Spatial clustering of seismic events in mines has been widely reported in literature. Despite obvious visual correlations between spatial clusters of seismic events and geomechanical structures in mines (such as pillars, dykes and faults), very limited research has been undertaken to utilise this information to filter seismic data. A linkage between spatial seismic event clusters and discrete rockmass failure mechanisms is tenuous and not well established using current seismic analysis techniques.

A seismic event clustering methodology is proposed. The first component of the methodology uses a complete-linkage (CLINK) clustering routine to identify relatively compact clusters of seismic events. The CLINK clusters are then subjected to a single-link clustering process, which uses spatial location and seismic source parameters as similarity measures. The resultant “Comprehensive Seismic Event Clustering” (CSEC) methodology can be used to identify individual seismic sources and rockmass failure mechanisms within a broad population of events. The CSEC methodology incorporates the vast majority of events within a population, such that the resultant clusters are a good representation of the entire seismic data set.

Within bounds, the clusters of seismic events identified in the CSEC methodology exhibit self-similar characteristics. Based on this clustering process a “Local Rockmass Seismic Source Model” is proposed. The model suggests that seismic events are primarily generated by local rockmass failure mechanisms. The local rockmass failure is a result of a combination of mining-induced stresses, geological features, rockmass conditions and mining influences (such as stope abutments, pillars and blasting practices).

Four seismic analysis techniques are proposed as tools to investigate clusters of seismic events. Magnitude-Time History analysis gives indications of seismic source mechanism, rate of rockmass failure and seismic hazard. Cumulative distributions of S-wave to P-wave energy ratios give quantitative data to assess seismic source mechanism. Apparent Stress Time History analysis is sensitive to stress increase and instability in a rockmass, creating an insightful link between mine blasting and the seismic response to mining. Finally, a combination of diurnal and phasor analysis can be used to identify if there is a statistically significant link between the time of day of seismic events and seismic source mechanism. Each of these analysis techniques is universal, reliable, valid, auditable and practical. Bounds and limitations of the analysis techniques are identified and discussed.

CSEC clustering, the “Local Rockmass Seismic Source Model” and new mine seismicity analysis techniques are used together to identify and characterise rockmass failure mechanisms in mines. Typical seismic behaviours are identified for mine abutments, pillars, faults and stiff dykes. The seismic nature of rockmass caving is also characterised, as well as the seismic profile for non-events associated with mine orepasses.

Finally, the CSEC methodology is shown to be useful in developing seismic hazard mine maps. An example is given in which seismic hazard is forecasted using CSEC clusters of seismic events.

Spatial clustering and seismic source parameter analysis are complementary tools that can be used to understand the seismic response to mining giving insight into rockmass failure mechanisms in underground hardrock mines. Numerous case studies from Australian and Canadian mines illustrate the application of these techniques.