Separating noise from signal is an old problem in geophysics. Most geophysical work involves some separation of the desired information from the information that is not of interest, that is, noise. Many seismic data, especially those acquired on land, are seriously contaminated with noise that impedes interpretation and interferes with further processing and analysis.
Noise may be either random or coherent. Random noise, as it is considered here, is a noise that has no predictability from one sample to the next. Seismic data often has a background of this random noise. This random noise is recognized by its dissimilarity from trace to trace. Seismic signals, on the other hand, are recognized by their lateral continuity, and this continuity is used to distinguish events of interest from the background random noise. Much of this continuity results from the sedimentary character of the data being considered. Coherent noise may be considered as undesired signal. Examples of coherent noise are ground roll, near-surface scatterers, refracted arrivals, and out-of-plane reflectors. While coherent noise may often be signal in some other context, these noises interfere with the primary use of the data. Eliminating background noise from a seismic section makes reflections significantly easier to recognize, especially when noise is strong.