Using in-stream extraction to accelerate the recovery of sand-bed rivers impacted by pulses of bedload sediment

© 2020 Alexander James SimsRivers can experience sudden pulses of sediment, from human and natural erosion processes, that can accumulate in the bed. Abundant studies have examined the sources and dynamics of bedload pulses, and problems they cause, particularly flooding, avulsions, and habitat simplification. Much less has been written about what managers can do about bedload pulses. This thesis examines the use of in-stream extraction to accelerate the recovery of sand-bed rivers impacted by pulses of bedload sediment. This thesis investigates the use of in-stream extraction by developing three conceptual models, and then testing each model in the Glenelg River catchment, SE Australia. The conceptual models are (1) bedload pulse dynamics at the catchment scale, (2) reach scale response of a river to a sequence of extraction pits, and (3) how a river responds to multiple interacting interventions at the tail of a bedload pulse. This thesis uses repeat channel surveys, hydraulic and sediment transport modelling, UAV surveys, geomorphic mapping and interviews with extractors to evaluate the conceptual models. Bedload pulses in the Glenelg River do not behave as migrating waves. Instead, sediment accumulates upstream of valley constrictions and a series of standing waves emerge along the river. Wave crests are static, but sediment is exchanged between each of the waves during floods. At the centennial timescale standing waves are eroded one after another, from upstream to downstream, so that their overall behaviour resembles a single, catchment-wide Aggradation Degradation Episode. Extraction pits did not refill with sediment solely by migration of the upstream end. Instead, pit refill is discharge dependant. At bankfull discharge a zone of bedload transport forms between the migrating pit head and the tail of the pit. At the same time a lobe of suspended sediment intrudes into the pit, depositing sediment on the bed of the pit and along the margins of the pit. Point bar rebuilding is controlled by the pattern of secondary circulation in low angle meander bends, and by recirculating eddies on sharp mender bends. A series of extraction pits excavated along a reach interact and cause erosion in downstream reaches. Thalweg erosion increases in a downstream direction, but the presence of in-stream vegetation (Phragmites australis) causes a greater increase in channel complexity than extraction alone. Extraction prevents existing pools from infill but does not cause new pools to scour on the outside of meander bends. A particularly surprising finding of this thesis was that in reaches at the tail of a bedload pulse, where the source of sediment has already been controlled, local-scale interventions can produce highly localised recovery. In-stream vegetation is a key component of the sequence-of-pools morphology that forms in reaches where stock have been excluded. Reaches at the tail of a bedload pulse do not interact, but become disconnected, which is in direct contrast to existing paradigms of rivers as connected systems. Managers can use these results to decide if, where and how in-stream extraction can be used to accelerate recovery in sand-bed rivers impacted by a bedload pulse