Compositional baseline assessments to address soil pollution : an application in Langreo, Spain

article

Boente, C.
Albuquerque, M.T.D.
Gallego, J.R.
Pawlowsky-Glahn, V.
Egozcue, J.J.

Elsevier

BOENTE, C. [et al.] (2022) - Compositional baseline assessments to address soil pollution : an application in Langreo, Spain. Science of The Total Environment. DOI 10.1016/j.scitotenv.2021.152383

10.1016/j.scitotenv.2021.152383

Compositional baseline assessments to address soil pollution : an application in Langreo, Spain

Potentially toxic elements
Soil pollution
Compositional indicators
Sequential Gaussian simulation

2022-01-03T19:28:07Z
2022
2024-01-03

Potentially Toxic Elements (PTEs) are contaminants with high toxicity and complex geochemical behaviour and, therefore, high PTEs contents in soil may affect ecosystems and/or human health. However, before addressing the measurement of soil pollution, it is necessary to understand what is meant by pollution-free soil. Often, this background, or pollution baseline, is undefined or only partially known. Since the concentration of chemical elements is compositional, as the attributes vary together, here we present a novel approach to build compositional indicators based on Compositional Data (CoDa) principles. The steps of this new methodology are: 1) Exploratory data analysis through variation matrix, biplots or CoDa dendrograms; 2) Selection of geological background in terms of a trimmed subsample that can be assumed as non-pollutant; 3) Computing the spread Aitchison distance from each sample point to the trimmed sample; 4) Performing a compositional balance able to predict the Aitchison distance computed in step 3. Identifying a compositional balance, including pollutant and non-pollutant elements, with sparsity and simplicity as properties, is crucial for the construction of a Compositional Pollution Indicator (CI). Here we explored a database of 150 soil samples and 37 chemical elements from the contaminated region of Langreo, Northwestern Spain. There were obtained three Cis: the first two using elements obtained through CoDa analysis, and the third one selecting a list of pollutants and non-pollutants based on expert knowledge and previous studies. The three indicators went through a Stochastic Sequential Gaussian simulation. The results of the 100 computed simulations are summarized through mean image maps and probability maps of exceeding a given threshold, thus allowing characterization of the spatial distribution and variability of the CIs. A better understanding of the trends of relative enrichment and PTEs fate is discussed.
info:eu-repo/semantics/publishedVersion

embargoedAccess

eng