Source Characterization of Polycyclic Aromatic Hydrocarbons by Using Their Molecular Indices: An Overview of Possibilities - Reviews of Environmental Contamination and Toxicology - page 78

Fig. 8 PCA ( principal component analysis ) allocates sediment PAHs to the contributing sources
by least-squares match of the PAH profile of the sample with the respective source. Adapted from
Burns et al. ( 1997 ), with permission, © SETAC
method (De Luca et al. 2004 ; Douglas et al. 2007b ; Neff et al. 2005 ). The first prin-
cipal component accounts for the largest possible variance in the data, and each
succeeding component explains as much of the remaining variability as possible
(Davis 2002 ). If principal components can be given a meaningful explanation (i.e.,
pyrogenic, petrogenic or natural sources), by assigning certain compounds to a cer-
tain principal component, then it is possible to quantify the contribution of the dif-
ferent PAH sources (Luo et al. 2008 ). In a cross-plot of principal components,
samples that have similar compositions will be close to each other (Fig. 8 ).
If there are doubts about the PAH source profiles, principal components can be
used to indicate the possible sources (Burns et al. 1997 ). If more is known about the
sources, PCA can be used to reveal the contributions of each source to the samples
(Boehm et al. 2001 ). Because the concentrations of individual PAHs often differ by
orders of magnitude in a given sample, they are usually normalized to the sum of the
analytes (Boehm et al. 2001 ).
Burns et al. ( 1997 ) used PCA to determine the major contributions of 36 identi-
fied sources in Prince William Sound, Alaska. The sources that contributed most
were then apportioned by a best-fit least square model to calculate a linear combina-
tion of contributing sources. Christensen et al. ( 2004 ) used externally normalized
PAH ratios as the loading variables in PCA, providing an integrated methodology
for oil spill identification similar to the tiered one presented in Fig. 6 .