Sidelobe Suppression for Likelihood Ratio-Based Seismic Deconvolution

In this work, we develop a novel abrupt-jump detection algorithm for seismic signal deconvolution. This new method significantly suppresses the effect of sidelobes. It begins with transforming a seismic trace to a series of innovations with a Kalman filter and then estimates the likelihood ratios of the reflectivity impulses from the innovations. Second, it modifies the likelihood ratios by imposing additional punishment on its asymmetry. Therefore, the likelihood ratios induced by the sidelobes are highly suppressed. Hence the reflectivity impulses recovered from the modified likelihood ratios are less affected by the sidelobes, leading to significantly enhanced resolution. The efficacy of the proposed method is numerically validated on a synthetic and a field dataset. The experimental results show that the proposed scheme is efficient and practical in enhancing the signal quality of seismograms than the original likelihood-ratio-based abrupt-jump detection.