DAILY PAPER REVIEW

1117_Characterizing dissolved organic matter fluorescence with parallel factor analysis

 

 

ESEL Paper Review_20101117 by Aamir Alaud-din

Title and Author

Title: Characterizing dissolved organic matter fluorescence with parallel factor analysis: a tutorial

Authors: Colin A. Stedmon1, and Rasmus Bro2

1Department of Marine Ecology National Environmental Research Institute Aarhus University, Frederiksborgvej 399, Roskilde Denmark
2Dept. Food Science, Faculty of Life Sciences, University of Copenhagen, 30, DK-1958 Frederiksberg, Denmark

Summary of Paper
Fluorescence spectroscopy can be used to trace dissolved organic matter (DOM). Excitation-emission matrix (EEM) data can be utilized to trace DOM. Parallel factor analysis (PARAFAC) modeling is a useful tool for modeling of DOM. DOM is a mixture of several compounds. When this DOM is excited with ultraviolet or blue light, fluorophores present in DOM fluoresce. Location and peak of emitted light gives the type of fluorophore and intensity respectively. Since DOM is a mixture of several compounds, PARAFAC decomposes the complex signal of the mixture and helps in determining the types of fluorophores. Seasonal data of water, using PARAFAC, can be utilized to find the source of pollutants.

PARAFAC modeling is based on some assumptions like (1) a change in concentration will alter the height of the peak of fluorophore emitted signal instead of the shape and location of the peak and (2) inner filter effects are assumed to be minimal.

The model equation for PARAFAC analysis is as follows

x_ijk= ∑_(f=1)^F▒?a_if b_jf c_kf + E_ijk ?

where
x_ijk= one element of three-way data with emission wavelength j, excitation wavelength k for ith sample
a= intensity of the signal
b= emission spectra
c= excitation spectra
E= residuals

For PARAFAC modeling, pretreatment of instrument as well as of data are required. Emission correction spectrum and inner filter effects are the correction for instrument. Correction of data includes removal of Raman and Rayleigh peaks. It is achieved by inserting zeros below Rayleigh region and subtraction of pure water spectrum from the samples for Raman scatter removal. Outliers from the data are removed by leverage of samples.

Model validation is the most important part of the analysis. Residual analysis and split half analysis is used for this purpose. The same model is run on splitted data. If the results are the same, it means the model is valid.

Contribution to ESEL
PARAFAC modeling can be used for analysis of drinking water monitoring waste water treatment control and metal concentration change in a watershed. So, because of a wide range of applicability of the tool, it can be used for analysis in different situations.

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

첨부 (1)
ESEL Paper Review 20101117.pdf
30.0KB / Download 2