Wavelet-Based Higher Order Correlative Stacking for Seismic Data Denoising in the Curvelet Domain

To whiten random noise and identify coherent noise while preserving the features of seismic events, a hybrid denoising scheme of wavelet-based higher order correlative stacking (HOCS) in the curvelet domain is proposed. The proposed algorithm uses HOCS to isolate the coefficients of seismic events in the curvelet domain. It then removes the noises and recovers signals recorded in noisy environment, without the need to choose an arbitrary threshold; the HOCS method selects a threshold automatically in the curvelet domain. Therefore, with the HOCS, it is possible to capture the features of useful signals with good correlations at all scales and all angles, then to remove the features of coherent noise with disordered correlations. Using interpretive seismic records of karst cavities and hidden sinkhole detections after artificial backfill, we show that the proposed scheme improves noisy seismic data significantly with respect to both signal-to-noise ratio and fidelity. To demonstrate the advantages of this hybrid denoising scheme, a comparison of the performances between different individual denoising methods is investigated for complex seismic records contaminated with different types of noise. Numerical case studies and three field data examples validate the effectiveness of the hybrid denoising scheme proposed in this paper.