Coherence-guided Interpolation improves the sampling of 3D volumes by inserting new shots and receivers along or across existing shot and receiver lines.

The algorithm, which is typically run after NMO and static corrections, may be applied in the shot and/or receiver domain, and it operates on neighboring traces within the current gather.

In order to synthesize a new trace, the algorithm begins by defining overlapping time gates within an analysis group of traces which is centered on the output location. For each temporal gate, dominant dip directions are identified via trial dip scan across the spatial window spanning the analysis group. Next, an interpolated data segment is constructed by local slant stack along these dominant directions followed by coherence-weighted combination of the individual slant stack sums. The process is repeated for all time gates, and the final output trace is created by merging the interpolated data segments from each temporal gate, where tapering is used to handle the overlap zones. Note that DSINTERP requires local input data support and it cannot work in the presence of large gaps; therefore it is often best used in conjunction with SFINTERP.