Multiple suppression using prediction-error filter

REORDERING THE OFFSET ORDER AND SIMULATING A RANDOM PROCESS

Starting with a NMOed CMP gather, we can rearrange the offset order in a random manner. Figure 1 illustrates a synthetic NMOed CMP gather which contains one primary and one multiple event. After offset randomization, the primary event remains horizontal and the multiple event turns into ``random noise''.

syn-in
Figure 1. A synthetic NMOed CMP gather contains one primary event and one multiple event.

If we apply the offset randomization multiple times, all the different samples of this random process can generate a 3-D data cube. As shown in Figure 2, the primary energy is coherent along the cube, whereas the multiple part is incoherent. Therefore, not only can we estimate a 2-D PEF in the CMP gather, but also estimate a 3-D PEF in the data cube. It is very natural to expect that the 3-D result will be better than the 2-D, since the 3-D approach exploits the difference between the primary and multiple more thoroughly.

syn-cube
Figure 2. A NMOed CMP gather is transformed into a 3-D data cube by conducting the offset randomization for multiple times.

Multiple suppression using prediction-error filter