Data Description

The dataset used in this paper (provided by ExxonMobil) is designed as a test for interpolation algorithms used for 3D surface-related multiple elimination. The model is constant-velocity, with multiple reflectors and 500 point diffractors contained in a prism near the water bottom that is oriented perpendicular to the acquisition direction. A schematic of the model is shown in Figure .

 

geom Figure 1 The model for the synthetic data: a prism containting diffractors below the sea floor, with reflectors underneath.

The data coverage is that of a modern marine 3D survey, acquired in a racetrack fashion. There are 12 streamers spaced at 50m, with 400 receivers in each streamer spaced at 12.5m, with a near offset of 100m for a maximum in-line offset of 5100m. There are two sources, spaced at 25m in the cross-line that are fired in a flip-flop fashion so that a single shot is fired every 37.5m. The total cross-line aperture is 550m. There are 10 sail lines with a spacing of 200m. This is very dense for a 3D survey, but the data density requirements of performing 3D SRME are much greater.

Ideal geometry for 3D SRME would include a source at every point where there is a receiver. In this case, a bin size of 12.5m would be ideal, meaning that in terms of interpolation, the following increase in data density would have to occur:

• Receiver spacing (in-line): none,
• Source spacing (in-line): factor of 3,
• Receiver spacing (cross-line): factor of 4,
• Source spacing (cross-line): factor of 8 or 16.

As can be seen, the combined ratio of needed data to sampled data is 96, meaning that the current sampling of the desired output cube is incredibly sparse. This is neglecting the extrapolation that is also required for receivers in the in-line direction to zero offset, which is minor, as well as for the extrapolation of receivers which is required in the cross-line direction, which would add another factor of 4 to this problem.

Most interpolation algorithms are tested by a factor of 2 in the in-line direction, meaning the infill of flip-flop shooting. This will be tested first, and afterward the infill of receiver cables.