Active Monitoring of a Fault with Seismogenic Potential Using an Ultrasonic Transmission at 1 km Deep in the Ezulwini Mine, South Africa

Monday, 15 December 2014
Hironori Kawakata1,2, Nana Yoshimitsu3, Masao Nakatani1,3, Joachim Philipp4, Makoto Naoi1,5, Anthony Ward6, Issei Doi1,7, Thabang Masakale8, Raymond J Durrheim1,9, Luiz Ribeiro6, Sylvester Morema6 and Hiroshi Ogasawara1,2, (1)SATREPS, Tokyo, Japan, (2)Ritsumeikan University, Kusatsu Shiga, Japan, (3)University of Tokyo, Bunkyo-ku, Japan, (4)GMuG Gesellschaft für Materialprüfung und Geophysik mbH, Bad Nauheim, Germany, (5)Kyoto University, Kyoto, Japan, (6)Seismogen CC, Carletonville, South Africa, (7)Disaster Prevention Research Institute, Kyoto University, Kyoto, Japan, (8)Open House Management Solutions, Potchefstroom, South Africa, (9)Council for Scientific and Industrial Research, Johannesburg, South Africa
Not only the stress state but also the elastic and inelastic properties of rocks around a target fault may be key information on seismogenic processes. Monitoring them may help us improve earthquake hazard assessment. In laboratory, elastic wave speed has been found to decrease prior to the main fracture (e.g., Lockner et al., 1977; Yoshimitsu et al., 2009). At the Ezulwini mine in South Africa, a fault with a high potential for mining-induced earthquakes with a relatively large magnitude (M ~2) was specified based on boring-core observation, tunnel wall observation, and mining plans. We started monitoring transmitted waves across the fault at about 1 km deep (Kawakata et al., 2011). We installed a piezoelectric transmitter as a wave source about 20 m away from the fault in the hanging wall. Three accelerometers of 3-component were installed along the acoustic beam from the transmitter; one in the hanging wall, and the other two in the footwall. Ultrasonic pulses were repeatedly transmitted, and the received waves were recorded at 400 ksps. Transmitted signals can be clearly recognized in stacked waveforms of all channels. In May 2012, increases in travel times (5-10 ms) from 2011 were observed for all the three stations. The common delay suggested that the decrease in wave speed was not localized at the fault. After that, there was a long strike in the mine, and the data were missed for about 100 days and 300 days due to the lack of maintenance of the monitoring system. By September 2013, when the data recording was resumed, travel times to the two receivers located after crossing the fault had increased greatly to ~60 ms, while that to the receiver before the fault increased only to 20 ms. Assuming that the thickness of the fault is 5 m, the difference in delay of 40 ms corresponds to a wave speed decrease of 8 %. We have not identified a relatively large event that might be associated with this change in travel times, but the event search is not completed.