Accounting for natural variability in 1D bedload prediction: a field case study

Any practitioner who has tried to predict bedload transport probably agrees with this final conclusion: it’s a damned task. The first uncertainty comes from the bedload equations themselves, as different equations can lead to several order of magnitude differences when used with the same data. But even in the best situation where a confidence exists on the equation (it can be the case when field measurements permit a comparison for instance), remains the very difficult question of how representative are the data available for computation. For example, it is not uncommon to see studies covering several kilometers of a watercourse, done with a single grain size resumed to a unique median diameter. This is not to blame anyone because data collection is expensive, and a trade-off must necessarily be found between the time and money available to collect the data and its expected completeness. Luckily, in research we have time. In this work we use the opportunity of an extensive campaign carried out in a French Alpine river, for testing how the data variability impacts bedload prediction. The conclusion is that in a complex system (such as braiding morphology) bedload prediction remains uncertain even with very well documented data.