CURVELET DOMAIN ADAPTIVE LEAST-SQUARES SUBTRACTION OF INTERNAL MULTIPLES

摘要

The elimination of internal multiples has always been a challenge in seismic processing. Compared to surface-related multiples, the amplitudes of internal multiples are more complicated and have a wider range due to different geological interfaces, and these complexities make the conventional prediction-subtraction algorithm not easy to be implemented. In this paper, a new method has been proposed to reduce the negative influence of energy diversity in internal multiples subtraction based on curvelet transform. First, we apply multi-resolution and multi-directional analysis to the seismic record with internal multiples, to map seismic events with different spectral and directional features into different curvelet domains. Then, internal multiples can be estimated by minimizing the misfit between the curvelet coefficients of the real seismic data and components of the predicted multiples under least-squares sense in curvelet sub-domains. A simple experimental data with three crossed events and a synthetic seismic record with complex internal multiples are used to validate the effectiveness of the proposed method. Results indicate that our approach is effective in suppressing internal multiples, preserving geological signals and avoiding distortion of primary events even when intersection or coincidence occurs.