Seismic Wavelet Estimation through Phase Retrieval

Summary A reflection seismogram, after some specified processing steps, can be regarded as a convolution of the source wavelet with the reflectivity series and added noise. A wavelet is defined by its amplitude and phase spectrum. Determining an appropriate phase in the time domain has always been a problem. In this paper, wavelet estimation through phase retrieval is proposed in two sequential steps. In the first, the reflectivity model is obtained only by the amplitude spectrum of the observed data. In the second, the seismic source wavelet is estimated using the obtained reflectivity. We deal with an ill-posed inverse problem which has to be solved through regularization. The proposed phase retrieval algorithm has no limitations for the phase of the estimated wavelet. The mentioned procedure is examined on both synthetic and real seismic sections to obtain reflectivity model and the source wavelet. Eventually the good performance of the proposed algorithm is demonstrated.