Wavelet packet polarization method for an automatic detection of P and S wave arrivals

A new method is proposed for detecting P and S wave arrivals in the problem of passive seismic monitoring. This method is based on the wavelet packet decomposition of three-component seismic traces, their automatic quality control, and the calculation of the principal components of the traces in overlapping packet frequency bands and in scale-dependent moving time windows. The wave arrivals times are estimated by means of a robust iterative procedure of adjusting hyperbolic traveltime curves to the initial estimates of arrival times determined from the maximum values of the multilevel measure of the nonstationarity of each trace. The method has been developed for applications in the cases of a large number of bad single-component traces and a high noise level. To illustrate its application, the method is applied to the data set of the Cotton Valley experiment on the seismic monitoring of hydraulic fractures in a fractured gas reservoir.