Multiple suppression provides a good context for interpolating seismic data, because the problems related to sampling are well known and the data usually have a very regular acquisition geometry and high signal-to-noise ratio. While cable feathering makes the data irregular in physical (x,y) coordinates, it tends to be perfectly regular in (s,g).
The shape of a streamer, while not straight, tends to change slowly from one shot to the next, since the currents tend to vary with tidal cycle at the fastest. With modern GPS and navigation controls, the source locations generally follow lines which vary quite slowly in relation to the shot spacing. This regularity makes marine data suitable for a wide range of algorithms. In particular, filtering is sensible because the shot and channel regularity and the slowly changing shapes of streamer cables make it easy to put the data on a meaningful grid. Figure recs2 shows the layout of receivers during numerous shots. The figure is vertically exaggerated to show the irregularities in the geometry. The streamer is not straight, so the data are not exactly regular in terms of shot and receiver surface locations. However, the data are regular in shot and receiver number coordinates, and because the streamer wiggles in a smooth fashion, they should also be predictable.
Land data often do not have this property. Receiver and source positions can make sharp turns, particularly when acquisition is constrained by permits and terrain. Adjacent receiver lines can have different numbers of receivers. The data are typically much noisier than marine data. Nonetheless, land data presents an interesting interpolation challenge because it is more difficult and expensive to acquire than marine data.