Cluster Analysis of Closely Spaced Mining Blasts as a Method of Event Location

摘要

The aim of this research was to characterize the high frequency content of noiseand signals, and to develop methods of discrimination between mine blasts and other seismic sources using high frequency and seismic data. Two previous reports (Israelsson et al., 1990; and Israelsson and Carter, 1991) describe much of the research conducted under this contract and covered the high frequency characteristics of local, regional, and teleseismic waveforms as recorded at the short period Scandinavian arrays and the NRDC high frequency stations in the Soviet Union. The final six months were devoted to a study of mining event characterization for location and discrimination purposes. Cluster analysis was used to group events with similar characteristics from the mining district of Karelian, northwest of St. Petersburg. The groupings compared well with a careful visual classification of the same data and were hypothesized to be associated specific mines. The results of this study are very promising and warrant expanding to other mining districts. Future studies based on this work, however, should verify the event sources using non-seismic data before an absolute event identification is assigned.