Blind wavelet estimation using a zero-lag slice of the fourth-order statistics

This paper presents an iterative method for blind wavelet estimation using a zero-lag slice (ZLS) of the fourth-order cumulant of the seismic signal. The proposed method can estimate non-minimum phase wavelets since the higher order statistics carry the phase information. By just using the ZLS that can be estimated from the seismic data with high reliability, our method achieves good accuracy for wavelet estimation, even if only short seismic signals are available. Furthermore, the proposed method is not computationally intensive because it is performed either by applying simple 1D operations ( frequency domain) or by solving a linear system ( time domain) in each iteration with fast convergence, and only needs calculation of the ZLS once. Application results on synthetic and real field data sets show that our method behaves well in relatively low signal-to-noise ratios, and when the wavelet length is overspecified.