Seismic phase detection and discrimination using adaptive filter coefficients

The authors present an approach to the detection and discrimination of seismic phases using multichannel data. When a seismic event (such as an earthquake or man-made explosion) occurs, the signal propagating from the source has several distinct phases which have varying propagation velocities and frequency content. By detecting and correctly identifying these phases one can estimate the distance between source and receiver. The detection algorithm considered consists of an adaptive correlation enhancer and a sliding window detector that make the detector insensitive to changes in the background noise level. The discrimination algorithm uses the adaptive correlation enhancer's weights as input discriminants to a neural net.<<ETX>>