Offset VSP data is often undersampled because the sources and/or receivers are not properly placed to provide adequate reflection coverage of the target. The quality of an offset VSP seismic image is a direct function of the source locations, receiver locations, target depth, velocity function at depth, and the near-surface velocity. Of these factors only the source and receiver locations are under the survey planner's control. Unfortunately there is no simple rule of thumb to follow in planning the optimal source and receiver placement for an offset VSP.

We have found that the most efficient and lowest-risk way to plan an offset VSP survey is to begin with a 2D velocity model of the survey area, like the model in Figure 1, and compute finite difference shot records with very dense source and receiver coverage. The coverage is much more dense than would be practical in a real survey.
Figure 1. Velocity Model
We then process subsets of the modeled sources and receivers through migration to find a set of economically feasible data acqusition parameters that meet the objectives of the survey. We work with the client and use a set of efficeint, reliable, and reproducible steps to determine the best set of data acquisition parameters. We consider important constraints such as the data acquisition tools that will be available for the survey and the amount of rig time allocated for the survey. We provide a complete report detailing data acquisition recommendations for source locations, receiver locations, and receiver spacing.

Why Finite Difference Modeling?

We use finite difference methods for modeling rather than ray tracing primarly because finite difference techniques properly model near-surface reverberations. Near-surface reverberations often dominate offset VSP shot records to the extent that some data cannot be used. Most ray tracing methods cannot produce reverberations, yet reverberations from P, S, and converted waves can be the factor that controls the maximum shot offset used in a VSP survey. Figure 2 shows modeled shot records from zero-offset, mid-range offset, and a far offset computed from the velocity field in Figure 1.

> Return to Top
Figure 2. Shot records: far, mid, and near offsets.
Reflections in the far offset shot (left) and mid-range shot (center) have strong interference from near-surface reverberations. Some data recorded in the upper 1/3 of geophones probably could not be used for processing.

A Modeling/Processing Example

In addtion to creating accurate estimates of reverberation noise, finite difference modeling, combined with migration, gives a more accurate prediction of imaging results than simple illumination modeling. Figure 3 shows the velocity model in Figure 1 overlayed by the image we obtained by migrating shots with a source spacing of 1000 ft and receivers from 100 ft depth to 5000 ft depth at a 50 ft interval.

> Return to Top
Figure 3. Illumination Model
The reflection image is dim in the area around the receiver well, particularly the interface sloping upward and to the left of the receiver well. By adding source points at an interval of 400 ft within 3000 ft of the receiver well the sloping area becomes much more clear (Figure 4). We also learned that the receiver depth range could be reduced to 500 through 3500 ft and still maintain the image quality.

> Return to Top
Figure 4. Illumination Model 2
This is an example of adaptive modeling that provides data acquisition parameters with the lowest technical risk for the lowest data acqusition cost. The outstanding computational infrastructure at STERLING allows us to quickly compute even high-frequency finite difference shot records to be sure of complete and optimal data acqusition recommendations.

> Return to Top