Monitoring seismic wave speed by an active seismic source

Decreases in elastic wave speed around cracked zones prior to faulting in rock fracture experiments have been reported (e.g., Yukutake, 1989; Yoshimitsu et al., 2009). These decreases in wave speed have been considered to be associated with crack and fault growth based on non-destructive observation using X-ray CT scan (Kawakata et al., 1999). Meanwhile, there were some reports on the decreases in seismic wave speed along paths that cross the hypocentral area in periods including some large earthquakes. Uchida et al. (2002) analyzed seismic waveform with explosive sources before and after the 1998 northern Iwate prefecture earthquake, and they showed that the decrease in seismic wave speed approximately 0.1-0.9 % by the earthquake occurrence. Justin et al. (2007) reported the reduction in seismic wave speed accompanied with the 2003 Tokachi oki earthquake around the rupture area by using the four repeating earthquakes that occurred before and after the 2003 Tokachi oki earthquake. However, seismograms of explosive sources or repeating earthquakes are hard to be frequently recorded, which makes the time intervals of estimated seismic wave speed be too long to distinguish preseismic changes from coseismic and post seismic changes. In order to monitor crustal structures and detecting the variation of rock properties in the crust, a kind of active seismic source systems ACROSS (Accurately Controlled Routinely Operated Signal System) has been developed(e.g., Kunitomo and Kumazawa, 2004). We used the controlled seismic source ACROSS, which installed at the Tono mine, Gifu prefecture, central Japan and has been routinely operated by Tono Geoscience center of JAEA (Japan Atomic Energy Agency), automatically. Frequency modulated seismic waves are continuously radiated from approximately 10-20 Hz by eccentric rotation of the source. In order to investigate the stability of ACROSS signals, we used seismograms recorded at the 110m depth of Shobasama observing site, which is located very close to the ACROSS source (< 2 km). The seismograms were stacked by every 50 seconds length that corresponds to the period of modulation sequences. The stacked signals of one year interval indicate highly correlated waveform (> 0.97) without lags, and 10 times larger signal than the noise level in the ACROSS signal band.