DISCRIMINATING SMALL EARTHQUAKES FROM QUARRY BLASTS USING PEAK AMPLITUDE RATIO Vmax/Hmax

MA, S., EATON, D., and DINEVA, S., Department of Earth Sciences, The University of Western Ontario, London, Ontario, Canada, N6A 5B7, sma44@uwo.ca, deaton@uwo.ca, sdineva@uwo.ca.

Widespread blasting activities in quarries, mines and road construction represent a significant problem for ensuring the accuracy and completeness for catalogs of small (M < 3) earthquakes in built-up areas. Although experienced seismologists can usually distinguish blast waveforms from those of earthquakes by inspection, the automatic detection and identification of blasts remains problematic, especially for real-time systems. Compared to earthquakes, horizontally propagating waves in the near-surface waveguide tend to dominate the waveforms of surface blasts. Therefore, we propose that an event can be classified as either a probable blast or a probable small earthquake based on the ratio of peak vertical amplitude Vmax to peak horizontal amplitude Hmax. For blasts, a small value of Vmax/Hmax and nearly synchronous (surface-wave) amplitude peaks are expected. For small earthquakes, a larger value of Vmax/Hmax and in most cases nearly synchronous (surface-wave) amplitude peaks are expected. In some rare cases the peak vertical amplitude is in the P-wave package. As a preliminary test of this method, we have examined broadband waveforms generated by 22 known blasts from two different quarries in southern Ontario, and compared them with recordings of 15 earthquakes of similar magnitude (1.3 < M < 3.8) from the same general region. The data were recorded in 2002-2005 by the Ontario POLARIS network at epicentral distance up to 30 km. The waveforms were filtered in the band 0.6 to 19 Hz. For one quarry (16 blasts) we found an average Vmax/Hmax of 0.299 with a standard deviation of 0.062; for a second quarry (6 blasts), an average Vmax/Hmax of 0.187 with a standard deviation of 0.060. The earthquakes examined are more variable, but show a significantly larger average Vmax/Hmax of 0.452 with a standard deviation of 0.138. We are continuing to test and develop this technique.