Data Background

The data were recorded by WesternGeco in 1999, as part of a multi-client (non-exclusive) survey which acquired a coarse grid of ten 2-D lines. A single 2-D line, consisting of 2943 shots covering roughly 120 km, was donated to SEP. The survey used the following acquisition parameters:

• Shot interval: 25 m
• Group interval: 12.5 m
• Number of channels: 912
• Active array length: 11.4 km
• Near offset: 185 m
• Record length: 9.8 sec at 4 msec

Of the 2943 total shots, we chose to limit our experimentation to the 960 from the center of the survey that were processed by WesternGeco using the Delft Surface Related Multiple Elimination (SRME) technique Verschuur et al. (1992). Before demultiple, the subset underwent the following preprocessing:

• Swell noise removal.
• Source signature deconvolution using nominal signature.
• Bandpass filtering and subsampling to 8 msec in time and 25 m group interval.
• Truncation to 9 km maximum offset.

After sorting into CMP gathers with 353 traces, the study area covered 26 km of midpoints. Cable feathering over the 120 km line was variable, but unfortunately quite strong over the subset that we studied. Differences between actual and nominal positions reach 2 km at 9 km offset. The SRME demultiple technique assumes nominal geometry, so the feathering issue in this case likely was a first order contributor to the incomplete suppression of multiples at far offsets Kostov et al. (2000). As we explain later, success of multiple suppression over the entire offset range is a crucial prerequisite to the successful identification of converted waves in both the prestack data and migrated images, since converted waves and multiples have similar apparent velocities. Other factors, such as 3-D effects and source/receiver directivity, also contribute to poorer performance of the SRME method.

WesternGeco also provided a depth velocity model, shown in Figure 1. The basalt layer exhibits a complicated vertical structure. A thin ``transition zone'' (perhaps interbedded basalt-sediment layers) with interpreted velocity of 3000 m/s gives way to more basalt-like velocities near 4500 m/s. The base of basalt is interpreted as a relatively flat reflector at roughly 3000 meters depth, underlain by a basement assigned a velocity of 5000 m/s.

 

vel-p
Figure 1 Depth migration velocity model provided by WesternGeco. We used the non-intuitive color scale shown because with the large velocity range (1400-5000m/s), it is impossible to resolve the layer boundaries on a simple grayscale intensity paper plot. Water depth ranges from 300-800 meters. The strongest reflection among the near surface sediments comes from the flat reflector at roughly 1000 meters depth. The transition zone consists of a package of reflectors from the basalt top. Base basalt interpretation is questionable, due to very low interpretable reflection energy.