Common flaws in the analysis of river sediments polluted by risk elements and how to avoid them: case study in the Ploucnice River system, Czech Republic

摘要

Purpose River channel sediments have been widely used to trace current and historical pollution in fluvial systems, although they are not the only media employed for that purpose. The current practice in such use of sediments from sampling strategy, sample pre-treatment and geochemical data processing frequently includes unsubstantiated and redundant steps (e.g., sieving by sub-millimetre meshes) and neglects certain relevant aspects (insufficient attention to what is actually sampled). The aim of this work was to improve that state and to remove redundant steps that make research more time-consuming and potentially introduce biases.Materials and methods The work presents data obtained in the identification of pollution sources in a small catchment with three tributaries (the Plounice River, the Czech Republic). The sediments were manually sampled in river channels in sites of recent accumulation of fine solids transported by the river. The samples were pulverised and analysed by X-ray fluorescence spectroscopy. The sediments are composed mostly of silt and sand with most chemical elements, including risk elements diluted by quartz and detritic organic matter, which are perfectly suited for geochemical normalization to correctly process compositional data. We avoided statistical tools based on Gaussian distribution, such as means and standard deviations, and instead used median-based statistics better fitting the known properties of geochemical datasets.Results and discussion Sediment sieving to sub-millimetre-size fractions was replaced by geochemical normalization best with Fe and possibly with Rb and Ti, with Al showing implausible performance. The normalization produces grain-size invariant compositional data. The performance of geochemical normalization and downstream variations of risk element concentrations (Pb and Zn) were tested by median-based criteria. Median smoothing of normalized risk element concentrations produced easily interpretable downstream variations of the pollution extent that were independent of grain-size effects and were robust towards the occasional presence of outliers.Conclusions Channel sediments are suitable for fluvial pollution monitoring if biased routines are avoided. The paper can inspire specialists who plan to perform environmental monitoring and desire to simplify their work and produce robust and unbiased estimates of pollution sources in fluvial systems under anthropogenic pressure. A similar approach could be tested for monitoring of pollution by further elements and hydrophobic organic compounds, except for the need to choose another normalizing element or TOC.