Colin A. Stedmon (taken from Ph.D. thesis introduction chapter (2004)
EEMs
Another, more sensitive, technique often used for tracing DOM’s dynamics in marine and freshwaters is fluorescence. When irradiated by UV and blue light, a sub-fraction of the CDOM pool fluoresces. As with absorption measurements, the concentration and chemical composition of CDOM determines the intensity and shape of the fluorescence spectra. In early studies, CDOM fluorescence was used for examining the export of terrestrial organic matter via rivers into the coastal environment (Kalle, 1966). More recently studies have shown how different components of the DOM pool can be distinguished using detailed spectral fluorescence measurements (Coble et al., 1990; Coble, 1996). Over the last decade fluorescence excitation emission matrix (EEM) spectroscopy has been successfully applied for the identification of terrestrial, marine and anthropogenic components of DOM. EEM’s are obtained by combining fluorescence (emission) spectra measured from a series of different excitation wavelengths. Figure 5(a-d) shows examples of EEM’s collected from different study sites visited during this project. In general CDOM fluorescence has a broad excitation between 250 and 400 nm and a broad emission from 350 to 500 nm. The location of the excitation and emission peaks varies with composition of the fluorescent CDOM pool.
Figure 4. Example of the (i) measured, (ii) PARAFAC modelled and (iii) residual (difference between measured and modelled) EEM’s for four samples from the catchment of Horsens, in Denmark. a) Sample from a forest stream, b) sample from a stream draining agricultural areas, c) sample from the out flow of Horsens town waste water treatment plant, and d) sample from a station outside the mouth of Horsens Fjord taken at 1 m depth.
PARAFC modelling
EEM’s provide a wealth of information about CDOM, which in itself can be very difficult to interpret. Until recently, the techniques for characterising EEM’s have generally relied on visual identification of peaks and ratios of fluorescence in different regions of the spectrum (e.g. Coble, 1996; McKnight et al., 2000). A novel chemometric technique for interpreting DOM EEM’s can be applied (Stedmon et al 2003). The technique is called parallel factor analysis (PARAFAC) and has previously been successfully applied for data analysis in other research fields (e.g. pharmacology, food science and telecommunications), where complex mixtures of signals need to be separated (Bro, 1997). The application of PARAFAC to DOM EEM’s allowed us to model the EEM’s and decompose their complex nature into the different independently varying fluorescent DOM fractions. The ability of the PARAFAC modelling to replicate the fluorescence in the EEM’s is demonstrated in Figure 4, where examples of measured, modelled and residual EEM’s for a series of samples from the Horsens catchment, in Denmark, are shown. The fact there is a limited amount of information left in the residuals indicates that model is successfully describing most of the fluorescence signal.
Spectral decomposition
The dynamics of the different fractions identified in different aquatic environments can then be traced and used as a proxy for the changing characteristics of the DOM pool as a whole. Figure 5 shows an example of 8 different fractions that could be identified in the DOM samples from the Horsens catchment in Denmark (Stedmon & Markager 2005a). Components 1-6 exhibit fluorescence characteristics similar to that of humic material (both of allochthonous and autochthonous origin), and components 7 and 8 represent a protein-like fluorescence (autochthonous DOM).
Figure 5 The spectral characteristics of the eight components identified by PARAFAC modelling of EEM’s from Horsens catchment, Denmark (Stedmon & Markager 2005a).
Much can be gained from observing the spatial and temporal trends in the components. For example, Figure 6 shows how the relative fluorescence intensities of each component can be used as a “fingerprint” for the different sources of DOM, revealing considerable compositional and concentration differences.