To study the preparation process of an earthquake, we should get closer to the hypocentre. In co-operation with ISS International, we had an experimental field (similar to200 in x similar to200 in) without dykes and faults in a South African deep gold mine. we have monitored more than 20,000 seismic events (M<2) with borehole tri-axial accelerometers. We used a data acquisition system with 15 kHz sampling and 120-dB dynamic range. In a remnant area at a distance of 100-200 in, an earthquake sequence associated with an Msimilar to2 event occurred. Significant changes in seismic parameters such as stress drop, b-value, energy index were observed associated with the sequence.
We installed Ishii's borehole strainmeters and could monitor as large a strain change as 10(-4) associated with mining. However, sampling was only 4 times per hour and 12 bit A/D. A distance from the M2 sequence was similar to100 m, being not close enough. Therefore, we found another field with several crosscut tunnels excavated across a fault where an M=2-3 event is expected. We installed the strainmeter within several metres of the fault to monitor both shear and normal strains on the fault. The strains are continuously digitized with a 24-bit 25-Hz A/D conversion. The Earth Tide (10(-8)) is dearly detected.
Seismicity induced by longwall mining can be compared to earthquake swarm induced by fluid intrusion to vertical dykes. We also found a decrease in energy index in the source area of M>5 events a few days before the major events.