Utilization of Multiples in Velocity Estimation and Migration
In the past the industry has made large investments in the removal of multiples from the seismic response ( ‘data linearization’ ), with the purpose to make the data suitable for our linear velocity estimation and linear migration algorithms. Now we are beginning to understand that multiples contain most valuable information that should not be removed (sometimes at high cost). Instead they should be utilized, making seismic migration a process that is fully consistent with the underlying physics. We need to realize that ‘multiple removal processes are the remains of traditional thinking’.

Therefore, today we see new developments at the horizon that make use of multiple scattering, such as Full Wavefield Migration (FWM) and Joint Migration Inversion (JMI). FWM and JMI are developed by the Delphi imaging team. Both algorithms use the fundamental property that, for a given acquisition geometry, multiples can reach parts of the subsurface where primaries cannot come. In FWM all types of multiples (both surface and internal) are included in the migration process, meaning that the wavefields are assumed to be nonlinear in reflectivity. In JMI surface and internal multiples are also used to improve the accuracy of the velocity estimation process.

In the past ten years I have worked on a theoretical framework for the utilization of surface and internal multiples in velocity estimation and migration. This full wavefield framework is an extension of the primary CFP concept. It has been summarized in an essay that consists of three parts (full wavefield modeling, joint modeling migration and joint migration inversion). In this trilogy I explain why the utilization of multiple scattering may cause a revolution in seismic imaging:

1. The illuminating wavefield is automatically enhanced by the downward- travelling multiple scattering (extra illumination for free). In the situation of incomplete data volumes and/or high structural complexity, the multiple scattering part may be the principal component in the illuminating wavefield.

2. The coda of the overburden is automatically removed from the response of the deeper boundaries (crosstalk removal for free). For deep reservoirs below a highly reflective overburden, this property is indispensable to reach quality images.

Both properties – enhanced illumination and clean responses – do not only allow a migration process that is consistent with the real physics in the subsurface, it also allows a velocity estimation process that is fuelled by the information in the multiple scattering events. I propose to make full wavefield velocity estimation an integral part of the full wavefield migration algorithm. This integration step will make seismic imaging less subjective; it will also give a significant contribution to the productivity in the seismic industry.

Delft, September 2014, A. J. (Guus) Berkhout, Delft University of Technology, The Netherlands