Soils have strange acoustic behavior. Their seismic velocities are usually less than or equal to the speed of sound in water (1500 meters/sec). It is not uncommon for soil velocity to be five times slower than the speed of sound in water, or as slow as the speed of sound in air (300 meters/sec). Where practical, seismic sources are buried under the weathered zone, but the receivers are almost always on the surface. About the only time you may encounter buried receivers is in a marshy area. There field operations are so difficult that you will have many fewer receivers than normal.
A source of much difficulty is that soils are severely laterally inhomogeneous. It is not rare for two geophones separated by 10 meters to record quite different seismograms. In particular, the uphole transit time (the seismic travel time from the bottom of a shot hole to the surface near the top of the hole) can easily exhibit time anomalies of a full wavelength. All this despite a flat level surface. How can such severe, unpredictable, travel-time anomalies in the weathered zone be understood? By river meanders, tiny shallow gas pockets, pocketed carbonates, glacial tills, etc. All these irregularities can be found at depth too, but they are worse at the surface before saturation and the pressure of burial reduce the acoustic inhomogeneity.
The shallow marine case is somewhat better. Ample opportunities for lateral variations still exist--there are buried submarine channels as well as buried fossil river channels. But the dominant aspect of the shallow marine case becomes the resonance in the water layer. The power spectrum of the observed data will be controlled by this resonance.
Likewise, with land data, the power spectrum often varies rapidly from one recording station to the next. These changes in spectrum may be interpreted as changes in the multiple reflections which stem from changes in the effective depth or character of the weathered zone.