In 1819 Alexander Marcet proposed that seawater contains small amounts of all soluble substances and that the relative abundances of some of them were constant. This hypothesis is nowadays known as Marcet's Principle or the principle of constancy of the composition of seawater. Based on this principle, the present research tried to prove that it is possible to detect polluted seawater samples using the seawater H(+) affinity spectrum by the application of the possibilities provided by chemometric tools. Seawater samples were classified using the principal component analysis (PCA) of the HBound spectra of the samples. It was concluded that the sampling points location does not have any influence in the cluster formation, while the season in which they were collected is significant. On the other hand, the seawater composition was calibrated using estuary water samples of different salinities. Once the major constituents were measured, the data analysis concluded that it is possible to make a calibration of the HBound spectrum vs. any of these constituents by means of partial least square (PLS) regression. Thus, the experimental evidence collected in this work confirms that it is possible to detect polluted sea or estuary water samples using these chemometric tools and the H(+) affinity spectrum because with polluted samples these multivariate methods lead to incoherent results. So, suspect polluted zones may be monitored in a simple way with a low cost method and spending much less time.
Keywords: Chemometrics; Partial least square regression; Pollution; Principal component analysis; Proton affinity spectrum; Seawater.
© 2013 Elsevier B.V. All rights reserved.