MULTIVARIATE STATISTICAL ANALYSES FOR THE CONTAMINANT SOURCES OF GROUNDWATER AT THE MASAN CITY, KOREA

Multivariate statistical techniques, such as cluster analysis, factor analysis and geostatistical analysis were applied to groundwater quality data of the Masan City in Korea. The Masan City is located at the southern part of the Korean Peninsula and is adjacent to the South Sea of Korea. Groundwater samples were collected at 64 shallow and deep wells, and 20 physical and chemical components were analyzed in the field and laboratory. By Piper's trilinear diagram, Ca(HCO₃)₂ type was the most predominant, and NaCl, NaHCO₃ and CaCl₂ types were dominant in order. By factor analysis, the primary source of groundwater contamination was seawater intruded from the South Sea of Korea. The secondary source was nitrates from sewage, fertilizer and the exhaust gas of vehicles. The tertiary source was irons from geology, although the contents of irons were lower than the Korean drinking water standard. In cluster analysis, Ward's method was used for sample classification, and the squared Euclidean distance was used for the measure of similarity between objects. Cluster analysis classified the groundwater quality data into three groups. Group 1 showed the characteristics of fresh groundwater. Group 2 was a little contaminated groundwater from seawater, nitrates and irons. Group 3 was heavily contaminated by seawater. Cluster analysis provided a valuable aid for the spatial classification and contamination assessment of groundwater quality. In geostatistical analysis, ordinary kriging and indicator kriging were used to trace the contaminant sources through the contour map of factor scores calculated from factor analysis. Indicator kriging of binary variables was more useful for the identification of contaminant sources than ordinary kriging.