Frequency-Magnitude Study Down To Magnitude -4.4 For Induced Seismicity Recorded At Mponeng Deep Gold Mine, South Africa

We investigate the lower magnitude limits and frequency-magnitude characteristics of the microseismic catalogue recorded with a high-sensitivity seismic network at the Mponeng mine, South Africa (Nakatani et al., SSA 2008 Proceedings). The network, composed of one 3-component accelerometer (50Hz to 25kHz) and 8 acoustic-emission sensors (700Hz to 200kHz), is located at a depth of 3543m and covers the limited volume of approx. 300x300x300m. The 3-component accelerometer was used to estimate the moment magnitude for the limited number of 135 very well recorded events (Mw in the range from -4.1 to -0.3). We use the relation between the moment magnitude estimated from accelerometer data and moment magnitude calculated from acoustic emission sensors to extend the catalogue to lower magnitudes. The magnitude of completeness of selected spatio-temporal subsets of the catalogue was estimated for: (1) post-blasting activity during working days, located more than 80 m from the network and (2) an aftershock sequence of a MW 1.9 event that occurred ~30 m from our network. Both datasets follow the Gutenberg-Richter (GR) frequency-magnitude relationship with no visible deviation from self-similar behavior of seismicity between Mw -4.4 and -1.9 for the aftershock sequence and between -3.5 and -1.5 for the post-blasting dataset. We estimated the magnitude of completeness of selected subset as low as -4.3 for the aftershock sequence and -3.4 for the post-blasting activity. Differences in magnitude of completeness are attributed to location of recorded activity and site effects. We do not observe Type A activity as reported in Richardson and Jordan (BSSA, 2002) study. To test the results obtained by GR analyzes and to get better insights into the influence of location and site effects, we estimated the probability-based magnitude of completeness for the aftershock sequence following Schorlemmer and Woessner (BSSA, 2008) was estimated for the aftershock sequence. This study reveals the strong influence of local geology, engineered structures and sensor directivity on the detection probability of events. We found the probability-based magnitude of completeness to vary throughout the observational space in agreement with results found by the GR study.