Application of subspace detection on a surface seismic network monitoring a deep silver mine

Abstract

Seismic monitoring is an important tool for understanding and mitigating seismic risk in many mining operations, especially those with deep, burst-prone conditions. However, establishing and operating a seismic network that produces quality seismic data can be very expensive due to equipment and labour costs associated with installation and maintenance of seismic stations. Researchers with the National Institute for Occupational Safety and Health are exploring data processing techniques that can improve the quality of a seismic event catalogue that do not require installing additional instrumentation. This paper presents the application of subspace methods to increase event detection capabilities of a surface seismic network monitoring a deep underground metal mine in the northwest of the United States of America. Events recorded on the network from late April to early July 2016 are used to identify similar, lower-magnitude events that occurred in a 15-day study period in June of the same year. False detection rates were evaluated by comparing results with a catalogue generated by an in-mine seismic monitoring system, and by visually examining filtered continuous waveform data at the nearest stations to the underground workings. The number of successful event detections more than doubled, with no false detections. However, detected events included production blasts that required screening based on proximity to blasting time. Acceptable estimates of magnitudes and locations for newly detected events were obtained. The application of similar methodologies to other networks may substantially augment event catalogues and provide additional data that can be used in seismic risk analysis to improve mine safety. When continuous waveform data are stored, such processing may be undertaken long after data collection is complete — a particularly valuable capability for investigating emerging stability issues.