Abstract

It is still difficult to accurately determine the concentrations of polycyclic aromatic hydrocarbons in water because of their low solubility during in-situ monitoring water quality. In this work, two methods including three-way parallel factor analysis (PARAFAC) and multi-way partial least squares (N-PLS) were used to quantify the concentrations of polycyclic aromatic hydrocarbons (PAHs) in reservoir water and river water. The results demonstrated that the combination of N-PLS with residual bilinearization (RBL) can achieve second-order advantage, thereby, predict the concentrations of phenanthrene, pyrene and anthracene as good as, or predict fluorine much better than, that by using PARAFAC. Combining PARAFAC with N-PLS/RBL may be an effective way to determining PAHs correctly in real-time and in-situ monitoring water quality.

© 2015 Optical Society of America