Storm dust source fingerprinting for different particle size fractions using colour and magnetic susceptibility and a Bayesian un-mixing model

A - Papers appearing in refereed journals

Nosrati, K., Akbari-Mahdiabad, M., Ayoubi, S., Degos, E., Koubansky, A., Coquatrix, Q., Pulley, S. and Collins, A. L. 2020. Storm dust source fingerprinting for different particle size fractions using colour and magnetic susceptibility and a Bayesian un-mixing model. Environmental Science and Pollution Research.

AuthorsNosrati, K., Akbari-Mahdiabad, M., Ayoubi, S., Degos, E., Koubansky, A., Coquatrix, Q., Pulley, S. and Collins, A. L.
Abstract

In the context of the continued increased global uptake of fingerprinting procedures to explore fluvial sediment sources, far less attention has been paid to dust source tracing and especially using different particle size fractions and low-cost tracers such as colour and magnetic susceptibility. The objective of this study, therefore, was to apportion local dust storm source contributions for the < 63-μm and 63–125-μm fractions of dust samples in a case study in central Iran. Colour and magnetic susceptibility properties were measured on 62 source samples and six dust storm samples. Statistical methods were used to select four different composite fingerprints for discriminating the dust sediment sources. These statistical approaches comprised (1) the Kruskal–Wallis H test (KW-H), (2) a combination of KW-H and discriminant function analysis (DFA), (3) a combination of KW-H and principal components and classification analysis (PCCA), and (4) a combination of KW-H and a general classification and regression tree model (GCRTM). Local dust source contributions were ascribed using a Bayesian un-mixing model using the final composite fingerprints. For both the < 63- and 63–125-μm fractions, the different composite signatures consistently suggested that alluvial fan material was the dominant source of the dust samples. The root mean square differences between the apportionment results using the different fingerprints ranged from 0.5 to 1.6% for the < 63-μm fraction and from 1.8 to 5.8% for the 63–125-μm fraction. The Wald-Wolfowitz runs test was used to compare the posterior distributions of the predicted source proportions created using the alternative final composite fingerprints and the results indicated that most of the pairwise comparisons were significantly different (p ≤ 0.05). For the < 63-μm fraction, the RMSE and MAE estimates of divergence between the modelled and known virtual source mixtures using the different final composite signatures ranged between 1.5 and 23.4%(with a corresponding mean value of 9.4%). The equivalent estimates for the 63–125-μm fractionwere 1.2–20.1%(8.3%). The findings clearly demonstrate that colour and magnetic susceptibility tracers offer low-cost options for apportioning dust sources.

KeywordsModified MixSIR Bayesian model ; Dust storm tracing ; Statistical techniques; Alluvial fans ; Aeolian sediments
Year of Publication2020
JournalEnvironmental Science and Pollution Research
Digital Object Identifier (DOI)doi:10.1007/s11356-020-09249-3
Open accessPublished as non-open access
FunderBiotechnology and Biological Sciences Research Council
Funder project or codeS2N - Soil to Nutrition - Work package 3 (WP3) - Sustainable intensification - optimisation at multiple scales
Output statusPublished
Publication dates
Online01 Jun 2020
Publication process dates
Accepted11 May 2020
PublisherSpringer Heidelberg
ISSN0944-1344

Permalink - https://repository.rothamsted.ac.uk/item/97y4w/storm-dust-source-fingerprinting-for-different-particle-size-fractions-using-colour-and-magnetic-susceptibility-and-a-bayesian-un-mixing-model

Restricted files

Publisher's version

Under embargo indefinitely

5 total views
0 total downloads
0 views this month
0 downloads this month