Multivariate Statistical Analysis of Geochemical Data of Groundwater in Veeranam Catchment Area, Tamil Nadu

Transcription

1 JOURNAL GEOLOGICAL SOCIETY OF INDIA Vol.74, November 2009, pp Multivariate Statistical Analysis of Geochemical Data of Groundwater in Veeranam Catchment Area, Tamil Nadu M. SUVEDHA 1, B. GURUGNANAM 2, M. SUGANYA 3 and S. VASUDEVAN 4 1 Department of Geology, Alagappa Govt. Arts College, Karaikudi Department of Earth Sciences, Annamalai University, Annamalai Nagar No.572, P.S.P.Street, Soodamani Nagar, Karaikudi Department of Geology, Bharathidasan University, Trichy swetha_earth@yahoo.co.in Abstract: The study of hydrogeochemistry of the Mio-Pliocene sedimentary rock aquifer system in Veeranam catchment area produced a large geochemical dataset. Groundwater samples were collected at 52 sites over km 2 area and analyzed for major ions. The large number of data can lead to difficulties in the integration, interpretation and representation of the results. Two multivariate statistical methods, Hierarchical cluster analysis (HCA) and Factor analysis (FA), were applied to a subgroup of the dataset to evaluate their usefulness to classify the groundwater samples, and to identify geochemical processes controlling groundwater geochemistry. Hydrochemical data for 52 groundwater samples were subjected to Q- and R- mode factor and cluster analysis. R-mode analysis reveals the inter-relations among the variables studied and the Q-mode analysis reveals the inter-relations among the samples studied. The R-mode factor analysis shows that Ca, Mg and Cl with HCO 3 account for most of the electrical conductivity, total dissolved solids and total hardness of groundwater. The single dominance nature of the majority of the factors in the R-mode analysis indicates non-mixing or partial mixing of different types of groundwater. Both Q-mode factor and Q-mode cluster analyses indicate an exchange between the river water and the groundwater in the vicinity. The rock water interaction like flood basin back swamp deposits of silty clayey formation is the major cause for the cluster II classification. Cluster classification map reveals that 58% of the study area comes under cluster II classification. Keywords: Groundwater, Multivariate statistical analysis, Geochemical data, Tamil Nadu. INTRODUCTION The objective of the study is to identify the processes controlling the geochemical evolution of groundwater by using two proven methods of multivariate analysis of the geochemical data sets, namely Hierarchical cluster analysis (HCA) and Factor analysis (FA). The relatively complex setting and geological history of the study area, use of HCA and FA aims at distinguishing respective roles of geological and hydrogeological factors in this hydrochemical evolution. We also assessed the relative applicability and complementarities of HCA and FA methods compared to conventional geochemical grouping in achieving the scientific evaluations. Multivariate statistical analysis has been successfully applied in a number of hydrogeochemical studies. Steinhorst and Williams (1985) used multivariate statistical analysis of water chemistry data in two field studies to identify groundwater sources. Usunoff and Guzma n-guzma n (1989) demonstrated the usefulness of the approach in hydrogeochemical investigations for understanding the geological and hydrogeological state of the aquifer. Multivariate treatment of environmental data is also widely used to characterize and evaluate groundwater quality (Vengosh and Keren, 1996; Suk and Lee, 1999; Helena et al. 2000; Reghunath, 2002; Lambrakis et al. 2004; Panagopoulos et al. 2004, Vincent Cloutier et al. 2008). It is also useful for identifying temporal and spatial variations caused by natural and human factors linked to seasonality. STUDY AREA The study area, the Veeranam catchment, occupies an area of km 2, falling in parts of Cuddalore and Perambalur districts, Tamil Nadu. It lies between the North latitudes 11 05'56" ' and East longitudes 79 15'30" '10" (Fig.1). Physiographically, the area is flat with gentle slope, experiences high rainfall from the northeast monsoon. Geologically, the area is underlain by alluvial deposits of Early to Middle Pleistocene. The nature and character of the alluvium have been studied, / /$ 1.00 GEOL. SOC. INDIA

2 574 M. SUVEDHA AND OTHERS Table 1. Mean and standard deviation of the chemical parameters of groundwater Parameter Valid N Mean Minimum Maximum Std. Dev 11 20'0"N Na K Ca Mg Cl HCO SO ph EC TDS TH RSC SAR '0"N Fig.1. Map showing water sample locations. Unit: Concentration in ppm except ph, EC (µs Cm -1 ), RSC and SAR (meq l -1 ). cations and anions were estimated by titration method. Residual sodium carbonate (RSC) was calculated by subtracting (Ca+Mg) from the values of carbonates and bicarbonates expressed as epm (Eaton, 1950). Sodium absorption ratio (SAR), was calculated by dividing sodium with root of half (Ca+Mg) expressed as epm (Richard, 1954). MULTIVARIATE STATISTICAL ANALYSIS based on the geological sections prepared from the well logs of the tube wells in the area. From these logs, it is evident that alluvial deposits which form the potential aquifers primarily consist of thick deposits of mottled sandstone, clay and lignite deposits of Mio-Pliocene age. The Quaternary formations are restricted to the alluvium of Cauvery, Kollidam and their distributaries which occurs as a isolated remnant patches over Cuddalore Formation. The area is bounded by the river Vellar in the north and Kollidam in south, running along the eastern to northwestern and northern to northeastern boundaries. MATERIALS AND METHODS GEOCHEMISTRY Fifty two water samples were collected in May 2006 from different shallow dug wells and deep bore wells, which are almost uniformly distributed over the study area. Only those wells were selected for sampling purpose which are in constant use and approachable. After half an hour discharge from the tube wells, the samples were collected in air tight bottles with stoppers and subjected to chemical analysis to see the variations in quality parameters. Major Factor Analysis (FA) Multivariate techniques can help to simplify and organize large data sets and to make useful generalizations, that can lead to meaningful insight (Laaksoharju et al. 1999). Cluster and factor analyses are efficient ways of displaying complex relationships among many objects (Davis, 1986). The two methods in cluster and factor analyses, i.e. Q- and R- mode analyses have been done for the data generated. R-mode analysis reveals the interaction among the variables studied and the Q-mode analysis reveals the interrelation among the samples studied. The software packages like Statistical Package for Social Sciences (SPSS) and STATISTICA 6 have been used to carry out the analysis. The data have been standardized by using standard statistical procedures. Hierarchical Cluster Analysis (HCA) Cluster analysis comprises a series of multivariate methods which are used to find true groups of data. In clustering, the objects are grouped such that similar objects fall into the same class (Danielsson et al. 1999). Hierarchical clustering joins the most similar observations, and then successively the next most similar observations. The levels of similarity at which observations are merged are used to

3 STATISTICAL ANALYSIS OF GEOCHEMICAL DATA OF GROUNDWATER, VEERANAM CATCHMENT, TAMIL NADU 575 construct a dendrogram. In this study, a standardized space Euclidian distance (Davis, 1986) is used. A low distance shows the two objects are similar or close together, whereas a large distance indicates dissimilarity. RESULTS AND DISCUSSION The chloride, calcium, sodium and bicarbonate content shows a significant difference between the medium and maximum values, the mean values being near the quarter values of the maximum values. It suggests that local contamination to the groundwater system. The wide range of bicarbonate contents, from to ppm is the result of the lateral geological variations of the layers. Box plots of the chemical concentration show that bicarbonate, calcium, chloride and TDS have the largest dispersions (Fig.2). The enrichment of chloride and TDS from values of to and to respectively, is observed in the groundwater on the eastern side of the study area. It is due to the high enrichment of the flood basin back swamp deposit of silty clayey formation. The increase in the salt concentration could be associated with different mechanisms like water rock interaction processes. R Mode Factor Analysis R-mode factor analysis of different chemical constituents of the groundwater of Veeranam catchment area has been carried out. All cations and anions, TDS, EC, ph and hardness have been considered for the present analysis. The Table 2. R Mode factor analysis with Varimax normalized rotation Parameter Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Na K Ca Mg Cl HCO SO ph EC TDS TH RSC SAR Eigenvalue % Total Variance Cumulative % analysis generated five factors which together account for 95.08% of variance. The rotated loadings, eigen values, percentage of variance and cumulative percentage of variance of all the five factors are given in Table 2. The first eigen value is 7.69 which accounts for 59.2% of the total variance and this constitutes the first and main factor. The second and third eigen values are 2.33 and 1.01 and these account for 18% and 7.84% respectively, of the total variance. Each of the remaining eigen values constitutes less than 10% of the total variance. The first factor (which accounts for 59.2% of the total variance) is characterised by very high loadings of Ca, Mg, Cl and EC, and moderate to high loadings of bicarbonate and ph. This factor reveals that the EC and TDS in the study area are mainly due to Ca and Mg and Cl, though bicarbonate also plays a substantial role in determining EC and TDS. This factor accounts for the temporary hardness of the water. The second factor (which accounts for 18% of the total variance) is mainly associated with very high loading of Na, Cl and bicarbonate, and also with moderate loading of TDS. This factor accounts for the temporary salinity of the water. The loading of bicarbonate is same as the first factor. Factors 3 5 are characterized by the dominance of only one variable each, such as SO 4 (factor 3), HCO 3 (factor 4) K (factor 5), and together these six factors account for 17% of the total variance. Q Mode Factor Analysis Fig.2. Box and Whisker plot for chemical parameters of groundwater samples. The rotated loadings, eigen values, percentage of variance and cumulative percentage of variance of the first three factors are given in Table 4. Q-mode factor analysis of the 52 groundwater samples generated three factors which

4 576 M. SUVEDHA AND OTHERS together accounted for 99.87% of the total variance (Table 3). The three factors obtained in this way were rotated using the Varimax procedure (Knudson et al. 1977), which could be more easily interpreted. The first factor (which explains 37.96% of the total variance) was considered as major factor controlling the relative proportions of major element existing in the groundwater samples and had the high loadings of almost all the samples except those from location nos. 28 and 46. On the other hand, groundwater samples from the three locations 28, 46 and 51 had high loadings in the second factor. As mentioned earlier, the Q-mode factor analysis described the relative proportions of these major elements in groundwater samples. Therefore, the relative proportions of major elements in these groundwater samples were controlled completely by the three factors which together explain 99.87% of the total variance. The distribution of wells explained by factors 2 and 3 do not conform to any kind of spatial pattern. However, the majority of the samples within factor 1 fall on either side of the main course of the river system. This strongly suggests that there is an exchange between the river water and groundwater in the vicinity. This has also been discussed by Reghunath et al. (2002). HIERARCHICAL CLUSTER ANALYSIS The HCA is a data classification technique. There are different clustering techniques, but the hierarchical clustering is the one most widely applied in Earth sciences (Davis, 1986), and often used in the classification of hydrogeochemical data (Steinhorst and Williams, 1985; Schot and van der Wal, 1992; Ribeiro and Macedo, 1995; Gu ler et al., 2002). The result of the hierarchical cluster analysis was given as a dendrogram (Fig.3). For this project, the Euclidean distance was chosen as the distance measure, or similarity measurement, between sampling sites. The sampling sites with the larger similarity are first grouped. Next, group of samples are joined with a linkage rule, and the steps are repeated until all observations have been classified. With this geochemical dataset, Ward s method was more successful to form clusters that are more or less homogenous and geochemically distinct from other clusters, compared to other methods such as the weighted pair-group average. Ward s method is distinct from other linkage rules, because it uses an analysis of variance approach to evaluate the distances between clusters (StatSoft Inc., 2004). Other studies in their cluster analysis (Adar et al. 1992; Schot and van der Wal, 1992). Gu ler et al. (2002) also found that using the Euclidean distance as a distance measure and Table 3. Q Mode factor analysis with Varimax normalized rotation S.No Factor 1 Factor 2 Factor Eigenvalue % Total Variance Cumulative %

5 STATISTICAL ANALYSIS OF GEOCHEMICAL DATA OF GROUNDWATER, VEERANAM CATCHMENT, TAMIL NADU 577 Sampling Location Linkage Distance Fig.3. Dendorgram of the hierarchical cluster analysis using the Ward method. Ward s method as a linkage rule produced the most distinctive group. There are three major clusters as shown in Fig. 2. Clusters 1, 2 and 3 correspond to the factors 1, 2 and 3 respectively. The similarity of the Q-mode cluster analysis to the Q-mode factor analysis confirms the interpretations made using the Q-mode factor analysis. To understand the spatial distribution of various cluster classes, the results were taken into GIS platform wherein spatial distribution map is prepared (Fig.4). The salient findings of spatial distribution map are given in the Table 4. aquifer lithology dominated by calcareous sandstone and clayey formation. The factor analysis reveals that the calcium and magnesium concentrations are the major sources for 11 20'0"N N Table 4. Results of Cluster Classification spatial distribution map Grid code Cluster classification Area in km 2 1 Cluster I Cluster II Cluster III '0"N CONCLUSION The scientific evaluation of the raw data by FA and HCA leads to the conclusion that the water rock interaction process is the major mechanism responsible for the groundwater salinity in the study area. The water samples are mainly of calcium-bicarbonate type, pointing to the Fig.4. Spatial distribution of map of cluster classes.

6 578 M. SUVEDHA AND OTHERS the hardness of groundwater. An anthropogenic contamination was identified in both the aquifers, due to local pollution inputs. The Q-mode factor and cluster analyses indicate that exchange between the river water and the groundwater plays a dominant role in the hydrochemical evolution of groundwater. Cluster classification map reveals that 58% of the study area comes under cluster II classification. References ADAR, E.M., ROSENTHAL, E., ISSAR, A.S. and BATELAAN, O. (1992) Quantitative assessment of the flow pattern in the southern Arava Valley (Israel) by environmental tracers and a mixing cell model. Jour. Hydrology, v.136, pp DANIELSSON, A., CATO, I., CARMAN, R. and RAHM, L. (1999) Spatial clustering of metals in the sediments of the Skagerrak/Kattegat. Applied Geochemistry, v.14, pp EATON, E.M. (1950) Significance of carbonate in irrigation water. Soil Sci., v.69, pp GÜLER, C., THYNE, G.D., MCCRAY, J.E. and TURNER, A.K. (2002) Evaluation of graph ical and multivariate statistical methods for classification of water chemistry data. Hydrogeology Jour., v.10, pp HELENA, B., PARDO, B., VEGA, M., BARRADO, E., FERNANDEZ, J.M. and FERNANDEZ, L. (2000) Temporal evolution of groundwater composition in an alluvial aquifer (Pisuerga River, Spain) by rincipal component analysis. Water Res., v.34(3), pp KNUDSON, E.J., DUEWER, D.L., CHRISTIAN, G.D. and LARSON, T.V. (1977) Application of factor analysis to the study of rain chemistry in the Puget Sound region. In: B.R. Kowalski (Ed.), Chemometric: Theory and Application. ACS Symposium Series, Washington, DC, pp LAMBRAKIS, N., ANTONAKOS, A. and PANAGOPOULOS, G., (2004) The use of multicomponent statistical analysis in hydrogeological environmental research. Water Res., v.38, pp PANAGOPOULOS, G., LAMPRAKIS, N., TSOLIS-KATAGAS, P. and PAPOULIS, D. (2004) Cation exchange processes and human activities in inconfined aquifers. Environ. Geol., v.46, pp REGHUNATH, R., SREEDHARA, M.T.R. and RAGHAVAN, B.R. (2002) The utility of multivariate statistical techniques in hydrogeochemical studies: an example from Karnataka, India. Water Res., v.36(10), pp RIBEIRO, L. and MACEDO, M.E. (1995) Application of multivariate statistics, trend and cluster analysis to groundwater quality in the Tejo and Sado aquifer. In: Groundwater Quality: Remediation and Protection. Proceedings of the Prague Conference, May IAHS Publ. No.225, pp SCHOT, P.P. and VAN DER WAL, J. (1992) Human impact on regional groundwater composition through intervention in natural flow patterns and changes in land use. Jour. Hydrology, v.134, pp STATSOFT INC. (2004) STATISTICA (Data Analysis Software System), Version 6. STEINHORST, R.K. and WILLIAMS, R.E. (1985) Discrimination of groundwater sources using cluster analysis, MANOVA, canonical analysis and discriminant analysis. Water Resources Res., v.21, pp SUK, H. and LEE, K. (1999) Characterization of a ground water hydrochemical system through multivariate analysis: clustering into ground water zones. Ground Water, v.37(3), pp VENGOSH, A. and KEREN, R. (1996) Chemical modifications of groundwater contaminated by recharge of treated sewage effluent. Contam. Hydrol., v.23, pp VINCENT CLOUTIER., RENÉ LEFEBVRE., RENÉ THERRIEN., MARTINE, M. and SAVARD. (2008) Multivariate statistical analysis of geochemical data as indicative of the hydrogeochemical evolution of groundwater in a sedimentary rock aquifer system Jour. Hydrology, v.353, pp (Received: 28 April 2008; Revised form accepted: 20 June 2009)