Earthquake Reaearch in China  2018, Vol. 32 Issue (1): 130-140
The Background Variation of Natural Source ELF and Its EM Abnormal Phenomena in Yunnan Earthquakes
Fan Ye1,2, Tang Ji2, Han Bing2, Cui Tengfa2, Xu Qin1     
1. China Earthquake Networks Center, Beijing 100045, China;
2. State Key Laboratory of Earthquake Dynamics, Institute of Geology, CEA, Beijing 100029, China
Abstract: In order to add earthquake monitoring methods and develop new method research, the ELF Network for Earthquake Monitoring selected 30 stations in the Capital Circle and the Sichuan-Yunnan region. Finding electromagnetic field background variation is the basis of distinguishing the seismic electromagnetic anomalies. This paper introduces the data acquisition and selection of the Shexian, Anqiu, Lijiang and Dali stations which have recorded for longer time with better data and are located on the similar latitude. Then we use the natural source electromagnetic field's auto-power spectrum to express the intensity of the electromagnetic field. By using power spectral data of many frequencies in the observation frequency band, after the data pre-processing and sliding average noising, the background variation of extremely low frequency stations and the range ability were acquired. Taking the Baoshan M5.1 earthquake on October 30, 2015 and Dali M5.0 earthquake on May 18, 2016 as examples, the authors analyzed the earthquake electromagnetic anomaly characteristic of ELF stations around the earthquakes.
Key words: Natural source     Extremely low frequency(ELF)     Electric and magnetic fields auto-power spectrum     Background variation     Earthquake electromagnetic anomaly (EM)    


In recent decades, high temperature and high pressure rock experiments in laboratory and actual observation records show that relevant electromagnetic field radiation disturbances have occurred before earthquakes (Zhao Guoze et al., 2009; Qian Fuye et al., 2009; Gao Shude et al., 2010; Huang Qinghua, 2005). During the seismogenic process, electromagnetic radiation produced in the process of stress accumulation to rock fracture might cause a change of electromagnetic field data from the observation of the natural source. The phenomenon of electromagnetic anomalies is more obvious before the occurrence of extremely low frequency (3-30Hz) and ultra-low frequency (30-300Hz) earthquakes (Zhao Guoze et al., 2003; Tang Ji et al., 2007; Hao Jianguo et al., 1995; Qian Shuqing, 2009; Guan Huaping et al., 1995).

Chinese seismic stations usually use traditional electric or magnetic methods for precursory electromagnetic observation, and they only observe electric or magnetic field signals lower than 1Hz, not including ultra-low frequency/extremely low frequency (SLF/ELF) bands which are sensitive to earthquake events.

During the period of the tenth "Five-year Plan", 12 magnetotelluric (MT) observation stations were built in different provinces and autonomous regions. These stations recorded abnormal electromagnetic field data from the 2008 Wenchuan MS8.0 earthquake (Gao Shude et al., 2010), Yingjiang MS5.8 and Myanmar MS7.2 earthquakes (Gao Shude et al., 2013), and before and after strong earthquakes in Yunnan Province and its adjacent areas between 2009-2010 (Sun Weihuai et al., 2012). Based on the early stage of the extremely low frequency magnetotelluric earthquake cases research, 30 extremely low frequency observation stations were built under the major national science and technology infrastructure construction projects in the eleventh "Five-year Plan" in the Capital Circle and the Sichuan-Yunnan region. These stations are not only able to observe electricity and magnetic signals simultaneously, but also the space electromagnetic field and the change of ground resistivity, suitable for the advantages of four-dimensional seismic monitoring. At the same time, they can also accept the stable signals of the known launch time, frequency, and the strength from the same launch station. When the signal shows a large electromagnetic anomaly caused by disasters such as earthquakes, it is more conducive to determine the time and place of anomalies appear (Zhao Guoze et al., 2012), which has important physical significance and scientific prospects.

At present, 30 new extremely low frequency (ELF) stations have been completed, and have been commissioned to observe natural source magnetotelluric data for more than two years. The source is plane electromagnetic waves of ionosphere activities, changing with time and place. Identifying the electromagnetic background change rule to determine whether the observed value is deviating from the normal curve can determine the accurate extraction of earthquake-related electromagnetic information, so it is necessary for the statistics of background variation of extremely low frequency (ELF) electromagnetic station, to lay the foundation for the recognition of future earthquake electromagnetic anomaly.


The earthquake prediction subsystem of extremely low frequency (ELF) exploration has built 30 stations in the Capital Circle and the Sichuan-Yunnan regions. Among them, 15 stations are in the Capital Circle and 15 are in Sichuan and Yunnan Provinces. Distribution of stations is shown in Fig. 1. All stations are equipped with German ADU-07e instruments and observation and data acquisition are standardized, all are arranged in the direction of north-south, west-east, and the perpendicular electric field (west-east electric field Ex, north-south electric field Ey) and 3 magnetic fields (north-south magnetic field Hx, west-east magnetic field Hy, vertical magnetic field Hz). The observation device is shown in Fig. 2.

Fig. 1 The distribution of extremely low frequency stations The red stars are station locations

Fig. 2 Layout of observing instruments

The frequency range of the Nature Source observed by the extremely low frequency (ELF) station is 0.001-1000Hz. The instrument continuously records the low-frequency 16Hz data which is stored in the ADU-07e instrument. At the same time, in the non-artificial source period, the 4s with 4, 096Hz data and the 64s with 256Hz data is recorded every ten minutes (Fig. 3). The final output data are: time sequence of electromagnetic field in the observation area, auto-power spectrum in the electromagnetic field and the components in the magnetotelluric response (apparent resistivity, impedance phase, lean, etc.). Regional data includes the change of the electromagnetic field and underground structure change. In this paper, the frequency range observed by the extremely low frequency (ELF) station is referred to as the extremely low frequency (ELF) segment.

Fig. 3 The observation of nature source recording

The ELF electromagnetic stations output three kinds of file formats every day: equipment running status in the Log file format which contains the current record task, GPS information, power supply voltage, grounding resistance, and data quality information; the Original time series in ATS format file; and the data file in EDI format. The natural source of low frequency 16Hz continuous data acquisition, after dealing with the instrument, a daily EDI format file is output which covers impedance, 48 frequencies (0-1, 376Hz) under electromagnetic auto-power and cross-power spectrum data, etc. In this paper, we extracted 48 frequencies of the electromagnetic auto-power spectrum data and analyzed them in the ELF background field.

Taking the ELF station instrument operation and electromagnetic interference into the environment situation, we choose four stations which are all continuously observational, with better data quality and on similar latitude, such as Shexian (36.54°N, 113.64°E), Anqiu (36.37°N, 119.22°E), Lijiang (26.98°N, 100.17°E), and Dali (25.70°N, 100.18°E). In the continuous observation of natural source of the EDI format file, from the 48 frequencies of auto-power spectral data of Ex, Ey, Hx, Hy, Hz electromagnetic fields, choosing the 9.4Hz from the lower frequency, 74Hz from the medium frequency, and 766Hz from the higher frequency to analyze the background field variation.


First of all, we pre-processed the original data, deleted the obvious abnormal data four orders of magnitude more than the 10 days of observation data, and replaced it with the intermediate values of observation data of the near five days. To reduce the influence of factors such as human disturbance and environmental interference, the spectrum characteristics of the natural source will be highlighted.

In order to further reduce the environmental noise from the natural source electromagnetic field, the authors take the sliding average process: after pre-processing the data sequence, the adjacent 11 data items composed a window, the intermediate values of the 11 items instead of the value of the center of the data. Then remove the first data in 11 items, and then add the next data, forming the sliding 11 datum sequences, and so on.

In Fig. 4(a) and (b), the frequency of 9.4Hz electromagnetic auto-power spectrum data in Shexian and Anqiu in 2015-2016 are respectively plotted, abscissa for time, ordinate for Hx, Hy, Hz, the Ex and Ey from auto-power spectrum intensity. It can be obviously seen that the electromagnetic fields in the above-stated two stations have obvious seasonal variation, the sine curve is based on a one year period, and the intensity of the auto-power spectrum is weaker in winter, stronger in summer, with a maximum amplitude change within the four orders of magnitude, in accordance with the background variation of the 12 ELF stations constructed during the tenth "Five-year Plan"(Fan Ye, Tang Ji et al., 2012, 2013). Fig. 4(c) and (d), are respectively from Lijiang station began recording on September 14, 2015 and Dali station with nearly 16 months data also can see these variation obviously.

Fig. 4 The auto-spectrum power curves of EM, the black solid curves are the data after moving average (a) f=9.4Hz at Shexian station; (b) f=9.4Hz at Anqiu station; (c) f=9.4Hz at Lijiang station; (d) f=9.4Hz at Dali station

In Table 1 from three frequencies in the electromagnetic field auto-power spectrum intensity in the above four stations, it can be seen that the horizontal magnetic field variation with the seasons is commonly in 2 to 3 orders of magnitude, and in the vertical magnetic field, the electric field variation is between 2 to 4 orders of magnitude, and the vertical magnetic field signal strength is 1-2 orders weaker than that of horizontal magnitude. In Fig. 4 the annual variation of the vertical magnetic field in the electric magnetic field curve is not as obvious as the horizontal magnetic field, which is easily disturbed, the curve noise is great, and the intensity curve of the electric field auto-power spectrum is not as smooth as that of the magnetic field. This is because the electric signal and the vertical magnetic field of the local effect is bigger and easily influenced by the regional underground structure. the influence of various human interference surrounding the station is also big, but the horizontal magnetic field signal has better SNR and regional comparability. Because the maximum latitude difference of the four stations selected is about 10 degrees, the variation of latitude is not obvious.

Table 1 The EM field intensity distribution of ELF stations

An earthquake is a complex process. The electromagnetic amplitude observed anomaly is associated with a variety of factors, such as earthquake intensity, epicentre distance, propagation medium, geological structure, earthquake source mechanism and other factors. It is also related to the productive process of data (Huang Qinghua, 2005; Du Xuebin et al., 2004; Jin Nengjun et al, 1995; Hao Jianguo, 1995; Yuan Jiazhi et al., 1996). As the ELF station is a new station, considering the initial observation time, data continuity, station observation environment and magnitude ≥5.0 earthquake distribution for nearly two years and other factors, the authors take the examples of the October 30, 2015, Baoshan MS5.1 earthquake and May 18, 2016, Dali M5.0 earthquakes in Yunnan Province, and find the ELF stations of epicentral distance within 300km, then analyze the electric and magnetic fields auto-power spectrum curve and the information of earthquakes, to find the rule of the electromagnetic anomaly before the earthquakes, as shown in Table 2.

Table 2 The information of earthquakes

The variation of electromagnetic abnormal amplitude signal generally occurs about one month before the earthquake (Wang Jijun et al., 2005). Seismic analysis shows that the earthquake occurred after the EM anomalies ended and underwent a quiet period (Yang Zhaowang et al., 1998; Guan Huaping et al., 2000), and some of the electromagnetic anomalies before the earthquakes will continue until the main earthquake (Ma Qinzhong et al., 2003). However, the range of the EM amplitude anomalies related to earthquake standards is yet to be determined, so this paper will exclude all kinds of known interference, compared with ELF electromagnetic background fields, considering the electromagnetic anomaly over an order of magnitude as an index of the anomaly, and also considering the geomagnetic storm impact, download index ∑Kp of one and a half month before and after the MS5.1 Baoshan earthquake and the MS5.0 Dali earthquake from the web site ( of the world geomagnetic data center. It is generally acknowledged that there would be a geomagnetic storm on the day when the index ∑Kp is more than 30. Due to the index ∑Kp is world time, the station data is correspondingly one or two days after the ∑Kp index.

The alternating electromagnetic field source related to solar activity is random, due to the long cycle signal being attenuated slowly in propagation. The signals have larger amplitude when it reaches the ground, that is, the lower frequency has a stronger signal (Shi Yingjun etc., 1985), so we choose the frequency < 3Hz data here, this frequency signal strength is not easily disturbed. From October 1, 2015 to November 15, 2015, index ∑Kp, from October 7 to 9, November 3 to 4, and November 7 to 10, ∑Kp > 30, so we avoid the above period of time of active magnetic days. According to Fig. 5, the 0.27778Hz in the horizontal magnetic field in Yingjiang slightly decreased in the upward trend in October 20, and the change of the east-west electric field changed for nearly 3 orders. The five EM fields of 0.11628Hz in Ljiang station have the highest jump of nearly four orders of magnitude on October 14 and 15, 2015, and the horizontal magnetic field has nearly three orders of magnitude jump on October 17, 2015. The 1.2Hz of electromagnetic field in Dali decreased slightly on October 18, 2015, jumping up to 2 orders of magnitude then reduced to the original amplitude on October 19, 2015.

Fig. 5 The auto-spectrum power curves of stations around the Baoshan earthquake and the space activity index ∑Kp From top to bottom is the NS magnetic field, the EW magnetic field, the Vertical magnetic field, the NS electric field, the EW electric field respectively and the ∑Kp (a) f=0.27778Hz at Yingjiang; (b) f=0.11628Hz at Lijiang; (c) f=1.2Hz at Dali

From Fig. 6, the 1.2Hz electromagnetic field intensity of Dali showed an order of magnitude rise on April 18, 2016, recovered and then rose again. However, on April 26 and 27, 2016 it showed a synchronous downward trend, while the E-W magnetic field and vertical magnetic field changed two orders of magnitude. On May 7 to May 10, 2016 there was also synchronous stepping down. Checking the index ∑Kp, there is a suspicion that it was connected to electromagnetic interference on 8-9. The 0.011111Hz electromagnetic field intensity at Dali is the opposite of the trend of 1.2Hz at the end of April, and the maximum change is nearly 2 orders of magnitude. The magnetic field also had a maximum of two orders of magnitude at the end of April, but the electric field did not change significantly. The jump in early May is suspected to be related to the geomagnetic storms on May 2, 8-9. The 0.011111Hz E-W magnetic field of Lijiang had a large step for half a year for unknown reasons, so we suspected that this electrode was broken. The remaining 4 channels have a synchronous step on May 12, with the maximum change of 3 orders of magnitude on the magnetic static days.

Fig. 6 The auto-spectrum power curves of stations around Dali earthquake and the space activity index ∑Kp From top to bottom is the N-S magnetic field, the E-W magnetic field, the vertical magnetic field, the N-S electric field, the E-W electric field and the ∑Kp respectively. (a) f=1.2Hz at Dali; (b) f=0.011111Hz at Dali; (c) f=0.011111Hz at Lijiang

Based on the comprehensive analysis of two earthquakes and a large number of earthquake electromagnetic literature, the authors believe that the electromagnetic anomalies associated with earthquakes exist in reality, but there is difference in the changing forms. Some electromagnetic anomaly weakened first, then strengthened. Some of them suddenly increased, some suddenly decrease, and then restore calm. For example, the Dali station is 80km away from the epicenter, and observed the opposite changing forms of different frequencies, it is suggested that the suspected received seismic electromagnetic anomaly shows frequency selectivity.

The auto-power spectrum of the magnetic field before the two earthquakes has changed a lot and the electric field is relatively small. Strength of different directions of the auto-power spectrum electromagnetic field amplitude changes sometimes is not the same, such as in the Baoshan earthquake, where the amplitude change of the 1.2Hz north-south magnetic field at Dali was two orders of magnitude larger than that of east-west magnetic field, and east-west electric field was one order of magnitude bigger than north-south electric field. This indicates that the suspected seismic electromagnetic anomaly received by the ELF station has directivity. Of course, the manifestation of pre-earthquake electromagnetic anomalies may be related to the location of the earthquake, the mechanism of earthquake preparation and the structural conditions.


After summarizing and processing the natural source data of 30 extremely low frequency (ELF) stations from 2015 to 2016, the stations with relatively coherent and less environmental interference were selected for in-depth analysis. The electronic and magnetic fields auto-power spectrum data are extracted from the EDI files of the daily observation output of extremely low frequency (ELF) stations to describe the intensity of electromagnetic fields. After data pre-processing, sliding average de-noising and a series of data processing, the authors got the statistical scope of electric and magnetic field intensity range of each frequency, and the law of the ELF electromagnetic field background change: the sine is based on one year cycle changes, the trend of the electromagnetic field intensity showed seasonal changes being strong in summer and weak in winter; the strength of the horizontal magnetic field varies with the seasons in commonly 2 to 3 orders of magnitude, while the vertical magnetic field and the electric field due to the effect of regional underground structure is larger, and also vulnerable to environmental disturbance, with an annual change of between 2 to 4 orders of magnitude. The vertical magnetic field is one to two orders of magnitude weaker than the horizontal one. Since the latitude difference of the four stations selected is only 10 degrees, the latitudal variation is not obvious.

In consideration of the extremely low frequency (ELF) station observation time, initial data continuity, observing environment and nearly two years of MS≥5.0 earthquake distribution and other factors, taking the October 30, 2015 Baoshan MS5.1 earthquake in Yunnan and on May 18, 2016 Dali MS5.0 earthquake in Yunnan as examples, the authors select the Dali, Lijiang and Yingjiang stations with epicenter distances less than 300km, and analyzed the observed auto-power spectrum electric and magnetic curve anomaly. During a magnetic static day, an electromagnetic anomaly can appear before an earthquake. Its performance deviates from the annual sine change, disrupting its background change. There may be an accidental phenomenon in a single earthquake case research, and we should test many earthquake examples using the statistical analysis method. Because of the time, validation needs to be done with follow-up records and accumulation of earthquake events.

Earthquake electromagnetic prediction is mainly based on the empirical analysis of observational data, which requires a more intensive electromagnetic network layout to identify the relationship between the seismic activity and electromagnetic phenomenon of statistical regularity as well as the analysis of the earthquake electromagnetic anomaly genesis mechanism.

In China, there are more than one hundred and twenty electric field stations and more than one hundred and fifty geomagnetic stations with frequency observation in the extremely low frequency (ELF) band. If these background changing laws of electromagnetic means can be ascertained, and adding to the statistical analysis of earthquake cases in recent years, the authors will be able to more clearly distinguish the electromagnetic signals abnormally related to earthquakes and have a more profound understanding of the earthquake electromagnetic anomaly phenomena, which should also be the later working plan.

Du Xuebin, Ye Qing. The medium short and short term anomalies of earth resistivity and the reason for on impending precursor before Minle-Shandan earthquakes in 2003 at Shandan station[J]. Northwestern Seismological Journal, 2014, 26(4): 334–338.
Fan Ye, Tang Ji, Zhao Guoze, et al. The background variation of ELF electromagnetic field and its application in earthquake monitoring[J]. Seismological and Geomagnetic Observation and Research, 2012, 33(1): 1–9.
Fan Ye, Tang Ji, Zhao Guoze, et al. Schumann resonances variation observed from Electromagnetic monitoring stations[J]. Chinese J.Geophys, 2013, 56(7): 2369–2377.
Fan Ye. Background Variation of SR and Its Application in Earthquake Monitoring[D]. Master's thesis. Beijing: Institute of Geology, China Earthquake Administration, 2012(in Chinese with English abstract).
Guan Huaping, Liu Guiping. The study of the relationship between electromagnetic radiation anomalies and earthquakes[J]. Acta Seismologica Sinica, 1995, 17(2): 237–246.
Guan Huaping, Chen Zhiyong. Study on seismic precursor feature of electromagnetic radiation in the capital circle and its surrounding area[J]. Earthquake, 2000, 20(1): 65–70.
Gao Shude, Tang Ji, Du Xuebin, et al. The change characteristics of electromagnetic field from before to after Wenchuan MS8.0 earthquake[J]. Chinese J.Geophys, 2010, 53(3): 512–525.
Gao Shude, Tang Ji, Sun Wenhuai. Electromagnetic anomaly before the Yingjiang MS5.8 and Myanmar MS7.2 earthquakes[J]. Chinese J.Geophys, 2010, 56(5): 1538–1548.
Hao Jianguo, Zhang Yunfu, Pan Huaiwen, et al. The ELF electromagnetic anomaly and spectral characteristics before earthquake[J]. Acta Seismologica Sinica, 1995, 17(1): 81–88.
Huang Qinghua. The state of the art in seismic electromagnetic Observation[J]. Recent Developments in World Seismology, 2005, 11: 2–5.
Jin Nengjun, Wang Yingliang, et al. The electromagnetic wave and short forecast[J]. Journal of Seismology, 1995, 1: 55–58.
Ma Xinzhong, Yin Jingyuan, Gu Xuezhang. The electromagnetic anomalies observed at Chongming station and the Taiwan strong earthquake[J]. Earthquake, 2003, 23(4): 49–56.
Qian Fuye, Zhao Biru, Qian Wei, et al. Impending HRT wave precursors to the Wenchuan MS8.0 earthquake and methods of earthquake impeding prediction by using HRT wave[J]. Science in China (D), 2009, 52(10): 11–23.
Qian Shuqing. Observation and experiment research on seismological electromagnetic radiation and earthquake prediction[J]. Recent Developments in World Seismology, 2009(7): 57–62.
Sun Wenhuai, Tang Ji, et al. Research on the characteristics of anomalies in ELF electromagnetic observations in Yunnan Province[J]. Seismology and Geology, 2012, 34(3): 488–499.
Shi Yingjun, Liu Guodong, Wu Guangyao, et al. Magnetotelluric Sounding Tutorial[M]. Beijing: Seismological Press, 1985: 1-278.
Tang Ji, Zhao Guoze, Chen Xiaobin, et al. Introduction of payload for electromagnetic emissions on seismic satellite[J]. Progress in Geophysics, 2007, 22(3): 679–686.
Wang Jijun, Zhao Guoze, Zhan Yan, et al. Observations and studies on EM phenomena caused by earthquake in China[J]. Journal of Geodesy and Geodynamics, 2005, 25(2): 11–21.
Yuan Jiazhi, Gao Qiao, Qian Shuqing, et al. The observation and study of seismic electromagnetic radiation in ULF and VLF[J]. Acta Seismologica Sinica, 1996, 18(2): 272–275.
Yang Zhaowang, Shen Haibao. Anomalous analysis of seismic electromagnetic emission of MS6.1 South Yellow sea earthquake[J]. Journal of Seismology, 1998(3): 30–33.
Zhao Guoze, Zhan Yan, Wang Lifeng, et al. Electromagnetic anomaly before earthquakes measured by electromagnetic experiments[J]. Acta Seismologica Sinica (English Edition), 2009, 22: 395–402.
Zhao Guoze, Wang Lifeng, et al. A new electromagnetic technique for earthquake monitoring-CSELF and the first observational network[J]. Seismology and Geology, 2012, 34(4): 576–585.