Application of Principal Component Analysis and Clustering to Spatial Allocation of Groundwater Contamination

This paper presented a case study that conventional statistical methods were applied to mining environmental monitoring database to extract the patterns of groundwater contamination. Eighty-four monitoring wells located at Chianan Plain groundwater subregion in Taiwan were selected as study area, and lab data of routine groundwater analysis including pH, EC, hardness, TDS, TOC, ammonia, nitrate, chloride, sulfate, Fe, Mn, As, Na, K, Ca and Mg were subjected to factor and cluster analysis. Principal component analysis (PCA) was utilized to reflect those chemical data with the greatest correlation, and PCA results identified four major principal components (PCs) representing 82.4% of cumulative variance. By utilizing PCA, salinization, As dissolution, organic pollution, and mineralization reasonably interpreted the possible underlying processes in the aquifer. Clustering technique was used to evaluate the similarities of water quality in groundwater samples, and 4 clusters were assigned in two-step cluster analysis (CA) in order to correspond with the number of PCs, i.e. the sources of groundwater contamination. Accordingly, CA results distributed all monitoring wells into the domain of each PC, and the domain of groundwater contamination can be spatially allocated by mapping the neighbouring wells within the identical cluster.