Publications

Land seismic noises generated by surface activity and sub-surface geology are generally much stronger than the primary energy and difficult to attenuate during seismic acquisition. Therefore, advanced seismic processing technologies are critical for reducing the noise, increasing the signal to noise ratio, and improving imaging quality to help geoscientists better interpret subsurface features. Noise cancelling without removing primary signal is a challenging task in seismic processing. In attenuating random and coherent noises, two state-of-the-art noise reduction technologies will be presented – Ground Roll Buster (GRB) and Curvelet Transform. GRB is a phase-matched filter that attenuates source generated surface-wave noise and requires no iterations to exactly match the dispersion relations of the noise wavetrains. It is more robust than other techniques such as FK filter. Curvelet Transform shows superiority compared to conventional techniques such as median filtering and f-x deconvolution filtering (Neelamani et al., 2008) in attenuating linear coherent and random noise. Curvelet is a non-adaptive method for multi-scale object representation. The fundamentals of the curvelet-based technique are already relatively popular in image processing and scientific computing. In the geophysics area, Herrmann et al. (2008), Neelamani et al. (2008) implemented curvelet-based strategies to a variety of seismic processing problems. Curvelet is able to separate noises from primary signal in terms of at least one of the following: dip, frequency, magnitude and location. In other words, the signal and noise map into different sets of coefficients after curvelet transformation. We implemented GRB and Curvelet Transform technique to denoise noisy Cepu 3D Seismic Data. The techniques effectively removes random, linear noises and even steeply dipping noises such as remnant high frequency scatter and nature noises. Keywords: Curvelet, Denoise, Random Noise, Linear Noise, Steep Dip Noise.