ESEL Paper Review_20110420 by Aamir Alaud-din

1.    Title and Author

Title: 2011_05_03_Estuarial Fingerprinting through Multidimensional Fluorescence and Multivariate Analysis
 
Authors: Gregory J. Hall, Kerin E. Clow, Jonathan E. Kenny

Department of Chemistry, Tufts University, 62 Talbot Avenue, Medford, Massachusetts 02155

2.    Summary of Paper

The paper presents estuarial fingerprinting for Mystic River Watershed region. The samples from Mystic River (R), Mystic Lake (L), Chelsea River (C), and Boston Harbor (H) were analyzed by fluorescence spectroscopy. Range of emission spectra was 220 ? 600nm in steps of 1nm, whereas, range of excitation spectra was also 220 ? 600nm in steps of 10nm to obtain 381 x 39 dimensioned excitation emission matrices (EEMs).

N-way partial least square (N-PLS) regression helped to develop calibration models with the ability of data classification. Combination of N-PLS with discriminant analysis (DA) helped in classifying data based on recognition of patterns. Parallel factor analysis (PARAFAC) was used to decompose this multi-variative data. 

Eight EEMs were used for the calibration from each location, L, R, C, and H. For the models with all four locations of sampling, six latent variables were used, whereas five latent variables were used for sites L and C.

Three factors were identified. Factor 1 for the locations L, C, R, and H, excitation/emission maximum wavelengths were 250/455nm. This corresponds to humic like substances. For sites C, L, and R, sharp peaks of emission spectra were observed at 350nm which correspond to tryptophan like substances being factor 2. For Boston Harbor, factor 2 was recognized to be marine humic like substance because of the excitation/emission values of 320/390nm. Factor 3 was identified as tyrosine like substance which is amino acid in biologically important proteins. This factor had the excitation/emission peaks of 221-280/306nm.

3.    Contribution to ESEL

This paper shows the identification of fluorophores in four different rivers, lakes, and harbor. It was done by just using N-PLS regression analysis which separated the samples from these sites for PARAFAC.

By: Aamir Alaud-din
aamiralauddin@gist.ac.kr