
Vol. 58 No. 5
May 2006
Three-dimensional Surface-Related Multiple Elimination (SRME) has proved a powerful method to address the seismic problem of either reconstructing missing data or reconstructing missing multiple contributions. The specific problems of diffracted multiple energy and out-of-plane (shifted-apex) 3D multiple energy have responded well. Typically, these multiples are caused by rugose or structured seafloors and have proved to be particularly difficult to remove.
Cie. Générale de Géophysique (CGG) has applied the 3D SRME method on projects worldwide with significant benefits. With the approach being based on full modeling of the multiple wavefield, this method is also independent of acquisition geometry, whether 3D streamer with conventional or sparse separation, multi-azimuth, or ocean-bottom cable.
The problem of multiples is approached by use of model-based prediction techniques rather than a data-driven method. The result is a multiple-attenuation technique that fully models the multiple arrivals. The strength of this method lies in its ability to cope with any kind of acquisition geometry.
Data-driven SRME techniques have recently represented attractive solutions to the multiple problem because they do not require a priori knowledge of the subsurface (reflectivity, structures, and velocities). However, these methods require one shot location for each receiver position, and this is not the case for most 3D acquisition geometries. The most common solution requires interpolation of the input data, thus creating the missing streamers and shot lines for the required prestack convolutional process. The strengths of this approach are that it is fully data driven and models all the multiples at once. However, the surface-consistent convolutions require sampling of identical source/receiver sets; in 3D, this requires significant acquisition and/or data regularization and extrapolation effort (typically, closer sail lines).
The modeling of the primaries is obtained by prestack demigration of the migrated section (either time or depth). The ringing of the primaries and multiples through the target volume is modeled using the one-way wave equation; as there is a propagated wavefield all along the acquisition surface, the method addresses problems of cable feathering.
Because the shot gathers are modeled separately, the method is suited for parallel implementation. It is currently optimized to run on large personal-computer clusters. For the ocean-bottom-cable environment, each receiver gather is processed separately, and there is no need for sail-line interpolation (or receiver-gather interpolation).
Deepwater offshore Brazil is an area that demonstrates many of the problems that until now have defeated multiple-attenuation techniques. Seabed channel development on the slope produces an environment that will generate complex diffracted multiple energy. Moveout discrimination and 2D SRME methods fail to properly identify or model such a complex multiple wavefield. The use of 3D SRME by wavefield modeling has proved particularly effective for modeling such complex diffracted and shifted-apex multiple energy. The nonhyperbolic form of the multiple moveout is clearly evident on the gathers (Fig. 1), and while 2D SRME attenuates some of the energy, significant residual energy remains to contaminate the stack. Application of the 3D-wavefield-modeling technique produces a significantly improved result. Accurate modeling of the complex multiple wavefield is particularly evident in the comparison of the modeled wavefields resulting from the 2D and the 3D methods (gathers on right of Fig. 1). There are significant differences between the stacks of the subtraction results (Fig. 2). Many residuals from the multiples are left in the 2D result.

Fig. 1—Representative common-midpoint (CMP) gather from offshore Brazil
showing main area of multiple interference.

Fig. 2—Zoomed stack section showing main area of multiple
interference.
In addition to Brazil and the Gulf of Mexico, the 3D SRME modeling technique has been used successfully in Canada, Europe, Africa, India, and the Far East.