Spectral fingerprinting and classification by location of origin of natural waters by multidimensional fluorescence

摘要

Multidimensional fluorescence is employed as an analytical tool for analyzing natural waters. Excitation-Emission Matrices (EEMs) are shown to contain spectral profiles of natural fluorophores as well as polyaromatic pollutants. A database of date-ordered EEMs was established to investigate fluorescence properties of several rivers and estuarial sites in the Boston area. Multiway Partial Least Squares Discriminant Analysis (NPLS-DA) regression models were constructed with calibration data from each sample site for classification of future test data to geographic origin. Parallel factor (PARAFAC) analysis resolved the pure component spectra for longitudinal and seasonal characterization studies. Time Resolved Excitation Emission Matrix (TREEM) spectroscopy exhibits extraction of further spectral information with the addition of a fluorescence decay time dimension. Location characterization of port waters by fluorescence fingerprinting is demonstrated. This spectroscopic technique shows promise as a regulatory tool for fingerprinting ships' ballast water to determine its harbor of origin. A deployable instrument for in situ analysis is proposed.