Elucidating intra-storm variations in suspended sediment sources using a Bayesian fingerprinting approach

There is an increasing need for reliable information on the origin of fine-grained sediment transported by river systems, but critically, at temporal scales that reflect the timing and pattern of sediment mobilization and delivery in response to effective precipitation events. Temporal variability is likely to be more pronounced in steep mountainous catchments. Accordingly, the main objective of this study was to investigate the temporal variability of sediment sources during and between flood events, in a mountainous catchment in Iran, using ageochemical fingerprinting approach and Bayesian un-mixing model. Potential sediment sources were classified as either tributary sub-basin spatial end members or individual source types represented by surface soils and channel banks. A total of 34 geochemical elements, elemental ratios and weathering indices were measured as potential tracers on a total of 155 samples, including 40 target samples collected across the hydrographs of four flood events at the overall catchment outlet. Virtual mixture tests, used to compare the computed source estimates with known source mixtures, revealed better fits in the case of spatial, rather than source type apportionment The Bayesian un-mixing model results revealed pronounced intra-storm variations in the relative contributions from both the spatial sources and source types, but with Quaternary surface soils and channel banks being dominant at different stages of the hydrographs. For all the suspended sediment samples collected throughout the three small to medium size events, the Babazid, middle upstream sub-catchment, was the major (84%–99%) sediment source while for the largest sampled event, suspended sediment samples predominantly (>98%) originated from the Kalkoshan, the most southern sub-catchment. By investigating both spatial sources and source types, temporal variations in the dominant sources were revealed. Quaternary geology and channel banks were identified as the main sediment sources. By combining instantaneous suspended sediment loads through the sampled flood events (across all floods) and using the corresponding source proportions derived from the MixSIR un-mixing model, channel banks and channel bank and surface soils generated 2911 kg and 34,171 kg of suspended sediment, respectively. Whilst the use of spatial source end members is practical in medium-size drainage basins, integration with estimates of sediment source type contributions is more informative to managers in selecting relevant sediment mitigation methods. The intra-storm results underscore the need to manage sediment loss from both surface soils and channel banks.