AbstractThe composition, grain‐size, and flux of stream sediment evolve downstream in response to variations in basin‐scale sediment delivery, channel network structure, and diminution during transport. Here, we document downstream changes in lithology and grain size within two adjacent ~300 km2 catchments in the northern Rocky Mountains, USA, which drain differing mixtures of soft and resistant rock types, and where measured sediment yields differ two‐fold. We use a simple erosion–abrasion mass balance model to predict the downstream evolution of sediment flux and composition using a Monte Carlo approach constrained by measured sediment flux. Results show that the downstream evolution of the bed sediment composition is predictably related to changes in underlying geology, influencing the proportion of sediment carried as bedload or suspended load. In the Big Wood basin, particle abrasion reduces the proportion of fine‐grained sedimentary and volcanic rocks, depressing bedload in favor of suspended load. Reduced bedload transport leads to stronger bed armoring, and coarse granitic rocks are concentrated in the stream bed. By contrast, in the North Fork Big Lost basin, bedload yields are three times higher, the stream bed is less armored, and bed sediment becomes dominated by durable quartzitic sandstones. For both basins, the geology‐based mass balance model can reproduce within ~5% root‐mean‐square error the composition of the bed substrate using realistic erosion and abrasion parameters. As bed sediment evolves downstream, bedload fluxes increase and decrease as a function of the abrasion parameter and the frequency and size of tributary junctions, while suspended load increases steadily. Variable erosion and abrasion rates produce conditions of variable bed‐material transport rates that are sensitive to the distribution of lithologies and channel network structure, and, provided sufficient diversity in bedrock geology, measurements of bed sediment composition allow for an assessment of sediment source areas and yield using a simple modeling approach. Copyright © 2016 John Wiley & Sons, Ltd.
Read full abstract