Curvelet-Based Noise Attenuation in Prestack Seismic Data

The seismic signal components depicting the underlying geology are usually corrupted by the noise components in field data. To obtain the high quality seismic records, geophysicists have to attenuate noise in prestack data. But the corruption of seismic signal components by the noise attenuation processing is unavoidable. Thus attenuating the undesirable noise components with minimal damage to the geologic signal is very important in seismic data processing. The curvelet transform developed recently is greatly suitable for seismic data processing, because the curvelets are little plane waves with enough spatial and frequency localization, complete with optimal sparsity. In this paper, we explore an effective noise attenuation approach based on curvelet transform. Applying this method to a synthetic data set and to a field data set shows that the curvelet-based approach outperforms the traditional method in noise attenuation with minimal impact on the desirable signal components.