Electromagnetic Signals Close in Time to Earthquakes B. V. Dovbnya, O. D. Zotov, A. O. Mostryukov, and R. V. Shchepetnov Borok Geophysical Observatory (BGO), Schmidt Institute of Physics of the Earth, Russian Academy of Sciences, Borok, Yaroslavl oblast, Russia (e-mail: dov@borok.adm.var.ru) Abstract. Search for electromagnetic effects of earthquakes has discovered pulsed signals preceding or following the time of event. The advance or lag amount to 0--5 min. The signals are observed as single or paired pulses in the frequency range 0--5 Hz. Dynamic spectra of the signals are presented and their characteristics are discussed. The hypothesis on the seismoelectromagnetic origin of the signals is suggested and verified. INTRODUCTION. Electromagnetic signals from earthquakes have already been discussed over more than a century. This interest is primarily related to the potential possibility of discovering short-term precursors of earthquakes. Their observation is also important for the study of physics of seismic processes. There are known single observations of electromagnetic signals correlating with earthquakes [Breiner, 1964; Moore, 1964; Eleman, 1966; Belov et al., 1974; Gokhberg et al., 1989; Krylov and Levshenko, 1990; Guglielmi and Levshenko, 1994; Yoshimori, 2002], and various
mechanisms are proposed for the conversion of the mechanical energy into the electromagnetic field energy [Guglielmi and Ruban, 1990; Guglielmi, 1992; Gershenson et al., 1993; Guglielmi and Levshenko, 1996, 1997]. However, the reliability of the relationship between the discovered signals and earthquakes is still debatable. The possibility itself of detecting the seismoelectromagnetic effects, at least by the presently available instrumentation, also raises doubts. Because of the difficulty of their recognition against a noise background, each observation of signals correlating with earthquakes raises great interest. The BGO regularly records and analyze variations in the natural electromagnetic field in the range 0--5 Hz, comparing them with other geophysical phenomena and, in particular, earthquakes. These observations detected, rather unexpectedly, anomalous pulsed signals in a reliably established close vicinity of the earthquake occurrence time (a few tens of seconds to a few minutes). Such observations were of undoubted interest for the search for electromagnetic effects of earthquakes. Investigations in this direction have been continued. The goal of this work is to describe the presently available results of the detection of electromagnetic signals correlating in time with earthquakes. First we briefly describe the method of analysis. The dynamic spectra of the discovered signals are then presented and their general characteristic is given. Further, we consider features peculiar to the observation of earthquakes in question. Finally, we discuss the results and formulate conclusions of our study.
INITIAL DATA AND METHODS OF ANALYSIS. The Borok observatory is located in a seismically passive zones, and the geographic coordinates of the observation point are (58.06N, 38.23E). Initial data were represented by analog magnetic tape records of ultralow frequency (ULF) variations in the electromagnetic field obtained by an induction magnetometer continuously operating at the observatory since 1973. The task was to seek pulsed electromagnetic signals correlating with earthquakes. The magnetic records were processed by methods of computer analysis (digitization and construction and analysis of spectra) and with the use of a spectrum analyzer of the Sonograph type. Amplitudes were estimated from visible photorecords obtained at a tape advance rate of 30 mm/min, and the frequency was determined from sonograms. The current fragment the magnetic tape recording (as a rule, 12 h long with the transformation coefficient Ktr = 1000) was represented as a diagram (sonogram) in the coordinates frequency time. The resulting instantaneous sonographic image of the alternating electromagnetic field provides, in the coordinates frequency time, information on signals in the interval analyzed. Such an approach is convenient in the search for new forms of electromagnetic radiation because, given a certain experience, it allows one to use the configuration of the dynamic spectrum for fairly rapid and reliable identification of anomalous signals (or a group of signals) possessing specific properties among the known signals. In our case, such property was a pulsed shape expected for signals of seismic origin [Guglielmi and Levshenko, 1996; Morgounov, 1985]. After preliminary inspection of sonograms, pulsed signals of the known (nonseismic) origin observed at the Borok observatory were eliminated from the further analysis.
They include lightning discharges observed in spring and summer time and possessing easily recognizable morphological features, irregular pulsations (Pi1B), and pulses associated with magnetic storm sudden onsets and with X- and gammaradiation of chromospheric flares [Guglielmi and Troitskaya, 1973; Troitskaya et al., 1979; Dovbnya et al., 1994, 1995]. In the analysis, each of these factors was taken into account and pulses (signals) correlating with the phenomena were eliminated from the further consideration. Noise signals were identified from their characteristic features in magnetograms and from their passage through high- and low-frequency magnetic recording channels. After this testing procedure, signals of a pulsed shape were compared with data of the world earthquake catalog of the International Seismological Center (ASC catalogues, www.isc.ac.uk). RESULTS OF OBSERVATIONS. Low frequency (0--5 Hz) electromagnetic signals uncorrelatable with other geophysical phenomena could be discovered for about 200 events (earthquakes) over the period from 1974 through 1976 in a specified time neighborhood of an earthquake. These signals either preceded (70%) or followed (30%) the earthquakes. Both leading and delayed signals were observed as either single (75%) or anomalous paired (25%) pulses (we shall define them as twinpulses). Signals of both types, as a rule, have discrete structure of the dynamic spectrum. According to several morphological features (short length and absence of dispersion, discreteness of the dynamic spectrum, an anomalous regime, and small
amplitudes), the detected signals differ from electromagnetic radiation disturbances known in geophysics. In the case of twin pulses, the time between the first and the second pulses varies from 40 to 150 s. The lead time of the first pulse before an earthquake varies within 50--250 s, and its delay time varies from 25 to 200 s. The amplitude usually does not exceed 10 pt, and the pulse length is no more than 20 s. Single pulses have an amplitude of up to 100 pt and a length of 40--60 s, and their lead and delay times vary from 0 to 250 s. Given the same average seismic activity, the observation probability of pulsed signals in a time neighborhood of an earthquake was found to be significantly different in different days and even weeks. Periods of complete absence of signals were suddenly followed by periods of unexpectedly high seismic activity. The probability of observation of signals recorded after a quiescence period did not depend significantly on the event magnitude but was higher in the case of nearer (2000--3000 km) earthquakes. It is also important to note that single pulses, although arising preferably close to the earthquake occurrence time, could sometimes be observed at the stage of earthquake nucleation and in other time intervals. Twin signals have not been observed outside the preferable time interval. Now, we consider examples of dynamic spectra of signals from earthquakes that differed in the magnitude M and occurred in different regions of the Earth. An arrow marks the moments of earthquakes in figures. The moments of earthquakes in figures are marked by an arrow.
Figure 1a illustrates the effect of the earthquake M = 3.0 that occurred on August 24, 1975 (11:58:21 UT) at a distance of 2500 km from the Borok observation point. Fig. 1. Examples of the dynamic spectra of single signals detected in the close vicinity of the occurrence times of the following earthquakes: (a) August 24, 1975 (11:58:21 UT), at the epicentral coordinates (37.4, 21.6); (b) November 21, 1973 (5:07:04 UT and 5:32:25 UT), at the epicentral coordinates (34.32, --116.36) and (-- 36.54, 147.75). The electromagnetic signal arose about 2 min before the earthquake onset. In the frequency time sonogram, it has the shape of a pulse of a discrete structure. Its length is about 30 s, and the amplitudes of its H and D components are, respectively, 100 and 150 pt. The frequency range is 0--2 Hz. The lower sonogram (Fig. 1b) presents examples of signals from distant earthquakes. Two events spaced at 25 min occurred on November 21, 1975, in different regions. In both cases, electromagnetic
signals arise 1.5--2 min before the earthquake time. They are also short, have the shape of a pulse of a discrete structure, and are observed in the same frequency range. However, their amplitudes are much lower and do not exceed 20 pt. The epicentral distance of each event is no less than 10000 km. The earthquake electromagnetic effect in the form of characteristic twin pulses is a relatively rare phenomenon. Such effects account for 25% of the total number of observed events. Fig. 2. Examples of the dynamic spectra of single signals detected in the close vicinity of the occurrence times of the following earthquakes: (a) January 17, 1977 (5:19:24 UT), at the epicentral coordinates (39.27, 43.7); (b) February 28, 1973 (6:37:54 UT), at the epicentral coordinates (50.514, 156.58).
Figure 2 presents examples of dynamic spectra of twin signals from distant and near earthquakes. The effect of the Turkish January 17, 1977, earthquake (M = 5.7) is illustrated in the upper sonogram (Fig. 2a). The electromagnetic pulse arose 3 min before the main shock. Its amplitude in Borok did not exceed 10 pt, and its length was 20 s. The second pulse of the same amplitude and length was observed after 35 s. Both pulses have a discrete structure in the range 0--3.5 Hz. The patterns of discreteness inherent in the majority of observed signals (single and paired) from earthquakes can differ for different events. In the case of twin signals, distinctions are also observed between the first and second pulses. Figure 2b demonstrates the electromagnetic effect preceding the strongest (M = 7.4) earthquake that occurred on February 28, 1973, at a distance of more than 10000 km from Borok. The signal arises 2 min before the earthquake onset and is represented in the sonogram by two discrete pulses 90 s apart in the range 0--2 Hz. They are no longer than 30 s, and their amplitudes in Borok do not exceed the sensitivity of the recording instrumentation (10 pt/mm). Note that the discreteness pattern and the time between the pulses are not the same as in the case considered above.
Fig. 3. Electromagnetic effect of the September 6, 1975 (9:20:11 UT) earthquake with the epicentral coordinates (38.513, 40.774). Figure 3 demonstrates one of the cases of anomalous occurrence of Pi1B geomagnetic pulsations during earthquakes. A destructive earthquake with M = 6.9 occurred in Turkey on September 6, 1975 (9:20 UT). We consider the succession of events as it appears in the sonogram. The leading twin signal arises 3 min before the main shock; it is represented by two pulses following each other with an interval of 1 min. The pulses of an amplitude of 20 pt were observed in the H component of the geomagnetic field. However, in this example, we are interested in the subsequent effect. One more signal arises at 9:20 UT and is followed by an irregular radiation pulsation about 5 min long that is similar in dynamic spectrum to the Pi1B pulsations [Guglielmi and Troitskaya, 1973]. Pulsations of these irregular oscillations are observed during the particle precipitation into the ionosphere and are typical of the evening time. The possible origin of the anomalous pulsations during earthquakes (LT = UT + 3 for Borok) is discussed below.
RELIABILITY OF THE EFFECTS. Single and twin signals differ in both the probability of their occurrence and their time relation to earthquakes. Therefore, in estimating their reliability, each type of the signals is discussed separately. (1) Single Signals. In the period from 1974 through 1976, 182 pulses were detected, and 155 of them were observed in the interval 0--5 min before or after an earthquake. We define this narrow time neighborhood as the interval of preferable occurrence (IPO) of signal and consider the relationship between pulses and earthquakes inside this interval (below, we return to the signals observed outside the IPO). Fig. 4. Distribution of single pulses in 5-min intervals before and after an earthquake. Fig. 4 shows the distribution of the single signals in the time intervale 0-5 min before and after the earthquake moment. Analyzing the resulting distribution
shown in Fig. 4, we should note that, if no causal relation exists between the measured value and the phenomenon compared and the sample is sufficiently large, the distribution will be uniform. And vice versa, a noticeable deviation from the uniformity can be regarded as evidence for the causal relation under consideration. The distribution in Fig. 4 is obviously inhomogeneous. Moreover, the histogram exhibits a well-expressed asymmetry between the left and right branches of the distribution; the asymmetry is observed not only in the numbers of pulses before and after an earthquake but also in the temporal pattern of the relationship between the pulses and earthquakes. Note that the nearer the earthquake, the more probable the signal occurrence. With the total number of earthquakes being 177, the number of pulses in the 5- min intervals before or after an earthquake is 155. The statistical significance of the occurrence of the paired pulse earthquake events is p = 0.88. (2) Twin Signals. The distribution of paired twin pulses in the time neighborhoods of earthquakes is shown in Fig. 5.
Fig. 5. Distribution of twin pulses in 5-min intervals before and after an earthquake. This figure shows that, likewise, anomalous signals are distributed irregularly in time and are asymmetric about an earthquake. It is noteworthy that both twin and single pulses arise preferably before an earthquake. As seen, 60 twin pulses were discovered, and 69 earthquakes occurred in the 5- min intervals before and after a signal. The statistical significance of the occurrence of the paired signal earthquake events is p = 0.87. Thus, since the above results are qualitatively different from the distribution of random quantities, we suggest that a causal relation exists between earthquakes and the signals observed in the close time vicinity of an earthquake. Therefore, we accept the working hypothesis on the seismic origin of the signals discovered.
DISCUSSION. This, we will consider the discovered signals as evidence for mechanical electromagnetic conversions in the crust. Usually, inductive and piezomagnetic mechanisms are usually regarded to be responsible for the generation of seismoelectromagnetic signals [Guglielmi and Levshenko, 1996]. However, the inductive mechanism implies abrupt appreciable displacements capable of generating of pulsed radiation and, with a high probability, this can be expected at the time moments of an earthquake, foreshocks, aftershocks, and coseismic signals [Iyemori et al., 1996]. Piezomagnetically induced generation of signals is controlled by piezomagnetic properties of earthquake source rocks. In our experiment, earthquakerelated signals were observed in geologically different regions and generally shortly before or after an earthquake. Some of these signals can be due to the action of these two mechanisms, but additional mechanisms are apparently required to account for the total set of the observed signals. We should note that the same forms and regimes of radiation (twin pulses) are repeatedly observed during earthquakes differing in their parameters and occurring in different seismically active regions. In this connection, it seems reasonable to suggest both a similarity of processes of mechanical electromagnetic conversions that result in the generation of electromagnetic pulses in the crust and the absence of an explicit dependence of such processes from the energy parameters of a forthcoming earthquake. The discovery of the preferable signal occurrence interval in the time vicinity of an earthquake is of a certain physical interest. Possibly, the processes of mechanical-
to-electromagnetic energy conversion are more probable in this specific time interval including the earthquake occurrence time moment. The observation of twin pulses is of particular interest. The regime of similar pairwise radiation is unknown in geophysics. The occurrence of two successive pulses possibly reflects the development of two interrelated processes immediately preceding the fracture of rocks in an earthquake source. The origin of such processes is unknown and deserves attention. Twin pulses have not been discovered outside the preferable time interval. An earthquake should be expected after the occurrence of the characteristic pair of pulses. The probable expectation time is 2--4 min after the occurrence moment of the first pulse. The detected signals actively affect the magnetosphere and ionosphere. Such an effect can result in precipitation of radiation belt particles into the ionosphere, which can produce anomalous Pi1B pulsations (Fig. 3). An earthquake-related abrupt rise in the flux of precipitating particles was observed in the experiment at the Salyut-7 orbital station in 1985 [Voronov et al., 1987]. The authors attribute this phenomenon to the resonance interaction of the Earth s low frequency (0--10 Hz) radiation with radiation belt particles. Now we consider signals recorded outside the IPO, i.e., signals occurring more than 5 min before or after an earthquake. Evidently, these signals arise between two successive earthquakes separated by a time interval of ten or more minutes. Accepting the hypothesis on the seismoelectromagnetic origin of the pulses in the close vicinity of the earthquake occurrence time, one cannot rule out possible
processes capable of generating such pulses at a different stage of the earthquake nucleation. However, this problem requires more careful examination because it is directly related to the main problem of geophysics, namely, electromagnetic prediction. On the basis of the above analysis, we can draw the following conclusions. (1) The above results indicate that the seismoelectromagnetic signals exist in reality. (2) The signals can be observed from both strong and weak earthquakes and at great epicentral distances. (3) The seismoelectromagnetic signals of this type are most likely to be found in the close vicinity of the earthquake occurrence time (within a few seconds or a few minutes). CONCLUSIONS. In this work, we tried to demonstrate experimentally the possibility of observation of seismoelectromagnetic signals and thereby to weaken the skepticism in relation to their actual existence. In any case, our results are promising for further search for electromagnetic effects of earthquakes. The experience of their search and analysis is of great significance for electromagnetic prediction studies because these directions of research are interrelated. ACKNOWLEDGMENTS. We are grateful to B.I. Klain and A.S. Potapov for helpful discussion.
This work was supported by the Russian Foundation for Basic Research, project nos. 04-05-64265 and 06-05-64178. REFERENCES 1. S. V. Belov, N. I. Migunov, and G. A. Sobolev, Magnetic Effect of Kamchatka Strong Earthquakes, Geomagn. Aeron. 14 (3), 380--382 (1974). 2. S. Breiner, Piezomagnetic Effect at the Time of Local Earthquakes, Nature 202 (4934), 790--791 (1964). 3. B. V. Dovbnya, A. V. Moldavanov, and V. A. Parkhomov, On the Origin of the Psfe High Frequency Component Produced by the Gamma Radiation of Solar Flares, in Studies on Geomagnetism, Aeronomy, and Solar Physics (Nauka, Moscow, 1994), Vol. 102, pp. 47--55 [in Russian]. 4. B. V. Dovbnya, V. A. Parkhomov, and R. A. Rakhmatulin, Long Period Geomagnetic Pulsations Accompanied by Intense X-Ray Flares, Geomagn. Aeron. 35 (3), 146--150 (1995). 5. F. Eleman, The Response of Magnetic Instruments to Earthquake Waves, J. Geomagn. Geoelectr. 16 (1), 43--72 (1966). 6. N. I. Gershenson, M. B. Gokhberg, and S. L. Yunga, On the Electromagnetic Field of an Earthquake Focus, Phys. Earth Planet. Inter. 77, 13--19 (1993). 7. M. B. Gokhberg, S. M. Krylov, and V. T. Levshenko, The Electromagnetic Field of an Earthquake Source, Dokl. Akad. Nauk SSSR 308 (1), 62--65 (1989).
8. A. Guglielmi, Elastomagnetic Waves in a Porous Medium, Physica Scr. 46, 433--434 (1992). 9. A. V. Guglielmi and V. A. Troitskaya, Geomagnetic Pulsations and Diagnostics of the Magnetosphere (Nauka, Moscow, 1973) [in Russian]. 10. A. V. Guglielmi and V. F. Ruban, On the Theory of the Seismomagnetic Induction Effect, Izv. Akad. Nauk SSSR, Ser. Fiz. Zemli, No. 5, 47--54 (1990). 11. A. V. Guglielmi and V. T. Levshenko, Electromagnetic Signals from Earthquakes, Fiz. Zemli, No. 5, 65--70 (1994). 12. A. V. Guglielmi and V. T. Levshenko, An Electromagnetic Pulse from an Earthquake Source, Dokl. Akad. Nauk 349 (5), 676--678 (1996). 13. A. V. Guglielmi and V. T. Levshenko, Electromagnetic Signal from an Earthquake Source, Fiz. Zemli, No. 9, 22--30 (1997) [Izvestiya, Phys. Solid Earth 33, 712--719 (1997)]. 14. T. Iyemori, T. Kamei, Y. Tanaka, et al., Co-Seismic Geomagnetic Variations Observed at the 1995 Hyogoken-Nanbu Earthquake, J. Geomagn. Geoelectr. 48, 1059--1070 (1996). 15. S. M. Krylov and V. T. Levshenko, On Ultralow Frequency Electromagnetic Radiation of Lithospheric Origin, Dokl. Akad. Nauk SSSR 311 (3), 579--581 (1990). 16. G. Moore, Magnetic Disturbances Preceding the 1964 Alaska Earthquake, Nature 203 (4944), 508 (1964).
17. V. A. Morgounov, On Electromagnetic Radiation Associated with Seismic Activity, Izv. Akad. Nauk SSSR, Ser. Fiz. Zemli, No. 3, 77--85 (1985). 18. V. A. Troitskaya, K. G. Ivanov, A. L. Kalisher, and B. V. Dovbnya, Daytime Magnetopause and Its Vicinities as a Source of Geomagnetic Pulsations in the Pc1 Range, Geomagn. Aeron. 19 (4), 657--662 (1979). 19. S. A. Voronov, A. M. Galper, V. G. Kirillov-Ugriumov, et al., Registration of Sporadic Increase of High Energy Particle Flux near Brasilia Magnetic Anomaly Region, in Proc. 20 th Int. Cosm. Ray Conf., Vol.4, 451--452 (1987). 20. H. Yoshimori, Seismoelectromagnetic Effect Associated with the Izmit Earthquake and Its Aftershocks (Jointly Authored), Bull. Seismol. Soc. Am. 92, 350--360 (2002). FIGURE CAPTIONS Fig. 1. Examples of the dynamic spectra of single signals detected in the close vicinity of the occurrence times of the following earthquakes: (a) August 24, 1975 (11:58:21 UT), at the epicentral coordinates (37.4, 21.6); (b) November 21, 1973 (5:07:04 UT and 5:32:25 UT), at the epicentral coordinates (34.32, --116.36) and (-- 36.54, 147.75). Fig. 2. Examples of the dynamic spectra of single signals detected in the close vicinity of the occurrence times of the following earthquakes: (a) January 17, 1977
(5:19:24 UT), at the epicentral coordinates (39.27, 43.7); (b) February 28, 1973 (6:37:54 UT), at the epicentral coordinates (50.514, 156.58). Fig. 3. Electromagnetic effect of the September 6, 1975 (9:20:11 UT) earthquake with the epicentral coordinates (38.513, 40.774). Fig. 4. Distribution of single pulses in 5-min intervals before and after an earthquake. Fig. 5. Distribution of twin pulses in 5-min intervals before and after an earthquake.