It is known that most of the tectonic earthquakes are related to existing fault motions, fault growth or formation of new faults. The processes of generation of such earthquakes, especially of large ones, should be reflected in the changes (variations) of geophysical fields near the fault zone (a zone of formation of the source of a future earthquake). Therefore, an important task in the search of earthquake precursors is to choose the forecast parameter reflecting the real-world geophysical processes in the lithosphere, the change of which is likely to testify for a forthcoming seismic event within a local lithospheric structure of the seismoactive zone. The most promising in these terms is the analysis of naturally occurring microseismic noises. Natural microseismic noises carry information about the whole spectrum of the crustal deformation processes of different energy levels – from tectonic plate motions and their related catastrophic earthquakes to lunar-solar tidal deformation processes and microearthquakes. In connection with a global transition to digital recording of seismic events and the growing number of seismic stations in the last decades, there is a worldwide development of methods for seismic monitoring based on the identification and analysis of components of the ocean-generated microseismic wavefield and microseismic noises, aimed at the search and acquisition of the prognostic data [1–7]. An example of a successful earthquake prediction based on microseism patterns is the prediction of the catastrophic March 11, 2011 Tohoky earthquake (Japan) with magnitude М = 9 [3, 4]. The well-proven precursor identification method is that based on the analysis of “tidally triggered” microseisms [1]. Short-term predictability for the Baikal rift system (BRS) based on the changes in spectral composition of microseismic noise before nearby large and moderate earthquakes is considered in [7, 8].

On December 9, 2020 at 21:44:34 UTC in the Baikal basin near the Selenga River delta, there was the Mw = 5.6 earthquake [9] named “Kudara” (Fig. 1). The earthquake source was located in the zone of the northeast-trending Delta fault typical for the central BRS [10]. The Kudara earthquake was accompanied by aftershock activity – over the period from December 9, 2020 to January 12, 2021, there were recorded more than 70 shocks of energy classes K = 5.3–12.2 [11]. Focal mechanisms of the Kudara earthquake obtained by different seismic agencies reflect the NW-SE crustal extension setting with normal faulting in the source, sometimes with a negligible strike-slip component (except for shear mechanism obtained by the Baikal Branch of the Federal Research Center “Geophysical Survey” of the Russian Academy of Sciences” – FRC GS RAS, no. 2 in Fig. 1). According to a population survey, the maximum intensity assessed VI–VII MSK-64 was observed in the Kudara village (at 16 km); in the near-field settlements at distances up to 50 km, the shaking intensity varied from V to VI [11]. Shocks with an intensity V were recorded at 22 to 253 km, including in large cities of Pribaikalye (Ulan-Ude, Irkutsk, Angarsk, Shelekhov, Usolye-Sibirskoe) [11].

Fig. 1.
figure 1

Location of seismic stations of the Baikal (yellow triangles) and Buryat (white triangles) branches of the FRC GS RAS, integrated monitoring test site “Buguldeika” (red triangle) and Kudara earthquake epicenter (asterisk). Numbers stand for the focal mechanisms: (1) GEOFONE, (2) Baikal Branch FRC GS RAS, (3) Institute de Physique du Globe de Paris, (4) Columbia University, USA, (5) Geoscience Australia.

Full size image

Near the western shore of Lake Baikal there are integrated geological disaster monitoring test sites of the IEC SB RAS – “Buguldeika” and “Priolkhonye” (assigned KYD and TRG codes from the names of the seismic stations, respectively, Fig. 1) which are a part of the SRF “Geodynamics and geochronology” of the Institute of the Earth’s Crust SB RAS and equipped for rock deformation monitoring. Besides, the test site “Buguldeika” is equipped with a broadband seismic station “Kuyada” (local code KYD, coordinates: 52.567° N, 106.136° E, height – 484 m) which has been in operation since November 30, 2020. The seismic station is equipped with three sensors (one vertical and two horizontal) and operates in continuous mode, sampling frequency – 100 samples per second, operating frequency range – from 120 s to 108 Hz. This frequency range makes it possible not only to record earthquakes but also catch and analyze microseismic noise variations. The rock deformation monitoring sites are installed on the basis of the authors’ instrumental complex [12]. The site “Buguldeika” is located in the zone of junction between local faults (Fig. 1, inset). The deformations therein are measured in orthogonal directions by two horizontal 10-m motion rod sensors buried in the ground at a depth of 2.5 m. At the site “Priolkhonye”, there are two installed measuring points in different structural situations: the first point is located in the fault zone, with the deformations measured in the well by a vertical 12-m motion rod sensor, the second point is located within an undisturbed block, with the deformations measured by one horizontal 10-m motion rod sensor buried at a depth of 2.5 m and oriented along the regional extension direction.

The deformation monitoring data showed the final phase of the Kudara earthquake generation at the sites “Buguldeika” and “Priolkhonye” (epicentral distances 37 and 45 km, respectively, Fig. 1). Though the monitoring sites “Buguldeika” and “Priolkhonye” are located relatively close to each other, the features of the generating earthquake manifested themselves in different ways. At the first site, these were the sinusoidal deformations with a month-long increase of oscillation amplitude prior the earthquake. At the second site, there was the start of exponential growth of deformations in the first point ten days before the earthquake, with no visible evidence of the earthquake generation in the second point [13]. The reason for the difference in the deformational features of the Kudara earthquake generation at the “Buguldeika” and “Priolkhonye” sites lies in different structural conditions of their location. The site “Buguldeika” is located at the intersection of two local fault zones within the wedge-shaped block isolated by the zones of regional faults – Primorsky and Morskoi (Fig. 1, inset). Due to significant amplitude of normal-fault displacement along the Morskoi fault, the upper part of the block on the southeastern side is in contact with the water lens and sedimentary infill of the South Baikal basin, incapable to transmit the acting regional extension thereon. An oscillatory nature of the deformations before the Kudara earthquake, identified based on the monitoring, is of secondary importance and related to rocking motion of the block under increasing extension of its underlying crustal horizons.

This paper attempts to identify possible precursors of the Kudara earthquake from the microseismic noise data. We have earlier identified possible precursors of the nearby large and moderate earthquakes in the BRS at the epicentral distances up to 80 km, which manifested themselves in the decrease in microseismic noise level over the periods from several hours to tens of minutes before the shock in the frequency range from 0.5 Hz and higher [7]. In case of the Kudara earthquake, no significant changes were found in amplitude-frequency characteristics of microseismic motions with these frequencies [7], so that we analyzed the low-frequency range from 0.01 to 1 Hz.

Analysis was made on the seismograms containing microseismic noise recorded before and after the earthquake at seismic stations located in the central BRS (with epicentral distances varied from 30 to 250 km, Fig. 1). The seismic station “Kuyada” started operating in continuous mode on November 30, 2020 (i.e. 10 days before the Kudara earthquake), which makes it difficult to analyze just the background oscillations before the earthquakes, so that the paper deals with the analysis of the time period from the beginning of recording to April 2022, in order to assess the characteristics of seismic noise in a quiescent state.

Based on the archival continuous seismic records, there were determined the average spectrum of microseismic noise and its polarization in the low-frequency range from 0.01 to 1 Hz for all seismic stations at distances up to 250 km. After that, the spectral-temporal analysis was made on 30-minute recordings of microseismic motions (Fig. 2), with plotted Swan Diagrams (spectrograms) and polarization diagrams showing the direction of motions in the horizontal plane. Finally, the currently obtained spectra and polarization diagrams were compared with the average spectra and polarization diagrams for each station.

Fig. 2.
figure 2

Spectrograms and polarization diagrams of microseismic noise in the frequency range from 0.01 to 1 Hz before and after the Kudara earthquake at the Kuyada station.

Full size image

The analysis of microseismic noise at the Kuyada station in the interval from 0.01 to 1 Hz showed a periodic increase in the amplitude of oscillations along the horizontal components in the frequency range from 0.01 to 0.1 Hz over the period from 10 days before the Kudara earthquake to 4 days after it (to 20:00 December 13, Fig. 2). Fourteen days before the Kudara earthquake and 9 hours after it, there was observed the maximum increase in the amplitude of oscillations – approximately 20 times relative to a quiescent background (Fig. 2). The frequencies of seismic wave radiation for the Kudara earthquake itself are 7–20 Hz for P-wave and 2–15 Hz for S-wave.

The whole period (December 1–14) at the Kuyada station also shows an abrupt change in the orientation of oscillations from chaotic to ordered northwest to southeast (Fig. 2). The background microseismic oscillations arise from a variety of endogenous and exogenous seismic sources and largely represent surface waves. The influence of multiple azimuthally different sources causes the absence of directional radiation of microseismic noise (Fig. 2). Amplitude-frequency characteristics of microseismic motions in a certain measuring point are relatively constant or vary depending on a season, anthropogenic load, meteorological parameters, breaking-wave effects, fault-zone influence, and other factors ([3, 8, 14] and others), but can change considerably under the influence of external factors such as environmental transformation at a large earthquake generation [1, 2, 4–7]. In our case, the influence of meteorological conditions and wave-breaking effects can be excluded—the wind velocity did not exceed 2–5 m/s, the wind was blowing in the direction of west and east, and only changed it to northwest on December 5 and 12 (https://www.gismeteo.ru). The anthropogenic factor is also excluded. Thus, it can be concluded that microseismic noises at the Kuyada station reflected the process of environmental transformation before the earthquake which is also confirmed by the observations on rock deformation anomalies [13].

The earthquake generation process should have been reflected on other stations near the epicenter, so that the paper involves analysis of the data obtained at seismic stations of the Baikal and Buryat branches FRC GS RAS, located 28 to 158 km from the epicenter in the South Baikal area (Fig. 1). Figure 3 displays the polarization diagrams of microseismic noise oscillations in the horizontal plane in the frequency range from 0.01 to 1 Hz for seismic stations located in the central BRS. No shift in the oscillation orientation was recorded at the remote stations (at a distance of more than 130 km from the epicenter), whereas the near-epicenter stations recorded an abrupt change therein. It is worthy of note that the records of oscillation orientations in the horizontal plane, obtained at the stations of different azimuthal location relative to the Kudara earthquake epicenter, differ markedly from each other but coincide with the azimuth for the epicenter. Some of these stations are sufficiently remote, with the orientation of oscillations perpendicular to the azimuth for the epicenter—they are located in large fault zones bordering the South Baikal basin or therebeyond.

Fig. 3.
figure 3

Polarization diagrams in the microseismic noise frequency range from 0.01 to 1 Hz before the Kudara earthquake at the stations of different azimuthal location relative to the epicenter.

Full size image

The comparison between spectrograms and polarization analysis on three-component (N–S, E–W and vertical) seismic data showed that an increase of oscillation amplitude occurred only for horizontal components, with no prominent changes observed for the vertical component amplitudes. The diagrams of directional motion of particles (polarization diagrams, Figs. 2, 3) also show the occurrence of oscillations in the horizontal plane. However, the polarization diagram of S-wave from the Kudara earthquake show the transverse-wave-typical polarization of oscillations in vertical and horizontal planes with a direction to the earthquake source (polarized SV- SH-waves, respectively). The absence of oscillations in the vertical plane is typical of longitudinal seismic waves, SH-waves, and surface Love waves. In case of the Kudara earthquake, the epicentral distances are too short for surface wave generation which indicates the predominance of body waves in the macroseismic noise field during the reporting period.

In order to identify the earthquake generation zone, there was performed a mapping of the segments with the origination point in the area of localization of seismic station with the azimuth oscillations obtained from the polarization diagram (these segments are shown by dashed lines in Fig. 4). The area of intersecting segments indicates the source of disturbance – a large seismic event generation zone. The observed deviation of fine azimuth for the epicenter and of oscillation orientation obtained from the polarization diagram are partially attributed to the use of oscillation-orientation azimuth average rather than the whole azimuth range wherein the oscillations are observed and to the dimensions of the epicentral zone of a large volume-expanded earthquake.

Fig. 4.
figure 4

Identification of a large generating earthquake source zone (yellow area) from the data on microseismic noise oscillation orientation. An asterisk indicates the Kudara earthquake; dashed lines show the azimuths of micriseismic noice oscillations, red-bordered triangles stand for the stations recorded the change in microseismic noise oscillation polarization with orientation to the earthquake epicenter. Legend: (1) DSS of the Baikal Branch FRC GS RAS; (2) DSS of the Buryat Branch FRC GS RAS; (3) state border; (4) borders of administrative division RF.

Full size image

The comparison between the deformation and microseismic monitoring data suggests that the microseismic noise field reflected the process of slow faulting before the earthquake and after the main shock with the subsequent attenuation. The proposed method provides the possibility for short-term identification of a large seismic event generation and of its location, and for taking the prevention measures at major hazard installations which are seismically monitored.