Analysis of the Gutenberg-Richter b-values of overlapping seismic clusters with application to Cooke 4 gold mine

The b-value of the Gutenberg-Richter frequency-magnitude relationship is an indicator of rock failure processes. Near-real-time analysis of the b-value has the potential to mitigate the risk posed by rockbursts, for example, by adjusting the geometry, sequence and rate of mining; or evaluating the re-entry time following a large seismic event. There are two main approaches to selecting a data set for b-value analysis: (i) select seismic events that fall within polygons or polyhedra associated with particular working places or seismic sources (e.g. a development end, stope or fault); or (ii) select seismic events that occur in the vicinity of each node of a 2D or 3D mesh that covers the entire region of interest. Challenges include the inevitable trade-off between statistical stability and space-time resolution, and overlaps of clusters of seismic events that arise from different sources. We wrote a Matlab code “Bplot” to conduct numerical simulations to investigate strategies to improve the resolution and reliability of b-value analysis. Bplot was also used to analyse seismicity during the extraction of the shaft pillar at Cooke 4 gold mine. Approximately 450 000 events, recorded from July 2011 to October 2011, were used to map spatial and temporal variations in the b-value. We find lower b-values close to the stope face. We attribute the higher b-value ahead of the stope to the occurrence of numerous small events caused by the fracture of intact rock by high stresses ahead of the mining front; while the relative increase in the number of larger events close to the face is considered to be the result of the growth and coalescence of these fractures.