A multivariate approach to oil hydrocarbon fingerprinting and spill source identification

Abstract

Analytical techniques, which are used for oil hydrocarbon fingerprinting, include gas chromatography–flame ionization detection (GC–FID), gas chromatography–mass spectrometry (GC–MS), and fluorescence spectroscopy. Oil hydrocarbon fingerprinting and spill source identification are, however, not limited to the chemical characterization using different analytical techniques, but consist of a combination of analytical techniques and methods for data preprocessing, analysis, and evaluation of the results. Rapid, reliable, and objective tools are a requirement for the characterization of complex chemical mixtures such as oil. This chapter describes the development of such tools for oil hydrocarbon fingerprinting and spill source identification. One of the most important advances in oil hydrocarbon fingerprinting is the systematic use of multivariate statistical methods for comprehensive and objective comparison and classification of oil from single and multiple sources. The use of multivariate statistical methods such as PCA and PARAFAC are the cornerstone of the integrated multivariate oil fingerprinting (IMOF) methodology. The multivariate methods enable the analysis and assessment of large datasets by extracting a number of principal components or factors that describe the prominent trends in data. A refined and more objective data analysis was obtained by WLS–PCA compared to PCA with variable selection. The limited human intervention required—and the extended amounts of chemical information that can be generated, analyzed, and evaluated—are the major and obvious strengths of the IMOF methodology.