Rock bursts and collapses are the most dangerous phenomena associated with seismicity in underground mines. The collapse of a mine roof can follow strong and shallow seismic events of magnitudes M > 3 [1
]. They pose a serious threat to life for miners working in the vicinity of the collapsing rock. The strongest underground mining-induced seismic events are Volkershausen [5
] and Newcastle [6
], both M5.6 Although the magnitudes of such shallow seismic events are usually small in comparison with natural earthquakes, the ground shaking might be felt by local citizens and can damage surface infrastructure. Rock-burst issues have occurred in the Polish copper region known as the Legnica Głogów Copper District (LGCD). Thousands of events of M > 1.0 are recorded every year in the LGCD. Among these, several are followed by rock bursts or tunnel collapses. Both ground shaking (i.e., a seismic event) and local ground failure occur as consequences of mining. While the shaking effects are comparable to what can be felt during small to moderate natural earthquakes, the surface deformation is specific to mining areas. The subsidence is either a ground response to the closure of excavated mining panels, e.g., [7
], or is related to strong tremors [10
]. In this paper, we present a joint seismological and geodetic study that expands our knowledge concerning the connection between mine collapses and the ground response to seismic sources manifested by permanent ground deformations.
The analysis focused on the strongest seismic event recorded by LGCD in the last several years—namely, the 29 January 2019 M4.6 collapse in the Rudna Mine, Poland. The phenomenon was widely felt and was followed by massive collapse. While the seismological analyses were carried out using records from local and regional seismic networks, the geodetic data are comprised of Global Navigation Satellite System (GNSS) observations collected at station LES1 (Komorniki; part of the European Plate Observing System project [EPOS-PL] network) and the European Space Agency’s (ESA’s) Sentinel-1 radar observations. The LES1 station was aligned with a strong motion sensor (KOMR), part of the local seismic system, the Legnica–Głogow Underground Mining INduced Earthquake Observing System (LUMINEOS), located above the mine. Interferometric Synthetic Aperture Radar (InSAR) techniques were applied to analyse the co-seismic surface deformation.
2. The Collapse and Seismological Analyses
On 29 January 2019, Rudna Mine in western Poland was struck by a seismic event followed by a rock burst. It was one of a few such events in the area in recent years [4
]. Using seismic signals recorded by the LUMINEOS surface local seismic network (Figure 1
a) and a local velocity model [14
], the epicentre was located at 51.5110°N, 16.1197°E. For our analysis, we assumed that the event depth was located at the level of the excavation; that is, 1 km below the surface. LUMINEOS consisted of 10 strong motion sensors (AC-73) and 17 five-second VE-53/BB instruments (both types manufactured by GeoSIG, Schlieren, Switzerland).
The source mechanism of the event was estimated using seismic signals recorded by broad-band seismometers (STS2 or Güralp CMG3ESP; Figure 1
b) at regional distances of 70 to 300 km. The signals and corresponding station responses were obtained from the ORFEUS-EIDA project page [15
]. The inversion was performed using the real displacement, derived from the original signals, for 11 seismic stations located around the epicentre (Figure 1
b). The mechanism (Figure 2
a) was estimated in terms of the full moment tensor, which can be directly used to interpret the mechanisms of mining tremors [16
]. The Pyrocko toolbox [17
] and KIWI tool software [18
] were used for the calculations. The software had already been tested as a tool for determining a stable point source model in mines [19
], including mine collapses [4
]. In this work, we followed the procedure previously used in [4
]. The inversion was based on the full waveform displacement and spectra in the specific frequency range. The regional velocity model proposed by [20
] was used. The lowest misfit was obtained for data filtered between 0.07 and 0.11 Hz. The final moment tensor solution was decomposed according to [21
] into its isotropic and deviatoric parts, and ultimately into the double-couple and compensated linear vector dipole. The resulting moment tensor (m11: −4.03 × 1015
m22: −3.91 × 1015
m33: −11.9 × 1015
m12: −0.07 × 1015
m13: −0.11 × 1015
m23: −0.22 × 1015
Nm) was dominated by the implosive component (isotropic part: −55%). This kind of mechanism can be explained by a tabular cavity collapse [4
], and it was strongly supported by the observations recorded on the strong motion sensors belonging to the LUMINEOS network(see Supporting Material: Table S1
). Figure 2
shows the distribution of the peak ground acceleration around the epicentre. The clearly visible peak ground acceleration pattern corresponds very well with the collapse source model.
5. Discussion and Conclusions
Mining-induced seismicity and the associated ground deformation, measured by advanced geodetic observations, have not been studied by many authors, except in the cases of controlled salt mine cavity collapse [46
], examples in the Upper Silesia Coal Basin [7
], and in the LGCD [51
]. Previous works concerning ground deformation studies in mining areas, however, were focused on geodetic data only (i.e., remote sensing or GNSS) ([46
] and references therein) or used just very basic seismological parameters, usually epicentre locations provided by various seismological centres [13
]. Some specific models of ground deformations have also been considered previously [54
]. Since a mining collapse is actually an isotropic seismic source, these models seem to be in very good agreement with the seismological model of mining collapse. On the other hand, the impact of mining tremor deals with a very minor number of existing subsidence basins in mining areas. Hence, it is difficult to judge, using the geodetic data only, whether the subsidence is directly connected with a tremor or not. In all these studies, it was postulated that the increased seismicity manifested itself on the ground as a higher subsidence rate, with the measurable effects of post-seismic vertical deformation reaching up to 10 cm [52
]. We applied several different techniques in order to gain a better understanding of the characteristics of the strong anthropogenic tremors induced by mining activities. These results not only support similar previous studies, but also contribute new findings.
First of all, we determined that shallow-induced seismic events with magnitudes of M4 and above can manifest not only as a relatively strong ground motion but can also provoke significant surface deformation in a very short time. Both effects have been observed in mining areas [6
], but rarely in such a detailed way as presented here. The combination of different long- and short-time monitoring techniques provided an opportunity to track the process of evolving surface deformation in a complex regime of mining subsidence and anthropogenic seismicity in detail. Thanks to the GNSS and InSAR analysis, the rate and delineation of the subsidence was determined quite well, and we confirmed the rapid subsidence of the ground in relation to the mine roof collapse events.
High-rate GNSS observations have been applied in earthquake analysis and structural health monitoring. Previous studies of seismogeodesy have been concerned with large natural earthquakes with magnitudes exceeding M4.5 and amplitudes reaching decimetres and even meters (e.g., [27
]). The event from 29th January is the first example presented of HR-GNSS application in short-term deformations caused by induced earthquakes. Studies related to induced shocks have involved long-term displacements and standard sampling frequencies of up to 1 Hz only [58
]. In this study, the greater potential of the HR-GNSS for small events with centimetre-scale amplitudes was confirmed by a comparison with the seismological data. The short-term GNSS observation revealed the possibility of a wider interpretation of the source mechanism in the future. Data from the collocated GNSS and seismic sensors showed very good agreement in the displacement time-series, which indicates that the GNSS stations can be used for additional, alternative data, recorded around an epicentre in an area with sparse seismic sensors. On the other hand, the long-term GNSS study revealed the possible detection of other high-energy seismic events in the area. However, this is highly dependent on the distance between the GNSS station and the seismic sources and requires extended investigation.
The spatial range of the surface deformation derived from the InSAR studies corresponded well with the estimated location and focal mechanism of the event. The DInSAR results showed an undoubted capability to detect and delineate the scale of the surface deformations caused by such an event, as proved by the comparison between the ascending and descending data, even though the ascending co-seismic interferogram (12-day) had a lower quality. The combination of the SBAS and DInSAR results demonstrated the intensive subsidence with varying velocity associated with mining activities. Continuous subsidence occurred in the area at different velocities, depending on the dynamics of the material being extracted and also the local seismicity.
In principle, the source parameters can be derived directly or indirectly from seismograms recorded around the hypocenter. The most important and interesting in the case studied in this work is the source mechanism, which in fact is quite different in comparison with natural earthquakes that occurred on faults. In such regions, the dominance of the double couple mechanisms is observed, and no isotropic sources are expected. Concerning the mechanism, since the subsidence was completed in a very short time after the quake happened, the satellite observations can be used as an indicator for the source type determination. In our case, the geodetic studies strongly supported the full moment tensor estimated with seismological analysis. We also gained better knowledge about the size of the rupture thanks to the additional study of the seismic and geodetic data. This is quite unique in the case of seismicity related to underground mining. Our results also help in better understanding the influence of the collapse like events on surface vibration measured in terms of PGA/PGV. By geodetic techniques, we investigated the surface for possible deformations before and after the event in order to verify the proper detection of the affected area. Last, but not least, it was previously shown that seismic sources with visible rock bursts (e.g., tunnel collapses) in this area are complex (see, e.g., [4
]). This is not obvious seismological observation and is crucial to understand the phenomena in the future. Since the subsidence is directly connected with the collapse, using geodetic analysis we would be ready to say more about the source physics.
The outcomes from the current study confirm the potential of the applied techniques for investigating ground vibrations and the surface deformation related to them. The studied event highlighted the need for a more detailed study, focused on several areas, starting with the highly complex nature of the ground deformations in areas where mining activities occur. The combination of factors provoking such ground changes, such as the very fast motions caused by moderate tremors with different magnitudes, the fast and slow nonlinear subsidence of the surface triggered by material extraction, and also the change in subsidence rates between the pre- and post-earthquake periods, the local geological structures, underground water-level changes, etc., produce a complex dynamic regime at the mine surface and in the behaviour of the cavity. The combination of different techniques provides a better understanding of the mining and its induced seismicity influence on the surface.
To increase the impact of such joint investigations, several recommendations can be made. A high density of GNSS stations, in areas with significant seismic risk also provoked by human activities, is recommended. A high density will allow a more detailed, long-term study of the horizontal and vertical trends of the surface in the area. On the other hand, the results from short-term GNSS measurements could be used to densify and verify the seismic recordings. Combined observation of long time series of seismological and geodetic data (GNSS, InSAR) could reinforce the possibility of detection of pre-seismic markers, enabling it to become a method for developing an early warning system in mining areas.
Undoubtedly, the investigation of this event has provided a new way for analysing hazardous rock bursts in underground mines. The current study has proved the high potential for obtaining a broader understanding of anthropogenic hazards by combining different Earth monitoring techniques. Joint geodetic and seismological observations can complement each other and help to overcome the individual disadvantages of the separate methods. This approach can be recommended as a complex tool, which can be used to gain a better understanding of the seismicity associated with all kinds of anthropogenic activities. The method would be especially useful in areas with sparse seismic monitoring and/or regions with already known isotropic source mechanisms, such as mining or volcanic.