Elsevier

Water Research

Volume 34, Issue 3, 15 February 2000, Pages 807-816
Water Research

Temporal evolution of groundwater composition in an alluvial aquifer (Pisuerga River, Spain) by principal component analysis

https://doi.org/10.1016/S0043-1354(99)00225-0Get rights and content

Abstract

A set of quantitative analytical data from the alluvial aquifer of the Pisuerga river, located at the north-east of Valladolid (Spain), has been processed by multivariate statistical techniques in order to investigate the evolution of the groundwater composition between two surveys. The original matrix consisted of 16 physico-chemical variables, determined in 32 groundwater samples collected in 32 sampling sites of the aquifer. The surveys were carried out in October 1994, at the end of the irrigation and low water period just before autumn rains, and in April–May 1995 at the beginning of the irrigation and high water period and after the recharge of aquifer. The experimental 64×16 matrix was analysed by Principal Component Analysis (PCA), and the resulting Principal Components (PCs) and Varimax rotated PCs (VFs) analysed by means of box and bivariate plots. The exploration of the correlation matrix allowed to uncover strong associations between some variables (sulphate, chloride, hydrogencarbonate, sodium, potassium, magnesium, calcium and electrical conductivity) as well as a lack of association between the others (pH, nitrate, and trace metals). PCA showed the existence of up to five significant PCs which account for 71.39% of the variance. Two of them can be initially assigned to ‘mineralization’ whereas the other PCs are built from variables indicative of pollution. Box and bivariate plots of the two first PCs showed that ‘mineralization’ is maintained through surveys and that trace elements increase from the first to the second survey in all zones, as a possible consequence of soil leaching caused by the rise of the groundwater level. Varimax rotation allowed to ‘break up’ the ‘mineralization’ PC into two VFs, assigned respectively to ‘natural’ mineralization and to ‘saline’ man-made contamination (sodium and chloride). Box and bivariate plots of VFs confirmed the existence of a differential ‘natural’ mineralization pattern, now ascribed exclusively to dissolution of geological materials, as well as the presence of a zone with unusually high levels of ‘saline’ contamination attributed to leakages from sanitary systems. The plot of VFs also allowed to visualise seasonal variations, identifying a process of dilution of the ‘natural’ mineralization during the recharge of the aquifer, and an increase of the ‘saline’ contamination, assigned to a variety of non-natural causes.

Introduction

Groundwater quality depends not only on natural factors such as aquifer lithology, groundwater velocity, quality of recharge waters and interaction with other types of water or aquifers, but also on human activities, which can alter these fragile systems, either by polluting them or by modifying the hydrological cycle, to such an extent that the use of groundwater becomes restricted. In the present work we undertook the study of the alluvial aquifer of the Pisuerga River located at the Northeast of the town of Valladolid (capital of the Comunidad of Castilla y León, Spain). Alluvial aquifers constitute a hydrological unit formed by the alluvial deposits and are characterised by a linear and shallow feature as they are spread along the fluvial valleys of the basin. Their shallow character as well as their high permeability make them highly vulnerable to pollution (Garcı́a-Hernán et al., 1988). In our case, intensive agricultural and industrial activities are carried out in the area, a plant for physico-chemical treatment of toxic and hazardous wastes and an industrial waste disposal landfill are also located in the surroundings. Therefore, the study of the behaviour of the aquifer results to be of priority importance, in order to prevent and to avoid health risks to the nearby human settlements located downstream: the village of Santovenia de Pisuerga and some farms and houses of the north-eastern suburbs of Valladolid, that use groundwater for drinking.

In previous works (Helena, 1996, Helena et al., 1995, Helena et al., 1999) this aquifer has been characterised, confirming its vulnerable character and the great influence that irrigation and polluting sources have on its behaviour. The existence of two different sub-units in the alluvial aquifer (left and right banks of the Pisuerga river) and three different singular zones in the left riverbank was also demonstrated. In this paper the influence of temporal variations on the groundwater is studied, which are of interest because meteorological parameters such as atmospheric precipitations, as well as some polluting sources, such as those derived from agricultural activities, that may affect the quality of groundwater, are strongly dependent on seasonality.

In such a study, a large amount of data is presented. The univariate statistical analysis, generally used to treat such data, could therefore cause misunderstanding and error both in the interpretation and in those to whom the conclusions are presented (Ashley and Lloyd, 1978). In order to avoid this problem, multivariate statistical techniques can be used instead, as they are unbiased methods which can help indicate natural associations between samples and/or variables (Wenning and Erickson, 1994) thus highlighting information not available at first glance. The multivariate treatment of environmental data is widely used to characterise and evaluate surface and freshwater quality, and it is useful for evidencing temporal and spatial variations caused by natural and human factors linked to seasonality (Andrade et al., 1992, Aruga et al., 1995, Brown et al., 1980, Elosegui and Pozo, 1994, Grimalt et al., 1990, Librando, 1991, Reisenhofer et al., 1998, Vega et al., 1998). However, these statistical methods have not attained a comparable diffusion in the case of groundwater studies to date (Aruga et al., 1993, Ashley and Lloyd, 1978, Giammanco et al., 1998). In this paper, Principal Component Analysis (PCA) was used, which allowed to reduce the dimensionality of a highly dimensioned data set by explaining the correlation amongst a large number of variables in terms of a smaller number of underlying factors (principal components or PCs) without losing much information (Jackson, 1991, Meglen, 1992). Varimax rotation (Knudson et al., 1977) has also been applied in order to find more clearly defined factors (called varifactors or VFs) which could be more easily interpreted.

Section snippets

Sampling area

The aquifer investigated is located at the North-east of the town of Valladolid in the Autonomic Community of Castilla y León in central Spain (see Fig. 1) and covers an area of about 25 km2. The studied aquifer was formed by the Pisuerga river, which flows generally from north-east to south-west. The main human settlements are the village of Santovenia de Pisuerga (ca. 1200 habitants) and the north-eastern suburbs of Valladolid (ca. 380,000 habitants). The area supports an important industrial

Results and discussion

The investigated aquifer has been described in previous works (Helena, 1996, Helena et al., 1995, Helena et al., 1999). These studies demonstrated the existence of a heterogeneity in groundwater mineral contents related to anthropic actions such as irrigation, source of irrigation water and location of polluting sources. There are two sub-units in the alluvial aquifer corresponding to the left (L) and right (R) banks of the Pisuerga river, and three different singular areas in the left bank:

Conclusions

Two surveys were made in order to evaluate the modifications of the groundwater quality by seasonal variations. The first one was conducted at the end of irrigation period, just before autumn rains and in a low water situation, and the second one at the beginning of irrigation and in a high water situation after the period of recharge of the aquifer. An experimental 64×16 matrix was found and analysed by multivariate statistical procedures.

The inspection of the correlation matrix of the 16

Acknowledgements

The authors would like to thank Junta de Castilla y León (Consejerı́a de Educación y Cultura), Spain, for the financial assistance which enabled the realisation of this research. B. Helena also wishes to thank the Instituto de Cooperación Iberoamericana (ICI) for its financial support.

References (22)

  • Standard Methods for the Examination of Water and Wastewater

    (1987)
  • Cited by (990)

    View all citing articles on Scopus
    View full text