To determine the average noise level, one-week period was taken before the self-isolation beginning and two weeks after it (from March 23 to April 12, 2020) and noise envelopes were constructed for each station with the overlay Yandex self-isolation index (Fig. 4). And although the average noise level and its temporal changes vary between stations in each city, the general trend of the usual level of social activity and its decrease during the population self-isolation period is similar. There are some differences in the ratio of the noise level recording in Yuzhno-Sakhalinsk between the broadband and short-period seismometers. As mentioned above, the broadband STS-2 is installed in a seismic pavilion located in a forest, and the short-period SPV-3k is installed in the basement of the Geophysical Survey building. In Fig. 4c, the noise level on the record from SPV-3k clearly agrees with the periods of activity of the service personnel increasing on working days and decreasing on weekends, while the noise level at STS-2 has a smoother line and evident noise peaks are clearly expressed in three impulses only. Such impulse anomalies can be traced on the records from the stations of all the cities; they were on March 25, April 1, and April 25, 2020, though the social activity has to be low during these days according to the Yandex self-isolation index.
The first two anomalies are large seismic events: on March 25, an earthquake occurred in the area of Paramushir Island with Ms = 7.5, and on April 01 s./st. has recorded a seismic event in the USA (West Idaho) with Mb = 4.9 [http://www.gsras.ru/new/eng/ssd.htm]. The April 5 anomaly is an atmospheric phenomenon (cyclone no. 28 according to [Mezentseva, Kaptyug, 2020]), affected the south of the Russian Far East during the day, the trajectory of which can be seen in Fig. 5a. In late March – early April, the studied area was significantly affected by cyclones and anticyclones (Fig. 5a). As for cyclones, at the beginning of the first decade of April, the Sea of Japan was influenced by a cyclone (no. 28), which developed south of Japan and, deepening significantly, moved to the northeast. The wind speed in the sea was 8–13 m/s, heaving of the sea was 1–2 m. On April 4, near the southern coast of Primorye, under the high-altitude center, the cyclone deepened, and then slowly moved eastward. On April 6, it entered the Sea of Okhotsk, causing a wind of up to 13–18 m/s in the Sea of Japan, heaving of the sea of 3–4 m and heavy precipitation. The very next day, cyclone no. 33 had its impact. Figure 5c presents the graphs of the power spectral density of the noise at the «Yuzhno-Sakhalinsk» s./st. for a six-hour segment of seismic noise record made on the day of the cyclone passage (April 5, 2020) and, for comparison, segments of similar duration in «noisy» (March 24, 2020) and «quiet» (April 11, 2020) days.
Figure 5. Map of cyclones (red lines) and anticyclones (blue lines) passage in the studied area (A). Power spectral density of the noise in the day of the cyclone passage, as well as in «noisy» and «quiet» periods for the «Yuzhno-Sakhalinsk» s./st. (B)
The impact of technogenic seismic noise can be estimated by means of the power spectral density (PSD) analysis. Figure 6 presents a spectral power of six channels of three stations during the period before self-isolation, we’ll call it «noisy», and a «quiet» period during self-isolation. It is evident from the given figure, that «quiet» period has affected the recorded seismic noise in the wide frequency range. On the records of short-period stations, its effect disappears at frequencies below 1 Hz, where anthropic noise is weaker and become manifest in the frequency range of 1–20 Hz for «noisy» days, that indicates a cultural activity. The records of broadband seismometers (Fig. 6a) demonstrate a similar trend of change in power spectral density at the frequencies of 1–20 Hz, as for short-period ones, and there is an obvious decrease in microseismic noise during a «quiet» period at a frequency of up to 1 Hz.
The study results indicate, that the measures taken by the Government have caused the decrease in social and economic activity, that, in turn, has affected the seismic data from all the stations. However, this influence character changes depending not only on the locality, but also on the stations location, that does not contradict the data obtained by foreign authors [Poli et al., 2020; Pandey et al., 2020] and some studies fulfilled in the Russian Federation for Moscow [http://www.ceme.gsras.ru/new/news/news-20200402.htm].
Using the «Method of coherent separation of weak continuous seismic signals» (included in the Spectrum program), it was revealed that the effect of noise at different frequencies is uneven throughout the day. We assessed the impact at the frequencies of 4, 11 and 17 Hz (typical for manifestations of technogenic noise sources) for all the stations within the period from March 23 to April 03, 2020. This period included the working week before the self-isolation regime was declared, weekends and the first non-working week after introducing the self-isolation regime. And although, as we have already found out, the average noise level and its change over time are different between the stations, there are several general characteristics. For illustration purposes (Fig. 7), graphs of the noise level in local time of each city were plotted according to the data of short-period seismometers. These graphs have a clearly pronounced diurnal variation of microseisms, that characterizes the population activity in the studied areas. Thus, morning and evening activities on «noisy» days are clearly expressed on the graphs of a signal amplitude fluctuations for the selected harmonics, while there is a flatter graph for the «quiet» days after self-isolation has been introduced.
We have also studied the spectral composition of the seismic noise and revealed its change depending on social activity. Figure 8 presents the averaged spectrograms of the signal recorded on the vertical components of the short-period seismometers of three stations at 12:00 local time from March 23 to April 3 in the range of 5–20 Hz. March 23–27 are the days before the self-isolation regime has been declared, March 28–29 – the weekends before a non-working week, March 30 - April 3 – non-working days. The graph clearly shows, that the seismic noise level within the 5–20 Hz frequency range significantly exceeds the noise level after the self-isolation beginning for most stations.
First of all, the excess of the amplitude of the seismic noise level can be distinguished in Fig. 8 within the entire frequency range for the «Yuzhno-Sakhalinsk» s./st. This testifies to the fact that, due to the isolation of the station from the noise of the city, the activity of the station personnel, limited after the self-isolation regime has been introduced, is, in fact, the only source of technogenic noise. For the «Khabarovsk» s./st., which is located in the central part of the city, the effect of a significant decrease in the level of the seismic noise amplitude is not observed. As for the «Vladivostok» s./st., which is also in the center of the city, there are separate frequencies, where significant excess of amplitudes is observed during the working «noisy» periods. Thus, there is a constant interference at a frequency of 16.7 Hz on the working days, that is typical of the operation of asynchronous three-phase industrial AC motors with a speed of 1000 rpm used in electric drives of various devices, mechanisms and machines (pumps, fans, compressors, etc.), which is obviously related to the operation of the building, where the seismic station is located.
As a result of reducing the noise level, the seismic events recording is considerably simplified. As an example, consider the record on seismograms of two earthquakes occurred in the Qinghai region (China), which have the same coordinates, depths and magnitudes. The first earthquake occurred on 06.05.2018 (09:23 UTC; 34.55N, 96.47E; H = 15 km; Mb = 5.0), the second one took place on 01.04.2020 (12:23 UTC; 33.13N, 99.00E; H = 10 km; Mb = 5.5) during the self-isolation period [http://www.gsras.ru/new/eng/ssd.htm]. Figure 9 shows the records of earthquake waveforms on the vertical components of the broadband channel of the three stations concerned, the records are made on the same scale and are presented in a «pure» form (without filtering). For each station, there are the records of two vertical channels, there is a record of 2018 earthquake on one of them (top), and the 2020 record – on the other (bottom).
There is a significant level of interference in the records of the 2018 seismic event, which makes it difficult to determine the time of seismic arrivals, as well as to determine the arrivals signs and clarity. This interference almost completely swamps the seismic channel recording, which makes it impossible to process the data without applying the filtration. On the record of the 2020 earthquake, not only surface waves are clearly distinguished, but also P-wave arrivals, the purity of the recording of which directly affects the determination of the earthquake focal mechanism by means of the method of the P-wave first-motion polarity [Prytkov et al., 2018]. The almost complete absence of interference suggests, that there is low social activity in the areas where the seismic stations are located.
In order to determine the frequencies, at which the interference occurs, the power spectral density (PSD) of noise and the Fourier amplitude spectrum (FFT) analysis of our earthquakes was carried out (Fig. 10). From the data obtained, it is seen, that the main source of interference on the seismograms is technogenic noise within the 1–20 Hz frequency band, which is generated by the locality and its industrial facilities.
So, we can conclude that the influence of technogenic noise makes it difficult to isolate and process the records of seismic events and shows the inexpediency of placing seismic stations in densely populated parts of cities.