Microseismic wavefield separation using multi-channel chirplet decomposition
Résumé
Small earthquakes and acoustic emissions making up microseismic events are characterized by interfering complex wavefields. Many authors have discussed the problem of achieving satisfactorily wavefield separation process mainly by considering propagation only as elastic and specular. Because microseismic wavefield is based on specular and not specular propagation with dispersion and scattering phenomena, conventional wavefield separation methods are no longer efficient. In this paper, a new type of wavefield separation approach is proposed allowing each event to vary independently in amplitude, phase, time-shift and waveform while it propagates across the microseismic array. This method is based on a multi-channel generalized wavelet transform characterized by the fact that the microseismic wavefield is decomposed into a weighted sum of atoms known as chirplets. A dictionary consisted of collections of parameterized chirplets allows a powerful adaptability to quantify the morphological attributes of the recorded microseismic events. The decomposition is performed through an algorithm that combines a conventional Matching Pursuit algorithm with a combinatorial algorithm providing the atomic decomposition in an optimal way.