A comparison of geological and statistical approaches to element selection for sediment fingerprinting

Abstract

Purpose: Elevated sediment loads reduce reservoir capacity and significantly increase the cost of 25 operating water treatment infrastructure making the management of sediment supply to reservoirs of 26 increasing importance. Sediment fingerprinting techniques can be used to model the relative contributions of different sources of sediment accumulating in reservoirs. The goal of this research is 28 to compare geological and statistical approaches to element selection for sediment fingerprinting 29 modelling. 30 Materials and methods: Time-integrated samplers (n=45) were used to obtain source samples from 31 four major subcatchments flowing into the Baroon Pocket Dam in South East Queensland, Australia. 32 The geochemistry of these potential sources were compared to sediment cores (n=12) sampled in the 33 reservoir. Elements that provided expected, observed and statistical discrimination between sediment 34 sources were selected for modelling with the geological approach. Two statistical approaches selected 35 elements for modelling with the Kruskal-Wallis H-test and Discriminatory Function Analysis (DFA). 36 In particular, two approaches to the DFA were adopted to investigate the importance of element 37 selection on modelling results. A distribution model determined the relative contributions of difference 38 sources to sediment sampled in the Baroon Pocket Dam. 39 Results and discussion: Elemental discrimination was expected between one subcatchment (Obi Obi 40 Creek) and the remaining subcatchments (Lexys, Falls and Bridge Creek). Six major elements were 41 expected to provide discrimination. Of these six, only Fe 2 O 3 and SiO 2 provided expected, observed 42 and statistical discrimination. Modelling results with this geological approach indicated 36% (+/-9%) 43 of sediment sampled in the reservoir cores were from mafic-derived sources and 64% (+/-9%) were 44 from felsic-derived sources. The geological and the first statistical approach differed by only 1% (σ 45 5%) for 5 out of 6 model groupings with only the Lexys Creek modelling results differing 46 significantly (35%). The statistical model with expanded elemental selection differed from the 47 geological model by an average of 30% for all 6 models. 48 Conclusions: Elemental selection for sediment fingerprinting therefore has the potential to impact 49 modeling results. Accordingly we believe it is important to incorporate both robust geological and 50 statistical approaches when selecting elements for sediment fingerprinting. For the Baroon Pocket 51 Dam, management should focus on reducing the supply of sediments derived from felsic sources in 52 each of the subcatchments.

Fingerprinting sediment sources in a large agricultural catchment under no‐tillage in Southern Brazil (Conceição River)

“…In addition to modeling and monitoring studies, the sediment fingerprinting technique was shown to provide a powerful tool to identify sediment sources in Brazil (Miguel, Dalmolin, Pedron, Moura‐Bueno, & Tiecher, ; Minella, Merten, Reichert, & dos Santos, ; Minella, Merten, Walling, & Reichert, ; Minella, Walling, & Merten, , ; Tiecher et al, ; Tiecher, Caner, et al, ; Tiecher, Caner, Minella, & dos Santos, ; Tiecher, Caner, Minella, Bender, & dos Santos, ; Tiecher, Minella, et al, ) and elsewhere in the world, such as in France (Foucher et al, ; Legout et al, ; Navratil, Evrard, Esteves, Legout, et al, ; Poulenard et al, ), Mexico (Evrard et al, ), South Africa (Foster, Boardman, & Keay‐Bright, ), Zambia (Walling, Collins, Sichingabula, & GJL, ), Canada (Barthod et al, ; Koiter et al, ; Stone, Collins, Silins, Emelko, & Zhang, ), United States (Devereux, Prestegaard, Needelman, & Gellis, ), United Kingdom (Collins, Walling, Webb, & King, ; Pulley, Foster, & Antunes, ; Smith & Blake, ), Spain (Brosinsky, Foerster, Segl, & Kaufmann, ), Luxemburg (Martínez‐Carreras et al, ), Turkey (D'Haen et al, ), Tunisia (Ben Slimane et al, ), Iran (Haddadchi, Nosrati, & Ahmadi, ), and Australia (Laceby, McMahon, Evrard, & Olley, ; Olley, Burton, Smolders, Pantus, & Pietsch, ; Wilkinson, Hancock, Bartley, Hawdon, & Keen, ). Therefore, the objective of the current study is to quantify the contribution of sources supplying sediment to the river in a catchment characterized by the widespread implementation of NT since the 1990s in Southern Brazil.…”

Tracing Sediment Sources in a Subtropical Agricultural Catchment of Southern Brazil Cultivated With Conventional and Conservation Farming Practices

“…This model simulation and solving process was repeated 2,500 times with the median proportional source contribution from these 2,500 simulations reported as the contribution of each source. Model uncertainties were determined through the sum of the modelled SD, the median absolute deviation of the modelled source contribution and the median absolution deviation of the modelled SD for the 2,500 simulations (Laceby et al ., ). Results are presented using the notation ± to express the standard error associated with one sample whereas the mean and SD are used for groups of samples.…”

“…Often, sediment fingerprinting studies rely on the discrimination of potential sediment sources according to land use (e.g. cropland, channel bank, road, pasture and damaged roads) (Collins et al, 2012;Sherriff et al, 2015) or catchment geology (Evrard et al, 2011;Hughes et al, 2009;Laceby et al, 2015). Recently, a geochemical fingerprinting approach was used to apportion sediment sources based on catchment soil types (Lepage et al, 2016).…”

Do forests represent a long-term source of contaminated particulate matter in the Fukushima Prefecture?

“…Model uncertainty was calculated by summing the standard deviation of each source's relative contribution, the modelled median absolute deviation (MAD) of these contributions, and the MAD of the modelled standard deviation for the 2500 model simulations (Laceby et al, 2015a). Paired two-tailed t-tests or Wilcoxon signedrank tests compare source contribution results derived from modelling different organic properties (i.e.…”