Velocity discrimination between primaries and multiples
|Series||Investigations in Geophysics|
|Store||SEG Online Store|
The CMP gathers in Figure 6.1-8a clearly illustrate the moveout difference between primaries and multiples. A primary p typically has less moveout than a multiple m. From the velocity spectrum in Figure 6.1-8b, note the difference between the velocity trends associated with primaries V P and multiples V M1 and V M2. The V M1 and V M2 velocity functions represent the water-bottom and peg-leg multiples, respectively. If NMO correction is applied using the primary velocities, as is normally done to generate final stacks, then the primaries are aligned while the multiples are undercorrected (Figure 6.1-8c). This suggests that CMP stacking itself is a viable method of multiple attenuation. The CMP stack derived from the gathers in Figure 6.1-8c is shown in Figure 6.1-8d.
The synthetic CMP gather in Figure 6.1-9c contains five primaries, including the water-bottom reflection W and the multiples associated with it. The velocity spectrum shows a significant separation between the velocity functions for multiples V M and primaries V P. Stacking with the primary velocity function should, to a large extent, discriminate against the multiples and result in a section that contains essentially the primary energy as shown in Figure 6.1-10. The stack trace in Figure 6.1-10c is repeated to better examine the relative amplitudes of the primaries and the multiples.
Stacking far offsets works to suppress multiples. However, stacking near offsets works against multiple attenuation, since the moveout difference between primaries and multiples is negligibly small on those offsets as in Figure 6.1-8c. The simplest way around this problem is to apply an inside mute to the CMP gathers before stacking. Another problem then emerges — the outside mute. The severity of this mute governs the amount of far-offset data left at early times for velocity discrimination (Figure 6.1-8c). If there is a severe multiple problem, an effort must be made to preserve the maximum amount of far-offset data associated with target events. The stacked section of Figure 6.1-8d with inside mute applied is shown in Figure 6.1-11a. When compared with Figure 6.1-8d, note that the deeper peg-leg multiple below 4 s has been further attenuated by inside trace muting. The difference between the conventional CMP stack (Figure 6.1-8d) and the inside mute stack (Figure 6.1-11a) shown in Figure 6.1-11b indicates the amount of energy, mostly multiples, that was removed by the inside mute.
Figure 6.1-8 (a) Three CMP gathers with strong multiples; (b) velocity analysis at CMP 186, where V P = primary velocity trend, V M1 = slow (water-bottom) multiples, and V M2 = fast (peg-leg) multiples. (V B is the velocity function used in generating Figure 6.2-15a.) For reference, the CMP gather is displayed next to the velocity spectrum. (c) The same CMP gathers as in (a) after NMO correction using the primary velocities. (d) CMP stack using the gathers as in (c). (Data courtesy Petro-Canada Resources.)
Figure 6.1-9 Synthetic CMP gathers containing (a) primaries, (b) water-bottom multiples, (c) superposition of (a) and (b). (d) The velocity spectrum derived from (c). Here, W = water-bottom primary, V M = velocity function for multiples, V P = velocity function for primaries, V B = a velocity function between V M and V P used in generating Figure 6.2-12b.
Figure 6.1-11 (a) The CMP stack derived from the CMP gathers in Figure 6.1-8c with inside mute applied. The inside mute pattern can be recognized on the left edge of the section. Compare this stack with that shown in Figure 6.1-8d. (b) The difference between the conventional CMP stack (Figure 6.1-8d) and the inside mute stack (a). (Data courtesy Petro-Canada Resources.)
A variation of conventional muting, such as optimum-weighted stacking can produce better results. In such a scheme, weights between 0 and 1 are assigned to each offset during stacking. The smaller weights are normally assigned to the near offsets.
In summary, because there is relatively less move-out differential between the primaries and multiples in the near-offset range, the inside mute (or some kind of weighted stacking) helps suppress multiples. Hence, it may help to cascade any one of the multiple attenuation techniques described in this chapter with inside mute during stacking.