Source input elucidation in polluted coastal systems by factor analysis of sedimentary hydrocarbon data

The sedimentary hydrocarbon composition (aliphatics, polychlorinated biphenyls and aromatics) of a coastal system that receives the discharges of six rivers (a data base composed by 115 variables and 66 samples) has been studied by principal components analysis and factor analysis (FA). The calculations have followed a stepwise procedure involving: (a) identification of strongly censored variables (X2 statistics on logarithmic data); (b) autoscaling; (c) iterative FA; (d) check for outliers (examination of FA scores in combination with the univariate Chauvenet and Dixon tests); and (e) iterative FA after outlier exclusion. This study has confirmed that FA is consistent with the theoretical background of organic geochemistry in which the molecular composition of the environmental systems is interpreted to originate from a small number of sources. The numerical procedure presented in this study is proposed as a useful systematic way of multivariate analysis of environmental organic molecular data bases.