| Abstract: | Temporal variations in bedload transport rates that occur at a variety of timescales, even under steady flow conditions, are accepted as an inherent component of the bedload transport process. Rarely, however, has the cause of such variations been explained clearly. We consider three data sets, obtained from laboratory experiments, that refer to measurements of bedload transport made with continuously recording bedload traps. Each data set is characterized by a predominant low-frequency oscillation, on which additional higher-frequency oscillations generally are superimposed. The period of these oscillations, as isolated through the use of spectral analysts, ranged between 0·47 and 168 minutes, and was associated unequivocally with the migration of bedforms such as ripples, dunes, and bars. The extent to which such oscillatory behaviour may be recognized in a data set depends on the duration of sampling and the length of the sampling time, with respect to the period of a given bedform. Several theoretical probability distribution functions have been developed to describe the frequency distributions of (relative) bedload transport rates that are associated with the migration of bedforms (Einstein, 1937b; Hamamori, 1962; Carey and Hubbell, 1986). These distribution functions were derived without reference to a sampling interval. We present a modification of Hamamori's (1962) probability distribution function, generated by Monte Carlo simulation, which permits one to specify the sampling interval, in relation to the length of a bedform. Comparisons between the simulated and observed frequency distributions, that were undertaken on the basis of the data described herein, are good (significant at the 90 per cent confidence level). Finally, the implications that temporal variability, which is associated with the migration of bedforms, have for the accurate determination of bedload transport rates are considered. |
