Improving the accuracy of sediment source fingerprinting under particle sorting

Abstract

Sediment fingerprinting has proven highly efficient and rapid in identifying the sources and hotspots of sediment, thereby facilitating the targeted implementation of control measures. However, the unpredictable effect of particle sorting during sediment transport on source identification has increased the uncertainty associated with this method. Here, laboratory sieving experiments were carried out to simulate particle sorting during sediment transport. The conservatism of the five widely used tracers (geochemical elements, mid-infrared, near-infrared, visible, and color properties) and the accuracy of sediment source fingerprinting were investigated in the process of sorting. The results highlight that selected tracers exhibited high source contribution prediction accuracy (mean MAE 6.6 %) without particle sorting, with the exception of the color tracer. In contrast, the predictive ability of all tracers significantly reduced (mean MAE 16.2 %) in simulated particle sorting groups (especially for 125–31 μm) due to the enrichment of fine particles. Moreover, the correction factor failed to improve the accuracy of source contribution prediction in this study. The uncertainty caused by particle sorting can be effectively resolved by narrowing the source and sediment particle sizes to <63 μm (for all five tracers) or <31 μm (except for color tracer). This study contributes a deeper understanding of the role of particle size in sediment source fingerprinting, providing valuable insights to guide future tracer selection and associated particle size criteria.