Sediment transport in braided rivers shows significant temporal variability, characterized by intense bursts of transport rates even under steady-state conditions. These fluctuations are often attributed to random variations in local hydraulic conditions and sediment movement. While such explanations emphasize small-scale processes, they do not explain how larger-scale channel structure organizes these fluctuations. The role of reach-scale channel morphology in controlling sediment transport remains unexplored. Here we show that temporal variability in sediment transport results from switching between a finite set of recurrent morphological configurations.

We used results from a former long-term (1,200-hour) laboratory experiment conducted under constant water discharge and sediment supply, in which sixteen morphological states were identified using image processing. This paper shows that the temporal evolution of these states can be modeled using a discrete-time Markov process. We assume sediment transport depends on the active channel state. State-dependent transport rates are simulated using either an empirical probability distribution fitted to the data or a Gamma distribution with parameters dependent on stream power. A third model ignores morphological state and simulates transport rates using bootstrap resampling.

We used the three models to generate long time series of transport rates, whose ensemble statistics were compared with experimental results. All three models perform similarly in predicting mean transport rate and variance, but the purely Markovian model better captures temporal dynamics. These findings indicate that variability in bedload transport mainly results from the persistence and succession of a finite set of morphological states.