Bedforms in a laboratory wave flume: An evaluation of predictive models for bedform wavelengths

Recent bedform dimensions measured in shallow waters in the nearshore zone (VINCENT and OSBORNE, 1993; OSBORNE and VINCENT, 1993; MARSH, 1996) compare poorly with published predictive models for bedform dimensions. A series o f experiments were conducted in a large flume with a computer-controlled wa ve generator and a sand bed, using waves of various amplitudes and characte ristics including waves from a field site and monochromatic waves. The ripp le wavelengths were then compared to the wavelengths predicted by the model s of NIELSEN (1981), GRANT and MADSEN (1982), MOGRIDGE et al. (1994) and WI BERG and HARRIS (1994), and to the semi-quantitative model of CLIFTON (1976 ), Under spectral waves from the field site the mean ripple wavelengths are anorbital remaining constant (within the scatter of the measurements) and showing none of the trends predicted by the models but falling between the dimensions predicted by NIELSEN (1981) for 'laboratory' and 'field' waves. Under monochromatic waves the ripples scaled with the wave orbital amplitud e (lambda approximate to 0.4A(0)) and were much closer to the model predict ions. It is suggested that it is rather difficult to change the wavelength of rip ples once they have formed. Field waves generally have a broad spectrum of frequencies (and hence of orbital excursions) so there is no length scale o f sufficient dominance to force the bed to reform. With regular waves every orbital excursion is the same and the bed rapidly scales to this length. O ur data suggest that bed form dimensions in an event may therefore be deter mined by the first waves capable of imposing their length scale on the bed, or by bed forms from an earlier.