Interpretative noise attenuation in the curvelet domain

Abstract

We present two real data examples of coherent and random noise attenuation in the curvelet domain. In the first example, the unwanted energy corresponds to cultural noise and surface waves. The dataset is processed in a series of passes on common‐shot, azimuth, in‐line, cross‐line, and midpoint gathers. In the second example, we process a migrated volume in a series of passes on in‐line, cross‐line and depth slices to remove some ambiguities near the base of salt due to linear, steeply dipping noise. In both examples, we use a simple weighting scheme in the curvelet domain. The insights of an interpreter were key to include geological knowledge and leverage at best the flexibility of the curvelet transform in the noise attenuation process. For this reason we consider our results to go beyond what has been published to date about curvelet noise attenuation of seismic data.