Discontinuity enhancement based on time-variant seismic image deblurring

• We propose a time-variant PSF estimation algorithm for seismic image deblurring.
• We construct a sequence of time windows with adaptive variable length.
• The whitening factors of the 2D Wiener filter are adaptively selected.
• The coherence cube image with high resolution is obtained using this algorithm.
• This algorithm is robust and can attenuate the high frequency noise effectively.

Post-stack 3D seismic data is spatially blurred by the effects of migration operators with limited aperture widths, which is not conducive to discontinuity (such as fault, channel, etc.) detection. By approximating the migration blur with a time-invariant point spread function (TIPSF), seismic image deblurring methods have been used to obtain data with enhanced discontinuity. Better discontinuity detection results can be achieved on the deblurred data than on the original data. Since the migration blurs are always time-dependent, a time-variant PSF (TVPSF) estimation method is proposed in this paper to approximate these blurs. In our method, initial PSFs corresponding to each horizontal time slice (HTS) from a 3D seismic data are first obtained. Then, PSFs corresponding to adjacent time slices are divided into the same categories based on their similarities. With average PSFs calculated in each category, linear interpolation is performed to estimate PSFs for the whole data set. Finally, we perform seismic image deblurring HTS by HTS with these estimated PSFs. To suit different signal-to-noise ratios (SNR) in these HTSs of the 3D seismic data, the whitening factor of the Wiener filter for each HTS is adjusted adaptively. Using field dataset examples, we demonstrate that the performance of our proposed TVPSF method outperforms the TIPSF method.