DETECTION OF WEAK SEISMIC SIGNALS USING STATIONARITY OF NOISE AND COHERENCY OF SIGNALS AT A SEISMIC ARRAY

A basic problem in seismic data analysis is the recognition of weak signals in the presence of ambient noise. Although the commonly used array processing method, which involves beam formation from the array data, provides a reasonable enhancement of the signal to noise ratio (SNR), it does not exploit the properties pertaining to stationarity of the background seismic noise. Prediction error filters, on the other hand, because of their ability to accentuate small non-stationary changes, can be effectively employed to detect weak signals in a stationary noise background. In this paper, it is shown how these two signal detection methods can be used in cascade to achieve an overall high SNR. Through the analysis of some artificial and real seismograms it is demonstrated that an average SNR improvement of up to a factor of ten is possible by the combined method of signal detection, using data from the Gauribidanur array.