INNER CHANNEL DEVELOPMENT AND KNICKPOINT RETREAT IN AN ANABRANCHING RIVER BY POTHOLING

Knickpoint retreat through a 700 m wide anabranching channel segment along the Orange River, Republic of South Africa near Augrabies Falls is creating an inner channel, which is capturing flow from adjacent anabranches. The knickpoint is retreating entirely by potholing. Pothole coalescence and growth creates an inner channel that varies between one and six meters wide. Width increases to 20 m after the walls of large downstream potholes collapse because of large-scale block collapse. Pothole depths and radii were measured for 194 potholes in the knickpoint. The largest potholes are >6 m in depth (d) and >2 m in radius (r). Radii are related to d by a simple power law, r=kd^ε, where k and ε are regression coefficients. Observed values are 2.383 and 0.5700, respectively. ε > 0.5 implies that, on average, d increases faster than r in the potholes. Geometrical models of hemispherical and cylindrical pothole growth were constructed. Hemispherical potholes were found to have a complex geometrical relationship to adjacent translating bed surfaces. These potholes must enlarge substantially faster than the bed and interior erosive fluids must possess relatively high efficiencies relative to those on the adjacent bed or potholes will be removed or suppressed. Cylindrical potholes have relatively simple geometrical relationships to adjacent bed surfaces. They may persist by increasing their depth at a pace ≥ to the pace of bed translation. Calculating the bedrock volumes eroded from cylindrical pothole walls (V_w) versus floors (V_f) using the regression coefficients we find that more bedrock is eroded from pothole walls than floors during growth: V_w/V_f = 1.14. Similarly, for ε-values from three other locales: V_w/V_f = {1.14, 1.70}. Hence, more bedrock is eroded from cylindrical pothole walls during growth than from their floors. The statistical and geometrical analyses reveal that knickpoint retreat is dependent upon abrasion of pothole walls by suspended sediment and secondarily by milling of pothole floors. Hence, ignoring velocity, the concentration of suspended sediment in floodwaters is the first order control on pothole growth. This is true at other locales as well where, in fact, even more material is removed from pothole walls relative to floors during cylindrical pothole growth (V_w/V_f > 1.14).