The Kola Regional Seismological Centre of the Russian Academy of Sciences (KRSC) is a small organization in the town of Apatity on the Kola Peninsula, North-West Russia. Since 1982 we have been continuously monitoring seismic events in North-West Russia and the adjacent seas.
In the past, we have primarily used our own set of seismological stations (see Table 1), but in recent years we have also been using data from IRIS stations (KBS, LVZ, KEV, ARU, ALE, NRI, etc.) and the Scandinavian seismic arrays (ARCESS, SPITS, FINESA, HFS, NORESS) for analysis of complicated events.
Name
|
Latitude (N)
|
Longitude (E)
|
Type
|
Worked Since
|
Till
|
APA
|
67.558
|
33.442
|
Analog
|
1956
|
now
|
PLQ
|
66.410
|
32.750
|
Analog
|
1985
|
now
|
BRB
|
78.073
|
14.197
|
Analog
|
1982
|
1990
|
PYR
|
78.659
|
16.216
|
Analog
|
1983
|
1987
|
AMD
|
69.744
|
61.648
|
Analog
|
1983
|
1995
|
KHE
|
80.600
|
58.200
|
Analog
|
?
|
1990
|
APA
|
67.558
|
33.442
|
Digital 3-C
|
1991
|
now
|
AP0
|
67.603
|
32.994
|
Array
|
1992
|
now
|
AMD
|
69.744
|
61.648
|
Micro-array
|
1993
|
now
|
As a result, a large amount of information has been collected, including seismic bulletins and catalogues, digital wave forms data, digitized seismograms for selected events, results of spectral processing, etc.
Because of the mining activity in our region, a lot of artificial seismic signals have been registered here including open-pit, underground and underwater explosions. These events give us a good basis for developing some criteria to discriminate explosions and earthquakes.
Before 1992, KRSC used only analog seismic stations. All of our stations were equipped with SKM-3 short-period seismometers with the same amplitude-frequency response (amplification 50000 in frequency range 1.25-2 Hz). In addition, the Apatity station included long-period sensors SKD (Ts=25 sec). Seismograms from all the stations (excluding KHE) are stored in Apatity. Data from KHE were transferred by daily telex into KRSC.
In 1991, a cooperative program between KRSC and NORSAR (Norway) began. As the result, three digital seismological stations have been installed.
Because the IASPEI-91 model is not suitable for the Barents region [1], we use local travel-time curves developed from a set of strong explosions with known locations. In addition, a 350 ton underground calibration explosion was carried out in the Khibiny Massif on September 29, 1996 [2]. Our velocity model is a combination of the NORSAR model for shallow depths (above 200 km) and the IASPEI-91 model for below 200 km.
Date |
Our location
|
JHD
|
Remark
|
18.08.83 |
73.289N, 54.893E
|
73.358N, 54.943E
|
|
01.08.86 |
72.945N, 56.549E
|
73N, 56.7E
|
Located by Marshall
|
02.08.87 |
73.298N, 54.398E
|
73.324N, 54.597E
|
|
07.05.88 |
73.275N, 54.436E
|
73.314N, 54.557E
|
|
24.10.90 |
73.304N, 54.634E
|
73.317N, 54.803E
|
To confirm it we relocated several previous seismic events. As seen in Table 2, a comparison of our locations with the known locations indicates the model's capability. In addition, we have sufficient data to locate the small nuclear explosion [3] on the Novaya Zemlya test site on August 26, 1984. The result is shown in Figure 1.
Figure 1. Location of the smallest Soviet nuclear explosion using data by stations PYR, BRB, APA and AMD.
The KRSC detection and location software is based on our algorithm which is close to the generalized beamforming approach [4]. It operates well when data from several seismic stations are available. The Amderma station, however, is far from other seismic stations so we frequently have to locate weak near events using only its data. The small aperture makes it impossible to use beamforming or other procedures to determine array-style back azimuths. Moreover, high noise levels (probably due to construction work) are commonly present. Under such circumstances, the totally automatic processing often results in wrong phase association (true phases may be associated with a noise, etc). To avoid this error, we use a semi-automatic routine. An analyst marks approximately the phases and the automatic procedure is executed for the filtration, STA/LTA detection, and joint polarization analysis both for P and S phases. Although accuracy of this method is not high, it is often suitable for preliminary locations (see examples below).
To use this method an analyst has to look through large data volumes. To assist the analyst , we developed a variant of site-specific monitoring (SSM) [5]. It scans for pairs of detected phases and for each pair assumes as a hypothesis that the first phase is P-wave and the second one is the S-wave from an event inside a given region. The hypothesis is evaluated by joint polarization analysis for P and S phases as well as several additional criteria such as frequency and amplitude compatibility. Those pairs with estimations greater than some threshold correspond to real seismic events. Of course false alarms do exist, but their number is within reasonable limits.On August 16, 1997, five seismic events occurred near Amderma station. Two of the events were very similar events and occurred in the same location of the Kara sea (distance from Amderma is about 320 km). The wave forms are shown in Figure 2.
Figure 2. Wave forms for two events on August 16, 1997, in the Kara Sea (Amderma stations). Low frequencies filtered below 2 Hz.
The other two events were explosions near Vorkuta (about the same distance but to South-West from Amderma) and one event was too weak to locate.
The result of the site-specific monitoring for this day is shown in Figure 3. The SSM procedure detected and located the two Kara events and the two Vorkuta explosions. False alarms are also shown. The results of semi-automatic location for the Kara event (6.20 GMT) and the Vorkuta explosion (7.02 GMT) are given in the insertions.
Figure 3. Results of site-specific monitoring using Amderma station for August 16,1997. Examples of semi-automatic location (isolines of rating functions and their maxima corresponding to epicenters) are shown in the insertions.
As mentioned, we often record seismic events which can not be identified by traditional criteria like P/S ratio. For instance, an event on February 9, 1998 near Murmansk (69.18 N, 32.63 E, 16.51:07) was recorded by seismic arrays ARCESS and SPITS very differently. The S-wave amplitudes for SPITS were much less than the P-wave amplitudes independent of the bandpass filter used. ARCESS recordings, on the other hand, include strong S-wave and even Lg and Rg phases.
A striking event occurred on April 18, 1998 in the Norwegian Sea near Bear Island. Its wave forms (recorded at APA, ARCESS and SPITS), together with our location are shown in Figure 4.
Figure 4. Wave forms together with our location of the strange event on April 18,1998. Signs "? S" mark places where S-onsets could be.
The nearest station is SPITS (about 470 km) and itÕs recording contains no noticeable S wave in any frequency band. ARCESS (670km) recorded strong S, whereas APA (1020 km) registered the P-wave only in the band 8-12 Hz.
Note that the calculated epicenter of this event is the same (within the limits of location errors) as the position where the nuclear submarine "Komsomolets" found its rest in 1989 [6]. Could the nuclear submarine be the source of this mysterious event ?
The authors greatly appreciate the support and scientific advice of Dr. Frode Ringdal. We thank Dr. Ralph W. Alewine for his consistent support of KRSC's developments. Finally, we appreciate the encouragement by Dr. Gregory van der Vink to write this article.
Return to: IRIS Newsletter
Information
Return to: Title Page and Table of Contents
Continue to: Next Article