Graded bed load transport in sediment supply limited channels under unsteady flow hydrographs

Abstract

Natural river channels that are sediment supply limited due to upstream water management practices and/or flow regulation are often subject to in-channel bed degradation and bed surface restructuring during natural flood hydrographs or controlled flow releases from upstream dams. The heterogeneity of sediment sizes present in these channels means that different size classes are transported more actively during different parts of the hydrograph. A series of laboratory experiments is conducted to investigate the in-channel response of a graded sediment bed to a range of design flow hydrographs when no sediment is supplied at the upstream boundary. The results show varying temporal lag and bed load transport hysteresis for different transported fractions, defined by fine, medium and coarse size classes in the graded sediment mixture. The coarse size class typically exhibits clockwise hysteresis, indicating more active transport during the rising hydrograph limb, whereas the fine size class demonstrates either no/mixed or counterclockwise hysteresis, as it becomes more active during the falling limb. On this separate limb basis, predictions of fractional bed load transport rates are improved by calculating unique reference threshold shear stresses for each size class at the initiation and cessation of fractional grain motions on the rising and falling limbs, respectively. Corresponding temporal variations in the bed load median grain size vary depending on the hydrograph total water work and unsteadiness, with peak values generally attained during the rising limb and overall bed load fining observed during the falling limb. Analysis of the three size classes also indicates that the medium-coarse and fine fractions are transported in larger relative proportions during smaller magnitude, more flashy hydrographs and larger magnitude, flatter hydrographs, respectively. The resulting armouring of the post-hydrograph bed surface layer is most significant at the upstream end of the channel and decreases exponentially in the downstream direction. Two empirical models, based on combined hydrograph and bed sediment descriptors, are also shown to predict reasonably well the overall bed load yields generated under different flow hydrographs from the satisfactory collapse of the current experimental data and previous datasets for both uniform and graded bed sediments. These bed load yields provide an empirical means to describe the extent of upstream bed armouring and the downstream fining of the bed surface layer after the passage of individual hydrographs.