Source fingerprinting sediment loss from sub-catchments and topographic zones using geochemical tracers and weathering indices
Farzaneh Derakhshan-Babaei,Kazem Nosrati,Peter Fiener,Markus Egli,Adrian L. Collins
DOI: https://doi.org/10.1016/j.jhydrol.2024.131019
IF: 6.4
2024-03-08
Journal of Hydrology
Abstract:Application of sediment source fingerprinting techniques to generate reliable information on sediment sources is growing, because of the global environmental importance of elevated fine-grained sediment mobilisation and delivery to rivers. Despite the application of fingerprinting at different spatial scales (e.g., land use types, geological units, and tributary sub-catchments), it has rarely been used to investigate the sediment contribution of topographic zones. The application of sediment source techniques from topographic zones can be helpful in better understanding the spatial distribution of sediment sources within a catchment and the contribution of different topographic features to sediment production and transport. Against this background, this study assessed both topographic zone and tributary sub-catchment spatial sediment sources in the Kan River catchment, Iran. For this purpose, targeted sampling was performed for each classification of spatial sediment sources, and different tracers (geochemical, elemental ratios, and weathering indices) were measured for 80 surface soil samples collected in topographic zones, 20 bed sediment samples collected in sub-catchments, and 11 suspended sediment samples retrieved at the outlet of the main catchment. Three new statistical approaches were applied to identify composite signatures for spatial sediment source discrimination. These approaches combined the Kruskal-Wallis H-test with Particle Swarm Optimization (KW-PSO), Multi-Layer Perceptron (KW-MLP), and Radial Basis Function (KW-RBF) algorithms. A Bayesian un-mixing model was used to apportion the spatial sediment sources. The three composite signatures identified for elucidating the contributions from either classification of spatial sources generated consistent results. The results clarified that the median contribution from topographic zone 1 equated to it being the major spatial sediment source on the basis of topographic zones, with the contributions estimated at 57%, 49%, and 52% (5%-95% uncertainty range) based on the KW-PSO, KW-MLP and KW-RBF signatures, respectively. In terms of the tributary sub-catchments, the Sangan sub-catchment was identified as the dominant spatial sediment source, with corresponding estimated median contributions of 84%, 83%, and 97%, respectively. A combination of weathering indices and elemental fingerprint properties generated statistically robust composite signatures, providing further evidence that the former can augment more conventional tracers. The predicted source proportions can be explained by topographic characteristics from the digital elevation model of the different spatial topographic zones. Therefore, we conclude that topographic features can be an effective predictor of sediment sources.
geosciences, multidisciplinary,water resources,engineering, civil