The Azimuthal Manifestation of Ground Roll in 3D Surveys.
For some time now there has been interest around the world for extracting subsurface azimuthal information from seismic data. Indeed azimuth-dependent stacks probably do have useful directional information contained in them. However, that information is almost certain to be weak embedded in noise - especially since the fold of such stacks is typically low. Therefore if we intend to assign geologic significance to the differences we see in such stacks, we must first be sure we comprehend the azimuthal influence of the noise. This study dwells on one particular case, the influence of ground roll in azimuth-dependent stacks from wide-swath 3D surveys.
The approach taken here is to model the propagation, acquisition and processing of various types of ground roll. Figure 1 shows a map view of the active spread that is considered. There are 12 receiver lines each with 48 channels on either side of the shot. (We will see later, though, that not all of these channels are used in the analysis). The inline receiver group spacing is 50 m and the crossline spacing is 200 m. The maximum source-receiver offset in the crossline direction is 1175 m.
36 geophones comprise each receiver array while 5 vibrator locations comprise each source array. The geophone array is areal, but the shot array is linear. Figure 2 shows the layouts of the patterns. These patterns were selected for the modeling experiments because they would be desirable from the standpoint of operational efficiency in an actual production survey.
The shot progression used is the popular double zigzag geometry. The sampling is such that there are 200 shots per square kilometer.
Three ground roll scenarios are considered. They are the short wavelength case (9-32 m), the medium wavelength case (54-82 m), and the long wavelength case (83-291 m). In each case, the modeled ground roll is dispersive. It is computed from the stationary phase solution of an integral presented by Aid and Richards (1980). Parameters for the different scenarios were obtained from the literature, from noise tests, and when necessary, from production shot records.
The first experiment to be examined here is the short wavelength case. This type of surface wave occurs when there is a thick covering of slow velocity material, like sand, at the surface.
The dimensions of our arrays are longer than the wavelengths of the ground roll, so we would expect the arrays to be fairly powerful at suppressing this noise. The inline component of the combined source and receiver array response is shown in figure 3. The crossline component is shown in figure 4. The portions of the diagrams that are relevant to the short wavelength case are shaded. We see that the arrays are indeed powerful for this type of ground roll, although the crossline suppression is not as great as the inline suppression. This is because the source array is linear in the in line direction.
Figure 5 shows portions of modeled shot records. These traces come from one side of the near cable. The maximum offset displayed is 1175 m. Recall that our spread geometry does not possess crossline offsets in excess of that. So in order to maintain parity among all of the azimuthal partitions in our analysis, the offsets are truncated to 1175 m - regardless of azimuth.