While sediment transport occurs on a grain-by-grain basis, the knowledge of transport processes is only of use at the much larger scales of beach variability. The study of large scale sediment processes (roughly greater than an incident wavelength) is based on the assumption that the integrated transport can be understood with models that are, to a greater or lesser degree, robust to small scale details. As with continental shelf processes, predictability has traditionally been better in the longshore direction (where at least the net direction is usually known) than in the cross-shore. Our focus will be on cross-shore models [see Komar, 1991 for a good review of longshore transport].
The bulk of research models for cross-shore transport have been based on the assumption of small amplitude bathymetric response. That is, the wave and current field over a pre-existing bathymetry force a sediment response whose pattern may be a recognizable morphology, for example crescentic or skewed sand bars. However, the wave field is considered unaltered by the generated bar. The fundamental concept involved is one of a template, provided by the wave field, towards which the beach morphology tends to progress.
Several prominent models exist for the generation of sand bar morphologies, perhaps the most apparent feature of nearshore bathymetry. King and Williams [1949] proposed that changes in the fluid field associated with the onset of breaking could cause bar generation at the break point, while Holman and Bowen [1982] (for example), showed how transport convergence at the nodes of an edge wave field could explain a variety of bar types.
The simplicity and apparent testability of these models has lead to a number of hypothesis-based field experiments, none of which has provided conclusive results (with regard to bar generation). Holman and Sallenger [1993], re-examined the objectives, achievements and shortcomings of the suite of experiments that have taken place at Duck, North Carolina from 1981 to 1990. They concluded that natural wave fields were never sufficiently narrow-banded (to provide a clearly dominant break point or standing wave node), that topographic response provides a strong feedback to the wave field (so is not small amplitude), and that natural topographies rarely have time to reach the modeled equilibrium with the wave field. In short, linear models are perhaps fatally simplistic.