Land-based reflection data sets always include many types of coherent noise. This article deals specifically with the ubiquitous problem of filtering noise known as ground roll. Ground roll is a particular type of Rayleigh wave and has high amplitude, low frequency, and low velocity. Ground roll is also dispersive and overwhelms the desired reflection signals unless special steps are taken in data acquisition and processing to get rid of it. Because of the dispersive character of this noise, shallow reflections at short offsets and deep reflections at long offsets are masked in the space-time domain and thus their waveforms are distorted. On the other hand, frequencies of ground roll and reflections overlap in the spectral domain (frequency or f-k). Figure 1 illustrates spectra of signal with ground-roll noise (black), signal (green), and ground-roll noise (red). Their frequencies clearly overlap in the low-frequency part of the spectrum.