Utilising CoDA methods for the spatio-temporal geochemical characterisation of groundwater; a case study from Lisheen Mine, south central Ireland (Datasets and Supplementary Materials).

Published: 22-06-2020| Version 1 | DOI: 10.17632/9hmz589r7g.1
Seán Wheeler


This data was used as part of a case study that examines the use of Compositional Data Analysis (CoDA) for spatio-temporal geochemical characterisation. The primary data file (GroundwaterGeochemistryData.csv) describes the results of geochemical analysis of groundwater from 8 wells over a 12 month period in the area of Lisheen Mine, south-central Ireland. The subsequent data treatment process (R_CodeForDataTreatment.R) allows for the original dataset to be compatible with CoDA.


Steps to reproduce

1. Groundwater Sample Collection and Analysis The samples were collected and analysed by an Irish National Accreditation Board-certified company (IAS Laboratories) for the presence of SO4, Cl, NO3, F, NH4, NO2, P, Ca, Na, K, Mg, Fe, Mn, Cu, Zn, Pb, Al, Ni, Ba, As, Hg, B, Cr, Cd, Mo, Ag, Co, Sr, Be, Sb, and U as well as a number of other parameters (e.g. alkalinity, pH, temperature, dissolved oxygen etc.). Geochemical parameters were also analysed at the ISO accredited IAS Laboratories in County Carlow, Ireland using a combination of ICP-OES (for cations) and ion-chromatography (for anions). The limit of detection (LOD) value for a given element or compound was used as a replacement for any parameter that could not be measured in a given sample. 2. Data Treatment Compositional Data Analysis (CoDA) requires a dataset composed of real positive numbers. Therefore missing cells in the raw data (approximately 2% of the dataset) were replaced with the mean value for that parameter at that sample location. Samples recorded as the detection limit were then replaced using expectation maximisation algorithms in R (version 3.5.2)(code attached). 3. Subsequent Data Analysis Output data tables from the R code were analysed using 'CoDA Pack' (version 2.02.21). In addition the data was analysed using Piper Diagrams in Aquachem (version 2014.2) and using Hierarchical Cluster Analysis (HCA) in SPSS (version 23).