Evaluation of suspended sediment sources in the Yeongsan River

Abstract

Large direct human impacts on the Yeongsan River resulting from the construction of two large weirs on the main stream and dredging of almost all sections of the main river have occurred since late 2009. Subsequent to these human impacts, some problems related to the sediment budget, such as channel changes, have occurred. To establish appropriate policies for control of sediment-associated problems, it is necessary to identify the physical and geochemical characteristics of the reservoir sediments in particulate form in the Yeongsan River. Furthermore, from a management perspective, determining sediment provenance is an essential prerequisite for handling excessive sediment problems in the source-transport-sink system. Hence, the objectives of this study were: i) to identify the physical and geochemical characteristics of the reservoir suspended sediments, ii) to evaluate promptly and quantitatively the relative contributions of slope-floor and channel-bank materials to reservoir sediment, using Cs-137 as a tracer, and iii) to apportion sediment sources of the reservoir sediment by applying composite fingerprinting procedures combined with a multivariate mixing model. Time-integrated suspended sediment samplers were installed in July 2012 at the Seungchon weir (YS-S1), the Juksan weir (YS-S3), and a point halfway between the two weirs, located at the Yeongsanpo waterfront park (YS-S2). Reservoir sediment samples from three samplers were obtained at monthly intervals until October 2014. Strikingly high rates of reservoir sedimentation were recorded in summer, with increased precipitation and discharge. Medium to coarse silt-sized particles dominated the suspended sediment load, however, the proportion of the coarser fraction increased at high flows. Concentrations of the major and trace elements of the collected sediments were analyzed and compared with the chemical composition of the underlying rocks. It was noticeable that easily mobilized elements, such as Ca, Mg, K, and Na, were considerably depleted in the suspended sediments relative to the bedrock of the river basin, indicating weak to moderate degrees of weathering. Trace elements including Ba, Co, Cr, Cu, La, Li, Mo, Nb, Ni, Pb, Rb, Sc, Sn, Sr, Ta, V, W, Y, Zn, and Zr were detected, with an abundance of several elements (e.g. Cu, W, Mo, Ba, Zn, Li, Cr, Rb, Ni, Co, and Pb) that are more likely to be affected by anthropogenic input. For the second objective of this study, potential sources of the suspended sediment were sampled using a soil scraper on the forest floor and channel bank. Source fingerprinting was conducted using Cs-137, which exhibits distinguishable characteristics between surface and subsurface (bank) materials. Probability distribution functions fit to the detected values of Cs-137 at the forest floor and channel bank were derived. The relative contributions of forest floor and bank materials to suspended sediments in the Yeongsan River were calculated using Monte Carlo simulation and a simple mixing model. The results indicated that the dominant source of suspended sediment was bank materials. Conveyance losses of delivered forest floor materials would be expected to occur due to various forms of impediment (e.g., hill-foot zones, alluvial fans, and agricultural dams) that act as barriers to sediment movement. Embankments largely prevent slope materials from flowing into the channel and consequently influence sediment connectivity. Dredging a riverbed and constructing river-crossing facilities can affect the channel in many ways

however, it remains unclear from this study as to how and to what extent human impact has influenced channel erosion, due to the absence of pre-interference data. To reduce the uncertainty and limitations of the Cs-137 methodology, a composite fingerprinting analysis, using fallout radionuclides, geochemical elemental concentrations, and the magnetic properties of each source material coupled with a multivariate mixing model, was conducted. Cropland topsoils, forest topsoils, and channel bank materials were selected as end-members. By applying the Kruskal–Wallis H-test and stepwise discriminant function analysis (DFA), final fingerprints were composed of different characteristics of the properties including excess Pb-210, Zr, V, Pb, and Co. According to the mixing model result, channel bank materials were the dominant source of reservoir sediment during the entire monitoring period. In winter, there was a slight increase in the contribution from forest topsoils, which can be influenced by freeze–thaw actions. An examination as to whether or not incorporation of particle size and organic matter content correction factors into the mixing model was appropriate showed that there is no consistent significant correlation between element concentration and specific surface area (SSA) or organic matter content (LOI). Careful application of tracer-specific correction factors combined with uncertainty analysis is needed to better estimate source apportionment.