Traditional approaches to predicting sediment yields from watersheds frequently involve coupling a soil erosion model with a sediment transport algorithm to predict the downstream sediment flux. This approach possesses several important limitations. The first is a failure to recognise explicitly that the sediment yield from a watershed will usually reflect contributions from a wide range of sources and that sediment mobilisation from all these sources must be represented to provide a meaningful prediction. In some watersheds channel erosion may be of greater importance than slope erosion as a sediment source. Equally, other potential sediment sources, such as unmetalled roads and areas of livestock poaching adjacent to channels can provide important contributions to the sediment yield at a watershed outlet. Failure to include such sources can result in significant underestimation of sediment yield. Sediment source can also exert a key influence on sediment quality and may be a critical consideration in its prediction. A second limitation is a failure to recognise the importance of storage and connectivity in controlling downstream sediment yield. In some, if not many, watersheds, connectivity and associated conveyance losses, as well as short- and longer-term storage exert a key influence on sediment yield and are arguably more important than erosion and sediment mobilisation in controlling downstream sediment fluxes.

Recent advances in sediment tracing, and more particularly sediment source fingerprinting and the use of fallout radionuclides to document the redistribution of soil and sediment within the landscape are now providing a growing body of empirical information on sediment sources, connectivity and storage and it is important that this should be integrated with developments in modelling and prediction of sediment yield. As attention shifts from problems of on-site erosion and downstream reservoir sedimentation to the impacts of sediment on aquatic ecology, it is frequently areas of lower sediment yield that are particularly sensitive to small changes in sediment yield. In these areas a meaningful representation of sediment sources and connectivity and storage is often the key to generating reliable estimates of downstream sediment fluxes and the potential impacts of mitigation measures.