Fang, S.C.; Chang, I.-C., and Yu, T.Y., 2015. Analysis of spatial features of coastal oil pollution using multivariate methods.

Coastal water quality plays an important role in people's lives. In this study, the coastal water of the Penghu archipelago was used as a research object. Three types of multivariate methods—factor analysis, cluster analysis, and multivariate scaling—were employed to analyze the spatial features of oil pollution, and the results of these three methods were compared with regard to the causes of pollution. Factor analysis showed that oil concentration subregions, composed of the first four principal components, accounted for 84.8% of the oil concentration variance. The contour maps of factor loadings of each principal component, combined with the distribution of pollution sources, the direction of seawater flow, etc., might be appropriate to interpret the pollution patterns represented by each principal component. Based on the comparison of the advantages and disadvantages of the three multivariate methods used to explain the geographical features of oil concentration, factor analysis was the best analytic approach in that it could provide information of proportional contributions from a variety of pollution sources on oil concentration variance; it also provided appropriate spatial classification results and a classification number. Factor analysis and multivariate scaling could divide the region into four subregions with consistent oil concentrations. The possible pollution sources in each subregion could be interpreted by locations of sources, contour maps of factor loadings, and relevant factors. Although cluster analysis could differentiate the concentrations measured by all observation stations, its geographical classification result was worse than those of the other two methods.