HOW DO GRAVEL-BED RIVERS BRAID

Sedimentary processes and bed forms leading to the onset of braiding were observed in small-scale hydraulic models of grave-bed streams. The laboratory streams had a variety of combinations of (constant) discharge and slope but identical bed-material particle-size distributions. From initially straight channels, braiding occurred by four different processes: deposition and accumulation of a central bar, chute cutoff of point bars, conversion of single transverse unit bars to mid-channel braid bars, and dissection of multiple bars. In these experiments the chute cutoff mechanism was the most common, but the predominant braiding mechanism depends upon sediment mobility (excess bed shear stress) and the bed-form regime. At very low excess bed shear stress the central bar process dominates, but at higher excess bed shear stress slip-face unit bars are more common, bed scour at confluences is more pronounced, and propogation of alternate convergence (scour) and divergence (deposition) is more likely; thus chute cutoffs and bar conversion dominate. The multiple bar mechanism is restricted to channels with very high width/depth ratio. All of these processes, along with avulsion, are significant for maintenance of an established braided channel. The direct physical sedimentary cause of primary braiding is essentially the same in all these processes: local aggradation (often by stalling of bed-load sheets) and loss of competence in a lateral flow expansion. The chute cutoff process occurs in a morphologically distinctive setting and may be aided by other factors, but it is usually triggered by the local thalweg shoaling that is the fundamental physical mechanism causing the onset of braiding by the other processes. Local short-term pulses in bed-load supply are often the trigger for the initiation and maintenance of braiding, regardless of the exact braiding process.