摘要：Despite the importance of groundwater in Terengganu, Malaysia, quality assessment has received little attention, and effort to use hydrochemistry data to solve particular problems are even fewer or non-existent. This paper, reports results from large hydrochemistry data analysed using multivariate statistical techniques such as Cluster Analysis (CA), Discriminant Analysis (DA) and Principal Component Analysis (PCA) with the objectives of determining the spatial variability of groundwater and to identify the sources of pollution that presently affects the groundwater. The water quality data was monitored at ten different wells, over the period of six years (2006-2011) using 24 water quality parameters. The CA allowed the formation of three clusters between the sampling wells reflecting differences on water quality at different locations. DA as a data reduction techniques was used to evaluate spatial variability in water quality, as it uses only 3 parameters (Ca+, NO2, and PH) affording 73.33% correct assignation to discriminate between the clusters using forward stepwise mode from the original 24 parameters, while backward stepwise mode yielded 83.33% correct assignation to discriminate nine parameters (Ca+, Mg2+, Fe2+, SO4-, Cl-, AS, Mn, NO2, and conductivity). PCA was used to examine the root of each water quality parameter due to nature and anthropogenic activities based on the three cluster regions. It identified eight PC’s, responsible for 76.45% of the total variance in the data set. The main factors obtained indicate that parameters influencing groundwater quality of the clusters are mainly related to natural (dissolution of soil and rocks), pointsource (municipal wastewater and industries) and non-point source pollution (agriculture) in the region. The results of this study clearly demonstrate the usefulness of multivariate statistical techniques in Geochemistry.
关键词：Multivariate statistical techniques; Cluster analysis; Discriminate analysis; Principal component analysis
