Full wave-form, automatic real-time monitoring of high sampling seismic data in a metal mine: detection, location, event classification and seismic repeater matching

Jannes Kinscher (1) , Kadek Palgunadi (2) , Natalia Poiata (3) , Francesca de Santis (1) , Pascal Bernard (2) , Emmanuelle Klein (1)

1 INERIS - Institut National de l'Environnement Industriel et des Risques
2 IPGP - Institut de Physique du Globe de Paris
3 NIEP - National Institute for Earth Physics [Romania] - Institutul Național pentru Fizica Pământului

Full-waveform based, automatic real-time seismic monitoring tools are becoming more and more standard in seismological surveys of natural earthquakes. These methods demonstrate significant improvement in the detection capacity of microseismic events and precursors of potential larger earthquakes. In mines, the implementation of such approaches is challenging due to the presence of a wide range of seismic noises related to mining activities with signatures similar to microseismic events. In addition, high sampling frequencies of seismic data (several kHz) used in these environments pose problems for real-time data transfer and processing. Here, we propose an adapted, full-waveform based automatic processing workflow for the Ineris seismic network located at the deep levels (> 1 km depth) of the Garpenberg metal mine (Sweden). To deal with high frequency sampling (8 kHz) we designed a pre-processing step based on a multi-frequency event detection scheme and first-order amplitude-based location. Final source location is then obtained by applying an array coherency based back-projection approach (BacktrackBB) on the preselected and reduced data set. We estimate that detection capacity compared to a usual triggered monitoring system is increased by at least a factor 100. Automatic event classification is achieved (at least for the strongest events) using several standard signal parameters based on shape, location, size and frequency content. Ongoing investigations aim to build an automatic identification and classification scheme for multiplet families with highly similar wave forms using cross-correlation and template matching techniques. Results from source mechanism and parameter analysis, relocation and spatio-temporal statistics have shown that multiplet families can be interpreted as classical seismic repeaters that repetitively slip over time periods of weeks to several months after certain production blasts. Following this hypothesis, an advanced repeater monitoring approach may allow to measure indirectly aseismic slip in the mining area and to provide criteria for seismic hazard evaluation such as asperity density and dynamic rupture potential.

Domaines: Sciences de la Terre

Jannes Kinscher, Kadek Palgunadi, Natalia Poiata, Francesca de Santis, Pascal Bernard, et al.. Full wave-form, automatic real-time monitoring of high sampling seismic data in a metal mine: detection, location, event classification and seismic repeater matching. AGU Fall Meeting 2018, Dec 2018, Washington, DC, United States. ⟨ineris-03239389⟩