ABSTRACT
Multivariate statistical analysis of sediment data (input matrix 122 x 15) collected from 122 sampling sites from the western coastline of the USA and analyzed for 15 analytes indicates that the data structure could be explained by four latent factors. These factors are conditionally named "anthropogenic", "organic", "natural", and "hot spots". They explain over 85% of the total variance of the data system, which is an acceptable value for the PCA model. The receptor models obtained after regression of the mass on the absolute principal components scores ensures reliable estimation of the contribution of each possible natural or anthropogenic source to the mass of each chemical component. It can be concluded that the region of interest reveals a different pattern of pollution compared with the eastern coastline treated statistically in a previous study.
