A Probabilistic Process‐Based Model of Bank Erosion and Its Application in the Middle Yangtze River

Abstract Bank erosion in a natural alluvial river is influenced not only by near‐bank hydraulic conditions but also predominantly by bank soil properties. These factors exhibit considerable uncertainties, which are seldom considered in previous bank erosion models. Therefore, this study proposes a probabilistic process‐based model of bank erosion by embedding the probability distributions of different bank soil parameters. The model was applied to simulate the bank erosion process in the Middle Yangtze River (MYR), and it was validated against field measurements. Results show that: (a) bank soil parameters including critical shear stress, friction angle and cohesion, followed the Log‐Normal or Gamma distribution, with large variation coefficients of 0.44, 0.59, and 0.50; (b) the expected values of the calculated bank erosion widths agreed closely with measurements (with a relative error of 5%), and the high probability of mass failure occurred within a seasonal timescale (during September–November 2019), despite the high uncertainties in soil properties; and (c) the incorporation of water content variation into the stochastic model further increased the uncertainty of the results by several‐fold, indicating that considering more influencing factors in the model may reduce prediction accuracy. Besides, from a large‐scale perspective, the high diversity of river morphology (channel width) is also closely related to these uncertainties.