter, is crucial for the GT0 calibration experiment. Audio-visual, air-shock and acoustic records were utilized for interpretation of observed unique blast effects, and for determination of blast wave parameters suited for yield estimation and the associated relationships. High-pressure gauges were deployed at 100–600 m to record air-blast properties, evaluate the efficiency of the charge design and energy generation, and provide a reliable estimation of the charge yield. The yield estimators, based on empirical scaled relations for well-known basic air-blast parameters—the peak pressure, impulse and positive phase duration, as well as on the crater dimensions and seismic magnitudes, were analyzed. A novel empirical scaled relationship for the little-known secondary shock delay was developed, consistent for broad ranges of ANFO charges and distances, which facilitates using this stable and reliable air-blast parameter as a new potential yield estimator. The delay data of the 2009 shot with IMI explosives, characterized by much higher detonation velocity, are clearly separated from ANFO data, thus indicating a dependence on explosive type. This unique dual Sayarim explosion experiment (August 2009/January 2011), with the strongest GT0 sources since the establishment of the IMS network, clearly demonstrated the most favorable westward/eastward infrasound propagation up to 3,400/6,250 km according to appropriate summer/winter weather pattern and stratospheric wind directions, respectively, and thus verified empirically common models of infrasound propagation in the atmosphere.     

The ratio between corner frequencies of source spectra of <Emphasis Type="Italic">P</Emphasis>- and <Emphasis Type="Italic">S</Emphasis>-waves—a new discriminant between earthquakes and quarry blasts

This study analyzes and compares the P- and S-wave displacement spectra from local earthquakes and explosions of similar magnitudes. We propose a new approach to discrimination between low-magnitude shallow earthquakes and explosions by using ratios of P- to S-wave corner frequencies as a criterion. We have explored 2430 digital records of the Israeli Seismic Network (ISN) from 456 local events (226 earthquakes, 230 quarry blasts, and a few underwater explosions) of magnitudes Md?=?1.4–3.4, which occurred at distances up to 250 km during 2001–2013 years. P-wave and S-wave displacement spectra were computed for all events following Brune’s source model of earthquakes (1970, 1971) and applying the distance correction coefficients (Shapira and Hofstetter, Teconophysics 217:217–226, 1993; Ataeva G, Shapira A, Hofstetter A, J Seismol 19:389-401, 2015), The corner frequencies and moment magnitudes were determined using multiple stations for each event, and then the comparative analysis was performed.The analysis showed that both P-wave and especially S-wave displacement spectra of quarry blasts demonstrate the corner frequencies lower than those obtained from earthquakes of similar magnitudes. A clear separation between earthquake and explosion populations was obtained for ratios of P- to S-wave corner frequency f ₀(P)/f ₀(S). The ratios were computed for each event with corner frequencies f ₀ of P- and S-wave, which were obtained from the measured f ₀ ^I at individual stations, then corrected for distance and finally averaged. We obtained empirically the average estimation of f ₀(P)/f ₀(S)?=?1.23 for all used earthquakes, and 1.86 for all explosions. We found that the difference in the ratios can be an effective discrimination parameter which does not depend on estimated moment magnitude M _w .The new multi-station Corner Frequency Discriminant (CFD) for earthquakes and explosions in Israel was developed based on ratios P- to S-wave corner frequencies f ₀(P)/f ₀(S), with the empirical threshold value of the ratio for Israel as 1.48.