Treatment of coherent linear noise by conventional processing
We shall apply a processing sequence to a marine 2-D data set that includes very basic steps without any special attempt to attenuate coherent linear noise. The objective is to examine the treatment of such noise by the three principal processes — deconvolution, stacking and migration.
Figure 6.0-10 shows selected raw shot records from the marine line under consideration. Note the presence of guided waves in all the records in the form of a prominent dispersive wave package. The dispersive nature of guided waves is pronounced especially in shallow water. Because of their high amplitudes, guided waves dominate recorded marine data before the correction for geometric spreading. Since they travel in the horizontal direction within the water layer, guided waves do not contribute to the useful reflection energy. Therefore, these waves are often muted in shallow records as shown in Figure 6.0-11. Unfortunately, some reflection energy at far offsets is inadvertently removed as a result of muting the guided waves.
Following t2-scaling of amplitudes to compensate for geometric spreading, we note the enhancement of coherent noise at late times. Note in Figure 6.0-11, the records at shot points 300 and 400 contain linear noise below 2 s, and all records except at shot point 200 contain coherent noise with a curvature below 3 s, all associated with side scatterers. Additionally, observe the low-frequency cable noise with large stepout especially at near offsets below 3 s on the records at shot points 300, 400, 600, and 700.
Deconvolution flattens the spectrum and as a result enhances the low-frequency cable noise as seen in Figure 6.0-12. After the application of a wide bandpass filter, very low-frequency and very high-frequency noise components are removed. Nevertheless, the side-scattered energy with varying moveout still remains in the shot records (Figure 6.0-13).
Figure 6.0-12 The same shot records as in Figure 6.0-11 after spiking deconvolution. Shown on top of each record is the average amplitude spectrum.
When data are sorted to CMP gathers, the linear nature of the coherent noise associated with side scatterers disappears (Figure 6.0-14). On the other hand, side-scatterer noise with curvature behaves like events with nonhyperbolic moveout. Following the normal-moveout correction and stacking, the side-scatterer energy reappears as in Figure 6.0-15. Note the dipping linear noise along the steep flanks of the diffractions associated with the side scatterers in the water bottom. The steeply dipping linear noise at water velocity should not be confused with the diffractions of the flanks of the salt diapirs at higher velocities.
Coherent linear noise associated with side scatterers are attenuated largely by dip-moveout correction (dip-moveout correction in practice). Compare the stacked section in Figure 6.0-16 with that in Figure 6.0-15, and note that DMO correction has enhanced the diffractions associated with the salt flanks while it has attenuated the linear noise associated with the side scatterers. Any remaining side-scatterer related noise at late times is overmigrated as a result of the higher primary velocities (Figure 6.0-17).
Figure 6.0-17 Migration of the DMO stack shown in Figure 6.0-16.
- Coherent linear noise
- Reverberations and multiples
- Treatment of reverberations and multiples by conventional processing
- Spatially random noise