Rupture processes and magnitude statistics of induced earthquakes and aftershocks: A detailed view from deep South African gold mines
Geophysics
Final Report Abstract
The combination of three sources of earthquake data − waveform recordings from a surface network, from a deep in-mine network, and a comprehensive catalog with about 70,000 earthquakes − allowed for detailed investigations of seismogenic processes in a spatially confined, seismically active volume. Using a variety of advanced waveform-based amd probabilistic methods, we achieved the main aims of our project, it means imaging and directivity of earthquake ruptures in short epicentral distances and a quantification and potential discrimination between anthropogenic and tectonic seismicity based on statistical characteristics. We separated the cataloged seismicity and grouped the events into three different classes: (1) the aftershock sequence of the Orkney M5.5 earthquake (≈ 15,000 EQ), (2) seismicity induced by flooding (≈ 48,000 EQ), and (3) seismicity induced by mining activity (≈ 7,000 EQ). We studied statistical properties of earthquakes in each of the three classes and compared jointly the results. We focused on the magnitude statistics, on space-time-energy patterns for cluster identification, and on the statistics of the dynamic stress drop. Another key aspect in the conducted research program was the imaging of the propagating rupture of the M5.5 earthquake. We applied successfully two different methods and retrieved comparable results which also agreed with further independent observations. Our main results and conclusions are: • the rupture of the M5.5 earthquake propagated predominantly unilaterally, nearly from North to South over a distance of about 5km; • the images of back-projected seismic energy revealed a highly complex rupture process; • the hypocenters of the aftershock sequence are unilaterally situated in respect to the hypocenter of the main shock and are aligned to the South confirming the obtained rupture propagation image and directivity; • the magnitude statistics of both aftershocks and induced earthquakes are noticeably affected by the finite size of the rock volume of stress perturbation which inhibits the occurrence of larger magnitude events; • the observed magnitude frequency distributions of aftershocks and induced earthquakes can be described with the Lower Bound model of magnitude probabilities; • the statistics of event waiting times are different for aftershocks and induced seismicity: the temporal distribution of aftershocks shows clearly the typical Omori-Utsu behavior (i.e., a strict power-law decay), the temporal distribution of induced earthquakes exhibits a Poissonian nature of independently occurring events; • the cluster identification analysis confirmed the existence of two statistically distinct subpopulations of events in the observed seismicity in Orkney: induced events are mainly stationary distributed in time but inhomogeneously distributed in space, aftershocks are clustered events which are closer to each other in time and notably in space; • the statistics of dynamic stress drop of aftershocks as well as of induced earthquakes comply with a log-normal distribution but the value range is different, i.e., aftershocks are generally characterized by higher stress drops and seismicity induced by flooding shows the lowest stress drops; • spatial variations of the stress drop of aftershocks and induced earthquakes anticorrelate slightly with spatial variations of the Gutenberg-Richter b value. Our findings and conclusions have basically and essentially the potential to contribute to an improved assessment and a possible mitigation of seismic hazard in active mining areas. In particular, we think that the Lower Bound model combined either with the Omori-Utsu model for aftershocks or with a Poissonian model for induced earthquakes would be a first step towards more realistic hazard estimations of mining operations.
Publications
- (2018): Rupture Directivity of Earthquakes below Gold Mines in South Africa, EAGE 80. Jahrestagung
D. Wehner, J. Folesky, C. Dinske, J. Kummerow and S. A. Shapiro
(See online at https://doi.org/10.3997/2214-4609.201801692) - (2018): Rupture Imaging and Directivity of the 2014 M5.5 Earthquake Below a Gold Mine in Orkney, South Africa, AGU Fall Meeting, S21C-0439
C. Dinske, M. Müller, M. Voigt, D. Wehner, J. Folesky, J. Kummerow, S. A. Shapiro and H. Ogasawara
- (2018): Rupture Processes and Magnitude Statistics of the 2014 M5.5 Earthquake Sequence Below a Gold Mine in Orkney, South Africa, Geophysical Research Abstracts, Vol. 20, EGU2018-7211
C. Dinske, M. Müller, M. Voigt, D. Wehner, J. Folesky, J. Kummerow and S. A. Shapiro
- (2019): Detecting Earthquakes Below South African Gold Mines Using Event Template Matching, DGG 79. Jahrestagung, Vol. 79, p. 136, SM2.01
M. Voigt
- (2019): Rupture Imaging and Directivity of the 2014 M5.5 Earthquake Below a Gold Mine in Orkney, South Africa, DGG 79. Jahrestagung, Vol. 79, p. 153, SO1.04
C. Dinske, M. Müller, M. Voigt, J. Kummerow, S. A. Shapiro and H. Ogasawara
- (2019): Rupture Imaging and Directivity of the 2014 M5.5 Earthquake Below a Gold Mine in Orkney, South Africa, Geophysical Research Abstracts, Vol. 21. EGU2019-7924
C. Dinske, M. Müller, M. Voigt, J. Folesky, J. Kummerow, S. A. Shapiro and H. Ogasawara
