Distribution et caractérisation de la matière organique dissoute dʹeaux naturelles de surfaceDistribution and characterization of dissolved organic matter of surface waters

The aim of this study was to characterize and compare the dissolved natural organic matter (DOM) of various river and reservoir waters. The determination of the “hydrophobic/hydrophilic DOM distribution” was used in parallel with more related drinking water parameters to compare the selected waters (i.e. chlorine and ozone reactivity). The “hydrophobic/hydrophilic DOM distribution” also called “humic/non-humic NOM distribution” of eight natural surface waters, four reservoir waters and four river waters, was determined by fractionating the organic matrix into three fractions onto two superposed XAD-8 and XAD-4 resin columns at acidic pH, the hydrophobic substances (i.e. humic substances) that adsorbed onto the XAD-8 resin, the hydrophilic acids that adsorbed onto the XAD-4 resin, and the non-adsorbed hydrophilic solutes that are contained into the XAD-4 effluent. The “hydrophobic/hydrophilic DOM distribution” was determined based on DOC or UV-254 nm measurements. The structural characterization of the NOM was completed by the determination of its biodegradable dissolved organic carbon (BDOC) content and its apparent molecular weight distribution using a 1000 dalton cutoff ultrafiltration membrane. Batch ozonation and chlorination were conducted on the various surface waters to compare their ozone demand and chlorine reactivity (chlorine demand, TTHM and TOX formation potentials). The waters collected from the reservoirs were found to be richer in DOC (6.8–7.8 mg l−1) and showed higher UV-254 nm absorbance values (0.22–0.29 cm−1) than the waters sampled from the rivers (2.8–5.3 mg l−1 of DOC, UV-254 nm absorbance ranged from 0.057 to 0.15 cm−1). Whatever the origin of the water, the higher the DOC, the higher the UV absorbance. However, a better correlation was obtained between these two parameters when considering the two groups of waters separately. The “hydrophobic/hydrophilic DOM distribution” showed that the DOC of both reservoir and river waters was evenly split between the hydrophobic and the hydrophilic fractions. Nevertheless, the hydrophobic fraction (i.e. humic substances) was found to be proportionally more abundant in reservoir waters (51–62% of the DOC) than in river waters (41–50% of DOC). This observation correlates well with the fact that the relative UV absorbance (also called specific UV absorbance: UV Abs.l/mg C) of the studied reservoir waters was generally higher than the relative UV absorbance of the studied river waters, since humic substances are known to be the most aromatic NOM fraction. On the contrary, data have also shown that the non-adsorbed hydrophilic fraction was significantly more abundant in rivers (23–26% of the DOC) than in reservoirs (14–21% of the DOC). One can note that the hydrophilic acids represent a pseudo constant fraction whatever the origin of the water (24–28% for seven of the eight surface waters). The presence of suspended sediments in rivers may explain the predominance of hydrophilic substances for this group of surface waters because of possible interactions (adsorption properties) between suspended materials and humic substances (hydrophobic organic fraction). The Loire river which corresponds to the largest river we have studied (rich in suspended sediments) was characterized by the lowest hydrophobic organic content, 40% of the DOC, a proportion that is similar to the ones previously observed by Semmens and Staples (1986) and Collins et al. (1986) for the Mississippi river and the Colorado river, respectively. Using a 1000 dalton ultrafiltration membrane no significant difference was observed between the apparent molecular weight distribution of the DOC of both river and reservoir waters. Organics above 1000 daltons generally accounted for up to 70–80% of the DOC. Results obtained with the Loire river were significantly different showing 50% of the DOC below 1000 daltons as compared to 20–30% for the other waters. These data are of particular interest because, according to the XAD-8/XAD-4 isolation procedure, the more hydrophilic the NOM, the lower its apparent molecular weight. One could note that the biodegradable dissolved organic carbon (BDOC) accounted for approximately 15% of the DOC of the studied waters. If one could expect that higher the hydrophilic organic content, the higher the BDOC content, no significant correlation was established between these two parameters. Reservoir waters were found to be more reactive with chlorine and ozone than river waters. The higher the UV-Abs or DOC of the water, the higher the ozone demand, chlorine demand, TOXFP, and THMFP. A reasonably good correlation was obtained between the ozone demand and chlorine demand. Regarding the DPB formation potentials, TTHMFP accounted for approximately 25% of TOXFP. Taking into account the DOC content of the different waters, the relative TTHMFP and TOXFP (μg Cl mg−1 C) were determined to be generally (with few exceptions) higher for the reservoir waters than for the river waters. However, no direct correlation between the hydrophobic organic content (i.e. humic substances content) and the DPB formation potentials could be established. Indeed, results of our previous works (Croue et al., 1993b) have shown that THMFP and TOXFP of hydrophilic acids and fulvic acids (90% of the humic substances) were similar.