2-D or 3-D

For marine data, there is very little difference between interpolating shots in 2-D survey and interpolating shots in a 3-D survey, except that 3-D data provides some interesting choices. We can treat a single source and a single streamer as a 2-D survey (giving us several 3-D input cubes), or separate the sources but leave the streamers together (4-D input), or just use the whole 5-D input. Results get better with more dimensions, because there are more directions for events to be predictable in; but they only get marginally better when we add the crossline directions in marine data, because there are only a handful of crossline points. The cost is large because a few points worth of zero padding are necessary. Padding the inline offset axis by a few points is a small increase in the data volume, but padding the crossline offset axis by a few points may double the data volume. Newer boats tow more streamers, so it may be worth using the extra dimensions on newer data.