Step‐pool stability: Testing the jammed state hypothesis

We investigate the stability of step‐pool channels by examining how the traditional approach to bed stability based on the critical Shields number is modified by particles jamming across the width of the channel. Experiments were conducted in a flume with slopes ranging between 3% and 18% and either smooth or rough walls. By varying the size of sediment and width of the flume we observed that the stability of the bed increases as the jamming ratio (channel width/ D 84step is the diameter at which 84% of the step stones are smaller) decreases for jamming ratios less than six. At low jamming ratios both grain‐on‐grain structuring and sediment entrainment phenomena affect the stability of the bed. Actual bed failure, however, depends upon the history of bed development and the chance arrangement of the stone structures in the bed. Thus, the experiments also demonstrate that the inherently stochastic nature of sediment transport affects not only the movement of individual grains but also the stability of the channel as a whole. Since stochastic processes affect the stability of the entire channel, there is no clearly defined separation between stable and unstable beds, rather, an overlapping field where both stable and unstable bed states can exist. This field was modeled using logistic regression to derive a probability of bed failure. A comparison of data from experiments with rough banks and smooth banks showed that rough banks significantly increase the stability of the bed.