Multivariate Statistical Techniques in the Analysis of Groundwater Quality Data – A Case Study

Multivariate statistical techniques such as factor analysis (FA), cluster analysis (CA) and discriminant analysis (DA) were applied to the groundwater quality data set of a tannery-polluted area of Chennai city, India. Groundwater samples were collected from 65 dug wells and analyzed for 25 parameters. FA applied to the data Get resulted in eight factors explaining 78.7% of the total variance. Though FA identified two polluting processes (major ion pollution and tannery pollution factors) explained by factors 1 and 2, it could not explain the remaining factors. Three major clusters corresponding to unpolluted, moderately polluted and severely polluted groups were obtained through CA on the basis of similarities between them. Standard mode DA classified the cases into three groups with 95.4% correct assignations. DA by stepwise mode suggested that Electrical conductivity (EC) is the discriminating variable with 69.2% correct assignation of cases. FA is more effective in identifying the polluting processes, whereas DA is more effective in grouping the sampling stations based on the extent of pollution.