Earthquake prediction research would benefit if a technique was available to continuously monitoring the physical characteristics of a volume of rock which immediately surrounds a seismic source. This study employs the polarization properties of microseismic events in order to estimate the volume containing the near-source scatterers associated with the coda arrivals.
A three-component accelerometer installed for research purposes approximately 180 m ahead of an active stope in a gold mine was located at a depth of 3,200 m. Microseismic events that were analyzed occurred within 300 m of the accelerometers, and these events had an average seismic moment of 1012 dyne-cm, with an average P-wave corner frequency close to 1.5 kHz. In general, the seismograms showed a strong acceleration coda following the direct P and S arrivals. A polarization filter was utilized to investigate the linearity of these coda signals, and this in turn was used to estimate the location of the scattering sources for these P-coda waves. The small values for the deflection angles (deflection is defined relative to the backazimuth from the receiver to the source) that were recorded for the strong P-coda pulses suggested that the region of the scattering was close to the seismic source. The series of initial strong pulses associated with the P-coda waves (5 msec) resulted largely from forward scattering caused by velocity fluctuations; these variations occurred over a length greater than 1 m. After these initial pulses, a second partition of the P-coda wave consisted of backscattered energy, and these pulses were attributed to velocity and density fluctuations over a length much smaller than 1 m. Results of the analyses suggest that the volume containing the scatters can be approximated by a sphere of radius 35 m that encompasses the seismic source.
Two parameters, F~2 and DP, were investigated, and these enabled a comparison to be made of the degree of polarization between different records. The degree of polarization F2(t) changes from having a value one, for linear polarization, to a value zero for the unpolarized signal. The second parameter DP was defined as the ratio of the integral of the square of the ground velocity multiplied by the polarization filter [sin2(2 arctan F2(t))] to the integral of the square of the ground velocity. This parameter facilitated the removal of the unpolarized signal from DP. The analyses of 25 events yielded an average value for F~2 of 0.57 ± 0.06 for the P coda and 0.47 ± 0.05 for the S coda. The corresponding values for DP were 0.66 ± 0.08 for the P coda and 0.42 ± 0.11 for the S coda. The P-coda was found to have a high degree of polarization over a significant portion of the coda; this was not the case for the S coda.