Some contributions to the uncertainty of sediment transport predictions

For moderate Shields number flows, the characteristic minimum sediment transport prediction error is estimated to be of the order of a factor of 2, even in stationary and horizontally homogeneous flows over a flat bottom. For oscillatory flows over ripples the same error is judged to be comparable to a factor of 4. In actual flows when the flow, sediments and bottom shape have to be predicted, moderate estimation errors for the near bottom turbulent stress results in characteristic minimum sediment transport error that is probably at least of the order of a factor of 10. The equations for the large-scale average flow require nonstandard parameterizations of the smaller scale ripple flow and boundary conditions. Only weakly founded closure relations are available for this. The error of the large-scale sediment transport can therefore not be quantified without extensive data comparisons. As extrapolated from the simpler estimates, it appears plausible that it could be significantly larger than a factor of 10. When the sediment transport variability is large, both in reality and in models, many models can probably be adjusted to fit the few data available. This may not be relevant for judging the prediction accuracy. (C) 2003 Elsevier Ltd. All rights reserved.