** Next:** Robust inversion using the
** Up:** Multidimensional seismic noise attenuation
** Previous:** Multiple attenuation: Theory and

Chapter introduces a Gulf of Mexico 3-D dataset donated by
the Compagnie Générale de Géophysique (CGG), where multiples are
attenuated with the pattern-based technique of Chapter
. With 3-D data, obtaining an accurate model of the
multiples becomes the biggest challenge because the acquisition
geometry is not regular. Therefore, the method of
Verschuur et al. (1992) cannot be directly applied on the data
without interpolation. Worst, the data need to be interpolated and
extrapolated on a regular and fine mesh in both inline and crossline
directions. With 3-D data, the cost of such processes can be
prohibitive and alternatives are needed. In this Chapter, I
propose doing the interpolation in the inline direction only
and model the multiples in a 2-D sense, one streamer at a time.
The shots are interpolated with a radon-based technique on
common-mid point (CMP) gathers. I show that the resulting noise model from the 2-D
prediction, although not perfect, still encapsulates the main features
of the true multiples present in the data. Because the pattern-based
method is robust to modeling uncertainties, the primaries
are well recovered regardless of the approximations made during the
model building.

** Next:** Robust inversion using the
** Up:** Multidimensional seismic noise attenuation
** Previous:** Multiple attenuation: Theory and
Stanford Exploration Project

5/5/2005