An Attempt to Take Into Account Natural Variability in 1D Bedload Prediction

Bedload transport can fluctuate considerably over relatively short periods of time and for a given quasi‐constant flow rate. What are the implications of replacing the fluctuating signal with a smoothed signal when calculating bedload transport using averaged values, as is common practice? This question was investigated with the BedloadR code, which allows 1D bedload calculation as well as Monte Carlo simulations using a new data set collected in the Severaisse River (French Ecrins massif). Four bedload equations (Camenen & Larson, 2005, https://doi.org/10.1016/j.ecss.2004.10.019; Meyer‐Peter & Mueller, 1948; Parker, 1990, https://doi.org/10.1080/00221689009499058; Recking, 2013a, https://doi.org/10.1061/(asce)hy.1943‐7900.0000653) were selected for their performance relative to the measured bedload (except for and Meyer‐Peter and Mueller) and because each equation has a different mathematical form and degree of nonlinearity. They were used in a Monte Carlo approach, with input probability distributions fitted to the measured river width, slope, bed grain‐size distribution, and to the associated (computed) Shields stress. The results show that accounting for natural variability in the calculation reproduces bedload fluctuations well. But overall, when calculating the bedload volume transported by a flow event, accounting for variability systematically leads to higher estimated volumes (of the order of 20%) than those obtained with a deterministic approach using average input parameters. This is a direct consequence of the nonlinearity of the equations. Plain Language Summary Transport of bedload in a river, that is, sediment in motion that stays close to the riverbed, fluctuates a lot over relatively short time periods for a given near‐constant discharge. In this paper, we try to reproduce this natural process. To do this, we use a protocol that involves running a bedload equation several thousand times, slightly and randomly modifying the input data with each run. The results show that by taking into account the natural variability of sediment size or channel geometry, we obtain realistic modeling of bedload transport fluctuations. Key Points A new field data set is used for computing bedload in a Monte Carlo approach Bedload modeling accounting for input variability reproduces bedload fluctuation well Because of non‐linearity, volumes computed with input variability are 20% higher