EN:

Natural organic matter (NOM) in drinking water is a complex mixture of organic compounds. Some of the compounds are not biodegradable, while others are quickly biodegradable and a third group is more resistant to biodegradation. To have a better understanding of the biofiltration process in drinking water treatment, it is important to identify the elements of the quickly and slowly biodegradable NOM and to characterize the biodegradation rate of each element. In this study, an ozonated NOM solution was used as the substrate. The NOM was isolated from a groundwater in Germany using ion-exchange resins. The ozone dose was 0.35 mg O3/mg DOC (dissolved organic carbon). Previously bioacclimated sand was used as filter media and biomass source and was homogeneously distributed in the filter prior to each run. The substrate removal was evaluated by DOC, biodegradable DOC (BDOC), assimilable organic carbon (AOC), aldehyde and ketoacid analyses. When expressed in terms of the empty bed contact time (EBCT), the results showed that filter velocity in the range of 1.5 to 15 m/hr had no impact on substrate removal. This implies that substrate utilization, not external mass transfer, is the rate limiting step for substrate removal in drinking water biofilters. In this study, compounds or NOM fractions are termed quickly biodegradable if they are removed in the first three minutes of EBCT. 15% of the DOC was removed by the biofilter within three minutes of EBCT and was termed the quickly biodegradable fraction. The BDOC fraction of the ozonated solution was determined to be 40 to 45% of the DOC. In terms of BDOC, about one third of the total BDOC was quickly biodegradable. The AOC results show that about 90% of the total AOC was utilized by Spirillum sp. NOX (AOC-NOX). Most of the AOC was quickly biodegradable and was removed within one minute of EBCT. For aldehydes, glyoxal and methyl glyoxal were removed to below the detection limit after two minutes of EBCT. However, only 60% of formaldehyde removal was achieved in the first two minutes of EBCT, and no additional removal was achieved with increasing EBCT. Additionally, no significant removal of acetaldehyde was observed. The results of ketoacids show that their utilization rates were very high. More than 90% of glyoxylic acid and pyruvic acid were removed within one minute of EBCT.

FR:

Aujourd'hui, il est démontré que les traitements biologiques utilisés pour le traitement de l'eau potable sont des méthodes efficaces pour la réduction de la matière organique de l'eau, des sous-produits de la désinfection, ainsi que pour le contrôle de la croissance bactérienne dans les réseaux.

Quelques études récentes ont tenté de modéliser ce qui se passait sur les filtres biologiques utilisés pour le traitement de l'eau potable. Les modèles de biofilms développés pour les eaux usées sont souvent utilisés pour l'eau potable. Contrairement au biofilm des eaux usées, épais et dense, celui de l'eau potable est dispersè en fraction dans le milieu du filtre ainsi que sur sa pami. Le biofilm est donc très dispersé et les paramètres importants de modélisation du biofilm en eau usée (densité, épaisseur) ne sont pas appropriés pour l'estimation du biofilm dans les filtres d'eau potable.

La matière organique est constituée d'un ensemble complexe de composés organiques. Quelques-uns ne sont pas facilement biodégradables, d'autres le sont facilement et une autre fraction est récalcitrante. Dans ces conditions, un modèle utilisant un seul substrat pour représenter la cinétique de biodégradation n'est pas approprié. WANG et SUMMERS (1994) ont récemment développé un modèle mathématique utilisant plusieurs substrats pour décrire la réduction de la matière organique de l'eau dans les filtres biologiques. Le modèle suppose que la filtration biologique est composée de deux étapes : un transfert massique exteme suivi de l'utilisation du substrat à la surface du filtre. Dans ce modèle, la matière organique est divisée en trois portions: des composés facilement ou lentement biodégradables et récalcitrants. Des vitesses différentes de réaction sont utilisées pour chaque fraction. Les résultats démontrent que la diminution de la concentration de la matière organique est souvent limitée parle taux d'utilisation des substrats, et non pas par la vitesse du transfert exteme de la matière.

Les objectifs de cette recherche sont d'identifier les éléments de la matière organique ainsi que les composés lentement ou rapidement biodégradables et de suivre l'évolution de ces composés dans les filtres biologiques.

Une fois traitée par ozonisation, la matière organique est utilisée comme substrat dans cette recherche. La matière organique a été isolée par concentration sur résine échangeuse d'ions d'une eau souterraine prélevée en Allemagne. Le taux de traitements en ozone a été de 0,35 mg O3 par mg de carbone organique dissous (COD). Un sable bioacclimaté a été utilisé comme milieu filtrant et comme source de biomasse. Avant chaque expérience, il a été mélangé pour avoir une répartition homogène de la biomasse dans le filtre, La quantité de substrats était mesurée par les analyses du COD, du carbone organique dissous biodégradable (CODB), du carbone organique assimilable (COA), des aldéhydes et d'acides cétoniques. Les résultats, exprimés en terme du temps de contact théorique, démontrent que la vitesse de filtration dans le filtre (dans des proportions de 1,5 à 15 m/h) n'a pas d'influence sur la réduction des substrats. Le facteur limitant pour l'élimination du carbone organique par filtration est donc l'utilisation du substrat et non par le transfert externe de matière.

Dans cette recherche, la fraction de la matière organique est appelée « rapidement biodégradable » si elle est éliminée dans les trois premières minutes du temps de contact. Elle constitue 15 ib du COD. La fraction du CODB de l'eau traitée par ozonisation représente environ 40 à 45 % du COD.

Presqu'un tiers du CODB est rapidement biodégradé. Les résultats démontrent que 90 % du COA est utilisé par le Spirillum sp . NOX (COA-NOX), et que presque tout le COA est rapidement biodégradé et s'élimine par biofiltration en une minute de temps de contact. Le glyoxal et le méthyl glyoxal sont totalement éliminés après deux minutes de temps de contact. Par contre, seulement 60 % du formaldéhyde est éliminé dans les deux premières minutes de temps de contact et l'augmentation du temps de contact n'engendre pas une élimination supplémentaire. En outre, on n'observe pas de réduction significative de l'acétaldéhyde. Les résultats démontrent que les acides cétoniques sont rapidement et fortement dégradés. Plus de 90 % d'acide glyoxalique et d'acide pyruvique sont éliminés dans la première minute de temps de contact.

FR:

Depuis le milieu du siècle dernier, nous savons que le fer existe à son degré d'oxydation supérieur VI dans l'ion tétraoxoferrate(VI) FeO42-, isostructural de SO42-, CrO42-, MnO42-. Cette espèce mise en solution aqueuse oxyde l'eau en O2, et ce quelque soit le pH de cette solution. La vitesse de cette réaction est très grande en milieu acide, plus lente en milieu alcalin où elle conduit à la précipitation de Fe(OH)3. On peut donc envisager son emploi comme réactif polyfonctionnel : oxydant et floculant, dans le traitement de certaines eaux.

La première partie de ce travail décrit une voie originale de synthèse à l'état solide, à la température ambiante, du sel de formule K2(Fe,S)O4, appelé sulfatoferrate de potassium, qui contient l'entité FeO42-. En effet, jusqu'à ce jour seules des synthèses par voie aqueuse, à rendements très faibles, chimiques ou électrochimiques, ont été utilisées pour obtenir FeO42-.

Dans une seconde partie, nous avons mesuré les propriétés oxydantes et floculantes du sulfatoferrate de potassium agissant sur deux types d'eaux usées urbaines, chargées (MEST=258 mg.L-1, DCO=549 mg.L-1) et peu chargées (MEST=9 mg.L-1, DCO=37 mg.L- 1). Cette étude effectuée en collaboration avec le Centre International de l'Eau de Nancy (NANC.I.E.) nous a permis de préciser les conditions optimales d'emploi de K2(Fe,S)O4 et de comparer ses performances à celles de FeCl3.6H2O et Al2(SO4)3.18H2O.

Sur une eau chargée, l'abattement de la MEST est meilleur avec FeCl3.6H2O et Al2(SO4)3.18H2O, tandis que K2(Fe,S)O4 est plus efficace vis à vis de l'abattement de la DCO.

Sur une eau peu chargée, c'est l'effet bactéricide de K2(Fe,S)O4 qui est le plus remarquable; une dose de 10 mg.L-1 en fer (VI) anéantit 99,70% des coliformes totaux et 99,90% des coliformes fécaux.

EN:

Tetraoxoferrate(VI), FeO4²-, possesses properties which make it potentially useful in certain areas of water purification: it is an excellent oxidizing agent, it has a powerful bactericidal action and it spontaneously decomposes over a short period of time. When reduced, the FeO4²- ion generates base in solution and a Fe(OH)3 type gel which precipitates and carries down with it other ions (precipitation of hydroxide metal salts). These properties make ferrate(VI) useful in water disinfection and in wastewater treatment because it acts by an oxidation-coagulation-precipitation process.

Iron, in its familiar form exists in the (II) and (III) oxidation states; ferrate(VI) ion has long been known, but, due to its instability and difficulty of preparation, it has not been studied extensively and has not been industrially produced. In this paper, we recommend a method of synthesis of alkali ferrate(VI) salts by a dry powder process, which occurs at room temperature and can be easily used for industrial production of Fe(VI). To minimize ferrate(VI) decomposition, we proceed in a strong alkaline medium (presence of a strong base like potash or caustic soda) where oxidation of an iron (bivalent or trivalent) containing salt becomes easier. By such a synthesis process, we produce stabilized ferrate(VI) which enters a solid solution with the formula:

M2(Fe,X)O4

where M designates Na or K, X is an element whose cation has the electronic structure of a rare gas, e.g. X=S. M2XO4 is an isomorph of K2FeO4 with closed unit cell parameters. The M2(Fe,X)O4 formula has been established by chemical means and by X-ray diffraction; [sup]57Fe Mössbauer spectrometry has been used to monitor the hexavalent iron oxidation state. This paper deals with sulfatoferrate K2(Fe,S)O4 properties for compounds having Fe/S ratios between 1 and 1.5. in aqueous solution, K2(Fe,S)O4 dissolves and gives rise to FeO4²- and SO4²- anions.

In collaboration with NANC.I.E. (Centre International de l'Eau de Nancy), the effectiveness of K2(Fe,S)O4 in wastewater treatment has been studied. The product exhibits a real bactericidal effect on both coliforms and total bacteria. This bactericidal efficiency is reached in a short period of time with iron starting levels as low as 5-10 mg.L-¹. Comparative coagulation jar-tests were carried out on raw wastewater and on secondary effluent using FeCl3·6 H2O, Al2(SO4)3·18 H2O and K2(Fe[inf]0.54,S[inf]0.46)O4. Total suspended solids (TSS) removal on the raw wastewater was better with Al(III) and Fe(III) than with sulfatoferrate. On the secondary effluent, 40% removal was achieved with sulfatoferrate as well as with Fe(III). For removal of chemical oxygen demand (COD), better results were reached with Fe(VI) than with either Fe(III) or Al(III). The effectiveness of FeO4²- after 30 minutes of contact with the bacterial medium (secondary effluent) has also been studied. At pH=8.5, a dose of 10 mg.L-¹ iron(VI) removes 99.70% of the total coliforms and 99.90% of the fecal coliforms.

FR:

Pour étudier les incertitudes d'un modèle hydrologique, on peut employer l'analyse de fiabilité, méthode couramment utilisée dans le domaine de la résistance des structures et du génie hydraulique. Cet outil peut être transposé dans le domaine hydrologique pour juger de la qualité d'un modèle. Un modèle hydrologique sera fiable si, sur une série de données test, l'ajustement obtenu avec un jeu de paramètres recommandé pour un fonctionnement passe-partout est, selon toute probabilité, d'une qualité proche de l'ajustement idéal correspondant au calage du jeu de paramètres sur l'échantillon test.

Ce papier montre comment adapter chacun des concepts de l'analyse de fiabilité en hydrologie et détaille la technique des deux premiers moments afin de calculer explicitement la fiabilité d'un modèle-pluie débit en réalisant un développement limité au voisinage du point de fonctionnement du modèle.

Cette approche par analyse de fiabilité est appliquée au modèle GR3 à titre d'illustration pour juger de la pertinence de ce modèle pluie-débit conceptuel à trois paramètres en situation de prédiction des crues à court pas de temps sur un bassin versant situé dans la région de la Côte d'Azur en France .

Les résultats numériques obtenus montrent le caractère opérationnel de cette approche très simple. D'autre part, ces calculs de fiabilité mettent en évidence la réponse du modèle à chacun des paramètres.

Enfin en mesurant la performance de représentativité d'un modèle selon plusieurs dimensions telles l'erreur quadratique d'ajustement, l'erreur sur le volume de crue prévue ou l'importance de l'écart de pointe de crue, l'analyse de fiabilité peut être naturellement étendue vers une approche multicritère en considérant des probabilités conjointes de satisfaction du modèle sur chacun de ces critères.

EN:

Reliability analysis is a method for evaluating the risk of failure of a system and this concept is widely used in structure design. The theory can be conveniently transposed to the field of water sciences so as to evaluate the quality of hydrologic models. Hydrological reliability of a model is defined as the probability that there will not be significant different performances between this model, with parameters assessed on the basis of a large sample including various hydrological conditions, and the same model running with ad hoc parameters estimated from the test sample. Consequently a tolerance threshold has to be introduced in reliability analysis for hydrological models. A number of conceptual differences between validation and reliability analysis are reviewed and categorized.

This paper shows how to adapt the concepts of reliability to hydrology and gives details of the first two moment technique that can be implemented to compute the reliability of a rainfall runoff model. This technique is based on a Taylor series developed in the neighborhood of the model working point. This simplification allows for an explicit expression of the expectation and the variance of the hydrological model performance as functions of the marginal sensitivity of the criterion and the parameter uncertainties. A Student ratio can be easily computed to evaluate the quality of a hydrological model from a reliability-based point of view.

The GR3 model is used to illustrate this technique. A reliability analysis of this three parameter conceptual rainfall-runoff model gives new insight into the performance of the model for short-term flood prediction in the Réal Collobrier river basin located in the French Côte d'Azur. Numerical results show that the reliability analysis approach is very simple and of practical interest. For 16 of the 18 samples used for validation, the GR3 rainfall runoff model proved to be reliable on the basis of the three criteria used in this paper to evaluate the quality of a hydrological model (general quadratic deviation loss of runoffs, flood volume error and flood peak error).

Using reliability analysis, contributions of each parameter to the variance of the performance function can be conveniently identified. For the GR3 model and the Réal Collobrier case study, the impacts of each parameter on the model efficiency are as follows:

- The contribution of a parameter varies from one criterion to another. This indicates that a parameter originally designed to play a specific role in the modeling of the rainfall runoff transformation process does contribute to the criterion that best describes this function, which confirms the conceptual "good sense" of GR3.

- For each criterion, the cross contributions are very low: the GR3 model satisfies the principle of parsimony and each parameter adds something different to the general behavior of GR3.

In this case study, the GR3 model exhibits a good balance between complexity and quality for hydrological models.

Finally, reliability analysis can provide a multicriteria point of view of the model qualities by changing the various indices of performance. Extension of reliability analysis towards engineering risk would include the estimation of joint probability distributions for these multiple criteria of interest.

FR:

Le fonctionnement hydrologique de la plaine côtière ancienne de Guyane française constitue une des contraintes majeures à sa mise en valeur agricole, du fait de l'existence de périodes d'excès d'eau prolongées.

L'objectif de cet article est d'analyser, sur un interfluve caractéristique de la plaine côtière ancienne, la forme et la dynamique de la nappe et ses sources d'alimentation en relation avec la variabilité des propriétés physiques des sols.

Sur le plan expérimental, le travail est conduit à partir d'un suivi hydrologique in situ des fluctuations de la nappe observées sur 21 stations et d'une caractérisation de la variabilité spatiale de la conductivité hydraulique des sols par la méthode du trou de tarière à charge variable. Sur le plan de la modélisation, l'identification de la répartition spatiale de la recharge sur l'interfluve est déterminée par une modélisation inverse.

Les suivis montrent d'une part la rapidité de réponse de la nappe aux pluies et d'autre part des temps de présence de la nappe en surface variables selon les sols. La modélisation conforte l'hypothèse d'isolement hydraulique de l'interfluve. La dynamique de la nappe est directement reliée aux entrées pluviométriques et dans une moindre mesure aux sols. La variabilité spatiale de la recharge est par ailleurs sous la dépendance de la topographie et de l'hétérogénéité spatiale de la conductivité hydraulique. Enfin, une estimation du ruissellement de surface souligne son importance sur l'interfluve.

Les résultats obtenus montrent que l'engorgement des sols de la plaine côtière est sous la dépendance directe des processus hydrologiques observés à l'échelle de chaque interfluve. La prédiction des zones à excès d'eau marqué et des zones où la recharge est homogène peut être envisagée à partir de la connaissance de la topographie et de la distribution des sols. Sur le plan agronomique et pour les sols à forte contrainte hydrique, la mise en place de systèmes de drainage permettant leur mise en valeur agricole est à considérer.

EN:

The hydrological behaviour of the old coastal plain in French Guyana causes intense soil waterlogging, which is a major constraint to the agricultural development of this area. The old coastal plain presents a succession of similar old offshore bars (Fig. 1). To elucidate the factors affecting groundwater fluctuations in the plain, the hydrological behaviour of one typical bar was studied. This paper presents the results of the survey of the bar and of the identification of groundwater recharge by inverse modelling. It also investigates the relationships between the spatial variability of recharge rates, the topography, and the soil distribution over the bar to provide means for extrapolation to the whole plain.

A bar, covering 0.16 km², was chosen for the present study (Fig. 2). The maximum relief of the bar is 5 m and its general elevation lies between 2 and 7 m above sea level. The bar is drained by two convergent thalwegs. An important lateral and vertical soil differentiation was observed over a depth of approximately 1 m. Four main soil types (FAO/UNESCO classification) were identified at specific positions on the bar: podzols on the top, ferralsols on the upper slope, alluvial gleysols on the thalwegs and planosolic soils on the mid slope and between ferralsols and podzols. Ferralsols exhibit a progressive increase of clay content with increasing depth. The other soil types present sandy horizons with an irregular textural discontinuity (TD) located at a depth of 70 to 100 cm. Underneath the textural discontinuity lies a sandy clay marine alluvium, which is heterogeneous, ferrallitized and hydromorphic, with lenses of sand and clay. An impervious clay layer (NI) occurs at the base of the bar at an average height of 2 m above sea level. The climate is equatorial and has two marked seasons, wet and dry. Rain is mostly confined to the period November-July, but with a maximum from May to July. Mean annual rainfall is 2700 mm.

A network of 21 piezometers was set up at the various topographic and pedological situations (Fig. 3). Six sampling sites were also equipped with tensiometers and access tubes for neutron probes. Groundwater monitoring lasted for three years from 1983 to 1986. The variability of soil hydraulic conductivity over the bar was measured by the auger hole method at the intersections of a 50 m square grid and at 25 m away from a few such intersections. A geostatistical analysis was performed and kriged maps of hydraulic conductivity were produced (Figs. 4 and 5). The comparison between the kriged maps and the soil map indicates that ferralsols exhibit higher conductivities than the other soils.

Groundwater monitoring showed three main points. First, a fast response of groundwater fluctuations to rainfall was observed on the bar, which suggests that the hydrology of the bar is little influenced by contributions from neighbouring bars or from the Precambrian basement situated upgradient. Second, time length of soil saturation varied markedly over the bar and was related to the soil types (Fig. 7). Lastly, the observed spatial variability of the hydraulic conductivities and the evolution of water table levels indicate the possibility of a variable distribution of recharge over the bar.

The deterministic flow model used for this study, WATASI (WAter TAble SImulation, Wolsack, 1982) is based on a Darcy-Dupuit hydraulic schematization. It is an integrated finite element and multilayer groundwater model with square cells of variable size. Here, three layers were considered: one representing the topsoil, from surface to the textural discontinuity, the second representing the sandy clayey alluvium, lying over the impervious layer, and the third representing the thalwegs surrounding the interfluve (Fig. 10). All layers were divided into cells whose length was either 25 m or 50 m. For the purpose of recharge identification, according to the results of the survey, the cells were grouped in six zones of homogeneous slope and soil type, with each zone assumed to exhibit constant recharge. Calibration of the parameters of the model and identification of recharge over the six zones were conducted over three periods, one exhibiting steady state flow and the two others transient flow, by minimizing the difference between the measured and simulated hydraulic heads. The results obtained by the simulation approach are:

- No assumption of lateral inflow is necessary to simulate properly the evolution of hydraulic heads (Figs. 11 and 13); thus groundwater recharge only originates from the seepage of rainfall through the soil cover of the bar.

- The spatial variability of recharge appears to follow firstly the slope distribution, and secondly the soil distribution: zones of high recharge correspond to zones of limited waterlogging and vice versa (Figs. 12 and 14).

- The estimated rates of recharge are small in comparison to rainfall, and soil water budget calculations demonstrate the existence of large runoff rates.

It can be concluded that the waterlogging of soils on the old coastal plain is mainly caused by the hydrological processes at the scale of each bar. Thus, for improving the agricultural suitability of the soils on the plain, local drainage of the waterlogged soils should be sufficient. To predict over the plain which zones should be drained, information on topography and soil distribution can be used as there are good correlations between the variability of these parameters and the variability in groundwater recharge and water table depths.

FR:

Il est admis aujourd'hui que la génotoxicité identifiée dans les extraits d'eau potable provient principalement de l'action du chlore sur la matière organique naturelle qui donne naissance à des dérivés organohalogénés.

Dans le présent travail, nous avons comparé la sensibilité de trois essais de génotoxicité (SOS chromotest, test d'Ames-fluctuation et test micronoyau triton) lors de l'étude de composés organohalogénés, d'acides fulviques chlorés et d'échantillons d'eau (non concentrés) en cours de traitement de potabilisation.

Les composés organohalogénés étudiés sont 4 trihalométhanes, 5 acétonitriles et 5 chloropropanones identifiés dans l'eau potable ou dans des solutions de substances humiques chlorées. Les résultats obtenus révèlent que le SOS chromotest est globalement le moins sensible des trois essais et que le test d'Ames-fluctuation et le test micronoyau triton permettent généralement de détecter les plus faibles concentrations de composés génotoxiques. Les essais ont également permis de démontrer que la nature des substituants halogénés (brome ou chlore), le nombre et la position des atomes de chlore influencent notablement la génotoxicité des composés organohalogénés.

Toutefois, les résultats obtenus indiquent qu'aucun des trois tests réalisés n'est suffisant à lui seul pour détecter l'ensemble des produits génotoxiques. Ces observations confirment la nécessité de réaliser une batterie de tests qui mette en oeuvre divers types cellulaires et différents systèmes de métabolisation, et détecte divers évènements de génotoxicité.

Les travaux portant sur les solutions concentrées d'acides fulviques chlorés montrent l'intérêt des essais sur bactéries (particulièrement le test d'Ames-fluctuation) pour la détection rapide de l'activité génotoxique de ces solutions.

L'étude concemant les échantillons d'eau prélevés à différents niveaux d'une station de potabilisation, et analysés sans concentration préalable, indique que le test d'Ames- fluctuation est le seul capable de détecter une activité génotoxique dans les échantillons non concentrés étudiés. On montre, conformément à la littérature, que l'activité mutagène observée résulte de la chloration de l'eau.

EN:

Since the identification of organohalides in drinking water in 1974, several investigators have detected genotoxic activity in drinking water concentrates. It is now widely admitted that the observed genotoxicity originates mainly from the reaction of chlorine on natural organic matter contained in the raw water, which leads to the formation of organohalogenated compounds.

The aim of this study is to show the benefit of three short-term assays for the evaluation of the genotoxic potency of organohalogenated compounds and of complex mixtures. In a wider context, the purpose is to identify a test or a battery of tests that can contribute to the control of natural and drinking water genotoxicity.

The three genotoxicity assays carried out during this work were:

- the SOS chromotest, a primary DNA damage in vitro assay on Escherichia coli;

- the Ames- fluctuation test, a point mutation in vitro assay on Salmonella typhimurium;

- and the newt micronucleus test, a chromosomal aberration in vivo assay on the amphibian Pleurodeles waltl.

These assays display a valuable advantage: the water samples under study can be analyzed without concentration prior to testing. Thus, the different concentration procedures, which may modify the original genotoxicity of the water samples, are avoided.

A previous study on seven reference genotoxic chemicals had indicated that the SOS chromotest was never the most sensitive of the three tests (for a given chemical, the most sensitive assay is defined as the test which detects the lowest concentration inducing a significant genotoxic effect). On the contrary, the Ames-fluctuation test proved to be the most sensitive for compounds showing direct genotoxic activity, and the newt micronucleus test the most sensitive for chemicals with indirect genotoxic effects. None of the assays was the most sensitive for every substance analyzed. These observations suggested the need to implement a battery of tests using several cell types, different metabolization systems and detecting several genotoxicity events. This earlier study also showed, in accord with several results in the literature, that the Ames-fluctuation test (in liquid medium) demonstrated a better sensitivity than the Ames test (in agar solid medium).

The first part of the present study involved testing the genotoxicity of 14 organo- halogenated compounds identified in drinking water samples or in chlorinated humic matter samples. The chemicals studied were four chlorinated and/or brominated trihalomethanes (trichloro-, bromodichloro-, chlorodibromo- and tribromomethane), five chlorinated or brominated acetonitriles with one, two or three halogens (monochloro-, dichloro-, trichloro-, monobromo- and dibromoacetonitrile) and five chlorinated propanones with one, two or three substitutions on one or two carbon atoms (monochloro-, 1,1-dichloro-, 1,3-dichloro-, 1,1,1-trichloro- and 1,1,3- trichloropropanone). Although the SOS chromotest was the most sensitive for 3 of the 14 substances analyzed, the results confirmed that this test was globally the least sensitive; the Ames-fluctuation test and the newt micronucleus test remained the most efficient assays. It is interesting to note that the Ames-fluctuation test appeared the most sensitive for all the chloropropanones tested and the newt micronucleus test, for all the haloacetonitriles analyzed. Moreover, several structure-activity relationships were demonstrated: the nature of the halogenated substituents (bromine or chlorine), the number and, above all, the position of chlorine atoms strongly influenced the genotoxicity of the organohalides studied.

In the second part of the work we analyzed the effects of complex mixtures containing several organohalogenated compounds: the three tests were performed on two chlorinated fulvic acids of different origin. Pornic fulvic acid was extracted from a surface water reservoir used to produce drinking water in Vendée (France) and Pinail fulvic acid came from a forest pond near Poitiers (France). The total organic carbon was about 1 g/l in the solution subjected to chlorination and the molar chlorination ratio was 1.5 Cl2/C. The results showed the advantage of tests using bacteria: the Ames fluctuation test was the only assay able to detect the genotoxicity of both chlorinated fulvic acids; the SOS chromotest detected the genotoxic effect of only one of the chlorinated fulvic acids (Pinail). In contrast, the newt micronucleus test did not show any genotoxicity of the chlorinated fulvic acids. However, it must be pointed out that, as insufficient fulvic acid was available, the genotoxic potency of these solutions on the newt was not tested under adequate conditions (e.g., subchronic concentrations were not studied). Nevertheless, the concentrations of fulvic acid analyzed were very close to those found in the aquatic environment.

The last part of the study attempted to approximate environmental and human exposure conditions: the three tests were performed on four water samples taken at several stages of a drinking water treatment plant. These samples were analyzed for genotoxicity in the three test systems without preconcentration. The plant studied is characterized by the following treatment steps:

- coagulation-flocculation;

- chlorination at 6 g Cl2/m³;

- sand filtration;

- ozonation at 1.9 g O3/m³;

- final chlorination at 1.4 g Cl2/m³ before sending the treated water into the distribution system.

The four samples were taken:

1. before any treatment (raw water, 10 to 13 mg total organic carbon per liter),

2. between the sand layer filter and the ozonator (chlorinated water),

3. after the ozonator (ozonated water),

4. and at the end of the treatment process (treated water).

The results obtained confirmed the advantage of the Ames fluctuation test, which was the only assay able to detect a genotoxic activity in the unconcentrated water samples studied. Regarding the influence of the different chemical treatments on the mutagenicity observed, it was demonstrated that the first chlorination step led to the formation of direct-acting mutagens. The treatment with ozone, at the rate used, did not significantly modify the mutagenicity of the samples that had previously been chlorinated. Similarly, the second chlorination step did not significantly increase the direct mutagenicity detected.

Practically, our study indicated that the Ames-fluctuation test is the only assay, among the three performed that is able to contribute efficiently to the control of drinking water genotoxicity. In this context, the benefit of the SOS chromotest appears only in case of accidental pollution: indeed, it is the only test able to yield results within 24 hours. The implementation of the newt micronucleus test could be useful for the control of natural or drinking water genotoxicity in case of extensive pollution.

FR:

L'Indice de Polluosensibilité Spécifique (IPS) est considéré comme l'un des indices diatomiques les plus performants pour l'évaluation de la qualité des cours d'eau. Son utilisation en réseau de surveillance reste cependant limitée en raison de la nécessité de travailler au niveau spécifique voire infraspécifique et de la systématique en perpétuelle évolution. A l'opposé, l'Indice Diatomique Générique (IDG) est plus accessible dans sa mise en oeuvre mais ne permet pas d'obtenir des résultats très fiables. Un nouvel Indice Diatomique Pratique (IDP) a donc été mis au point sur un bassin versant expérimental à partir d'un jeu de 86 relevés. Dans un premier temps, les inventaires ont été classés en fonction des écarts observés entre IPS et IDG. Dans un second temps, ont été identifiées les espèces responsables de ces écarts en prenant en compte celles présentant une abondance relative supérieure à 5 % et une différence de polluosensibilité avec le genre correspondant supérieure ou égale à 0,4. Plusieurs IDP ont été mis au point et leurs performances, par rapport à l'IPS, étudiées. Il apparaît que la prise en compte des espèces responsables des écarts supérieurs ou égaux à 2 constitue le meilleur compromis entre fiabilité et applicabilité en réseau. Cette méthodologie a été appliquée aux 480 relevés effectués dans le bassin Artois - Picardie et aux 550 espèces inventoriées. Elle permet de proposer un indice diatomique pratique basé sur l'identification de 45 genres et 91 espèces.

EN:

Macroinvertebrates constitute the main biological support for an evaluation of the quality of water courses and are, therefore, widely put to use in monitoring networks. However, for major water courses and canalized waterways the use of other methodologies is imperative. Diatoms and diatom indices are well adapted to the study of these environments. Among these, the Specific Polluosensitivity Index (SPI) established by CEMAGREF seems to be one of the better performing diatom indices. Calculation of this index relies on the Zelinka & Marvan formula derived from the saprobic system: SPI=[Epsilon]A[inf]j v[inf]j i[inf]j / [Epsilon] A[inf]j v[inf]j where A[inf]j is the relative abundance of the species j, v j is its indicative value ( 1 [smaller or equal] v[inf]j [smaller or equal] 3) and i[inf]j its pollution sensitivity (1 [smaller or equal] i[inf]j [smaller or equal] 5). The values initially falling in the range between 1 and 5 are transformed into values comprised between 1 and 20, in order to make comparisons between the various existing indices easier. Five categories of water quality can be distinguished according to the value of the index: SPI [Bigger or equal] 16: zero pollution or low eutrophication; 13.5 [smaller or equal] SPI < 16: moderate eutrophication; 11 [smaller or equal] SPI < 13.5: moderate pollution or heavy eutrophication; 7 [smaller or equal] SPI < 11: high pollution; SPI < 7 : very heavy pollution. However, the SPI index is rarely used because of two main obstacles: it requires data at a specific or even infraspecific level, and it is based on constantly changing systematics. Progress towards increased accessibility and, therefore, larger application was made with the elaboration of the Generic Diatomic Index (IDG) based on the same principle as the SPI. However, this GDI does not yield reliable results, in so far as certain genera, such as Navicula and Nitzschia, contain species with a widely differing ecologies. In order to provide a methodology that can be used as a matter of routine, a protocol for the elaboration of a Practical Diatomic Index (PDI) was established and tested on 86 inventories from the water basin of the river Aa (North of France). These were first classified into four categories according to the variations observed between SPI and GDI: category 1: |SPI-GDI| [bigger or equal] 3 ; category 2: 2 [smaller or equal] |SPI-GDI|; category 3: 1 [smaller or equal] |SPI-DGI| < 2 ; category 4: |SPI-DGI| < 1. For each of the first three categories, the species responsible for the variations were identified, taking into consideration those with a relative abundance of more than 5%, the pollution sensitivity of which showed, compared to the corresponding genus, a variation higher than or equal to 0.4. Thus, three indices corresponding respectively to category 1 (PDI1), 2 (PDI2), and 3 (PDI3) were proposed and tested against the SPI taken as reference index. The results of this comparative study can be summarized as follows:

- GDI=0.57 SPI + 5.47 r=0.801 (242 species),

- PDI1=0.86 SPI + 1.12 r=0.972 ( 27 species),

- PDI2=0.95 SPI + 0.55 r=0.991 ( 39 species),

- PDI3=0.96 SPI + 0.45 r=0.994 ( 42 species).

To test the implications of replacing the presently used SPI by this practical index, a comparative study of the classification of inventories in four categories of hydrobiological quality was also carried out. This study shows that the mean, at - 1.76 ± 2.25 for the GDI, is reduced to 0.14 ± 0.94 for PDI1, to - 0.07 ± 0.51 for PDI2, and to - 0.07 ± 0.45 for PDI3. Given the variability of the index at one and the same site and in one sampling, PDI2 considered to be the best compromise between reliability and network applicability. The methodology corresponding to PDI2 was applied to the 480 samplings carried out in the Artois-Picardie basin and a new Practical Diatom Index is thus proposed for the monitoring of the 200 sites making up the monitoring network of the Artois-Picardie water basin. This PDI, built on a base of more than 550 species and varieties, rests on the joint determination of 45 genera and 91 species of which the pollution sensitivity coefficients and the indicative values are given.

FR:

Le suivi de la matière organique totale (COD) et spécifique (substances humiques et principalement acides aminés dissous totaux) dans les eaux d'une filière de production d'eau potable a mis en évidence des abattements moyens dans l'eau traitée représentant 40 % du COD et 0 à 75 % des acides aminés. Parallèlement, le potentiel de demande en chlore à long terme est diminué de près de 70 %, alors que la fraction biodégradable connait une élimination de l'ordre de 50 %.

Aucune corrélation directe n'a pu être mise en évidence entre les teneurs en acides aminés et la valeur du carbone organique dissous biodégradable (CODB) et de la demande en chlore des eaux brutes ou en cours de traitement. Cependant, compte tenu des teneurs en acides aminés dissous totaux quantifiées dans les eaux traitées de l'usine, leur participation peut être estimée entre 5 et 23 % du potentiel de demande en chlore et entre 5 et 25 % du CODB.

La matière organique analysée spécifiquement dans le cadre de cette étude (substances humiques et acides aminés dissous totaux), représente 20 à 35 % du COD des eaux en fin de filière, et la présence de ces composés organiques serait responsable d'une part notable de la demande en chlore à long terme des eaux (40 à 60 %).

EN:

Dissolved organic compounds (dissolved organic carbon (DOC), biodegradable dissolved organic carbon (BDOC), total amino acids and humic substances) and their chlorine demand were analyzed monthly at different steps of the water treatment plant of Méry-sur-Oise (Paris, France, 270 000 m³/d, Fig. 1).

Total dissolved amino acids were determined by HPLC analysis with fluorimetric detection after hydrolysis of combined amino acids (proteins, polypeptides) by hydrochloric acid (DOSSIER BERNE et al., 1994 a); the separation of 17 amino acids was performed after orthophthaldialdehyde (OPA) pre-column derivatization. Humic substances were determined according to the method developed by THURMAN and MALCOLM (1981), by gravity feeding XAD-8 resins with acidified samples. The humic fraction was expressed as the difference between DOC before and after passage through the column. The method developped by JORET and LÉVI (1986) using a biologically active sand inoculum was used for the determination of BDOC.

For the determination of chlorine consumption, the kinetic model described by JADAS- HÉCART et al. (1992) was used; it takes into account the long-term chlorine demand in terms of potential demand. This method was automated (DOSSIER BERNE et al., 1994 b) and computer-assisted. The chlorine dose was chosen depending on the DOC value and on the N-NH4+ content of the water; the applied dose was: 3 mg Cl2/mg DOC + 10 mg Cl2/mg N-NH4+. Chlorine determination was performed automatically by the spectrophotometric N, N-diethylphenylene-1,4-diamine (DPD) method.

In raw water, DOC values may reach 5.6 to 6.5 mg C l-¹ during the cold season, but the average yield of elimination was generally close to 40 % (Fig. 2). The biodegradable fraction of the organic carbon (BDOC), which represents 25 to 50 % of the DOC in the raw water, was partially removed in the plant and the residual concentration in treated water varied between 0.4 to 1.8 mg C l-¹ (Fig. 2). A transitory increase in the BDOC values was generally observed during the ozonation step; it reached 0.2 to 0.5 mg C l-¹ (Figs. 6 and 7).

Whatever the period of the year, extracted humic substances constituted about 50 % of the DOC found in raw water (Fig. 3). This hydrophobic fraction was significantly removed in the treatment plant reducing the proportion of humic substances in the DOC of treated water to 16 - 23 % (Figs. 3, 6 and 7).

The concentrations of total dissolved amino acids ranged from 100 to 260 µg l-¹ C in raw water and from 50 to 150 µg l-¹ C in produced water (Fig. 4); the main part of this elimination occurred during the clarification step (Figs. 6 and 7). No important seasonal variations could be observed for chlorine demand (Fig. 5); its removal occurred at each step of treatment and the average global elimination by the plant was of the order of 70 % (Figs. 6 and 7).

On account of the low concentrations of amino acids, no direct relation could be shown between amino acid concentrations and the respective values of BDOC or of chlorine demand potential (Figs. 9 and 10). With regards to BDOC and chlorine demand potential, no correlation could be shown between these two parameters either (Fig. 8). However, it was possible to calculate the contribution of specific classes of dissolved organic compounds to chlorine demand and to BDOC. This calculation is based: i) on the concentrations of humic substances and amino acids measured during the monthly experiments at each point of the treatment plant, ii) on bibliographic data concerning the contribution to BDOC (biodegradability) and chlorine consumption of a large variety of model molecules (free or combined amino acids, aquatic humic substances).

The results obtained in this way are reported in Table 3 for the treated water of the M?ry sur Oise plant. These results indicate that the small amounts of total dissolved amino acids present in treated water may account for 5 to 25 % of the BDOC value and for 5 to 23 % of the total chlorine demand potential. With regard to humic substances, their biodegradability was assumed to be very low but their contribution to the chlorine demand of the treated water was estimated between 10 to 40 %, whereas their DOC contribution ranges from 16 to 35 %. As a consequence, both amino acids and humic substances could account for 40 to 60 % to the chlorine demand of treated water. A special point should be noticed for total amino acids: their contribution to the DOC values of treated water was only 2 to 7 %, but they may account for a larger proportion of BDOC or chlorine demand potential (5 to 25 %).