A mixing model to incorporate uncertainty in sediment fingerprinting

Abstract

Information on sediment sources is required for effective sediment control strategies, to understand nutrient and pollutant transport, and for developing soil erosion models. Uncertainty associated with sediment fingerprinting mixing models is often substantial, but this uncertainty has not yet been fully incorporated in these models. The main objectives of this study are to apply geochemical fingerprints to determine relative contributions of sediment sources and to develop a Bayesian-mixing model that estimates probability distributions of source contributions to a mixture associated with multiple sources for assessing the uncertainty estimation in sediment fingerprinting in the Hiv catchment, Iran. In this analysis, 28 tracers were measured in 42 different sampling sites from three sediment sources (rangeland, orchard and stream bank) and 12 sediment samples from reservoir check dams. Discriminant analysis provided an important data reduction as it identified four tracers, i.e. B, C, Sr and Tl, that afforded more than 97% correct assignations in discriminating between the sediment sources in the study area. Using a stable isotope mixing model, the median contribution from rangeland, orchard and stream bank sources was 20.8%, 11.2% and 68%, respectively. Sediment source fingerprinting was used to explore the uncertainty in the contributions of sediment from the three sources. Uncertainty is considerable, as the range of probable values was wide: 2–24% for rangeland, 1–26% for orchards and 66–83% for stream banks respectively. While these results can be useful as a scientific basis of sediment management and selecting the soil erosion control methods for decision makers of natural resources they also show that it may not always be possible to identify sediment sources with great precision. Consequently, uncertainty needs to be accounted for when evaluating different management options.