Modelling of the groundwater hydrological behaviour of the Langueyú creek basin by using N‐way multivariate methods

The groundwater hydrochemical behaviour of the Langueyú creek basin (Argentina) has been evaluated through a systematic survey, followed by application of hydrological and chemometric multivariate techniques. Ten physicochemical parameters were determined in groundwater samples collected from 26 wells during four sampling campaigns (June 2010; October 2010; February 2011 and June 2011), originating a tridimensional experimental dataset X . Univariate statistical and graphical hydrochemical tools (contour maps and Piper diagrams) applied to individual campaigns, allowed to reach some preliminary conclusions. However, a best visualization of the aquifer behaviour was achieved by applying Principal Component Analysis (MA‐PCA) and N‐way PCA procedures, Parallel Factor Analysis and Tucker3. Results were consistent with two‐term models, being Tucker3 [2 2 1] the most adequate, explaining a large amount of the dataset variance (50.7%) with a low complexity. The first Tucker3 [1 1 1] interaction (38.2% of variance) is related with (i) calcium/magnesium versus sodium/potassium ion exchange processes; (ii) an increase of ionic concentration and (iii) a decrease of nitrate pollution, all processes along the direction of the groundwater flow. The second [2 2 1] interaction (12.5% of variance), accounts for the predominant role played by conductivity, bicarbonate and magnesium in the dataset. The seasonal variations are closely related to concentration/dilution phenomena originated by the variations of the phreatic levels, although this point will require additional sampling to establish a definitive hydrochemical model. Copyright © 2013 John Wiley & Sons, Ltd.