Bedload transport in braided gravel-bed rivers : a hydraulic model study

A 1:50 scale hydraulic model was designed based on Froude Number similarity using hydrological and sediment data from a braided gravel-bed river (the North Branch of the Ashburton River, Canterbury, New Zealand). An experimental programme was devised to investigate firstly, the relationships between discharge-slope product and bedload transport rate under steady and varying discharges; and secondly, the changes in bedload transport rate resulting from flow abstractions. Seven experiments using steady flow conditions were conducted (with six different discharge-slope combinations), and eleven experiments using unsteady flow conditions were also conducted (with five different discharge-slope combinations). The experiments were carried out in a 20 m x 3 m tilting flume equipped with a sediment feed device and an automated data acquisition and control system. In all experiments water at 30°c was used to reduce viscosity-related scale effects. Braided stream development was in some experiments found to be limited by the 3 m flume width; however, using available data from the narrower reaches of the North Branch of the Ashburton River, it was shown that good hydraulic similarity was achieved. Analyses of the experimental results revealed that bedload transport rates in braided channels are highly variable, with relative variability being inversely related to mean bedload transport rate. Variability was also found to be cyclic with short-term variations being caused by the migration of bedforms. Assessments of two bedload transport prediction rate equations were made: the Schoklitsch (1962) equation and the Bagnold (1980) equation. It was concluded that the Schoklitsch (1962) equation is unlikely to be useful for braided gravel-bed rivers, however the Bagnold (1980) equation was found to be very reliable, with the experimental data displaying a close adherence to the empirical 3/2 power-law dependence of bedload transport rate on excess stream power which is the basis of this equation. Average bedload transport rate was found to be dependent on channel form, although insufficient measurements were made to define the relationship. Bedload transport was found to be more efficient under steady flow than under unsteady flow, and it was postulated that this is caused by a tendency for channel form to evolve towards a condition which maximises bedload transport for the occurring flow. Using the results from the unsteady flow experiments, it was shown that stream power is a reliable basis for predicting the bedload transport capacity reductions caused by flow abstractions from braided rivers