FR :

Quatre bioessais in vitro ont été réalisés pour évaluer la réponse des communautés phytoplanctoniques naturelles à différentes concentrations de zinc (2,5; 10; 20 - 25 et 40 - 50 mg.l^-1). Les essais réalisés au cours des quatre saisons , ont été menés en conditions contrôlées pendant un mois avec des échantillonnages tous les 2-3 jours.

La réponse algale fut variable selon la saison et dépendante des concentrations en Zinc, des espèces algales, de la densité de l' inoculum et de la température. 2,5 et 10 mg Zinc.l^-1 dans le milieu de culture ont stimulé la croissance de certaines diatomées; à une concentration de Zinc de 20-25 mg.l^-1 Chlorella vulgaris (Chlorophyceae) s'est particulièrement développée, et, dans des systèmes contenant 40-50 mg Zinc.l^-1 une mortalité importante a été généralement trouvée, à l'exception de Chlorella vulgaris et de quelques espèces de diatomées qui se sont montrées tolérantes au zinc: Cyclotella meneghiniana, Gomphonema parvulum, Navicula sp., Nitzschia palea, Nitzschia sp., Pinnularia biceps, Synedra acus et Synedra ulna var. amphirrhynchus.

Les Cyanophyceae, les Euglenophyceae, les Tribophyceae, les Chrysophyceae, les Zygophyceae et les Dinophyceae ont été particulièrement sensibles. En présence d' inoculums abondants, la sensibilité au Zinc a été observée à partir de 25 mg.l^-1 (automne-20°C, printemps-20°C et été-25°C); par contre, en présence d'inoculums peu denses, elle est apparue dès 10 mg.l^-1 (hiver-15°C).

De manière générale, à mesure que la concentration de Zinc augmentait, Chlorella vulgaris Biej. est devenue graduellement, le taxon dominant; ce qui a eu pour conséquence une diminution de la diversité du système algal testé. Des communautés plus simples se sont alors développées, dominées presque toutes par des espèces tolérantes au Zinc.

EN :

In order to study the response of native phytoplanktonic communities to different zinc concentrations (2.5, 10, 20 - 25 and 40 - 50 mg.l^-1), four bioassays in vitro were carried out. They were conducted in autumn, winter, spring and summer under controlled conditions during a month and the samplings were performed each 2-3 days.

The algal responses were variable according to the season and zinc-concentration, species, inoculum density and temperature dependent. 2.5 and 10 mg.l^-1 of Zinc in the test medium caused a stimulatory effect on the growth rate of certain diatoms; with 20-25 mg.l^-1 of Zinc the development of Chlorella vulgaris (Chlorophyceae) was stimulated, and with 40-50 mg.l^-1 of Zinc it was generally found important mortality rates except for Chlorella vulgaris and for some tolerant diatom species: Cyclotella meneghiniana, Gomphonema parvulum, Navicula sp., Nitzschia palea, Nitzschia sp., Pinnularia biceps, Synedra acus and Synedra ulna var. amphirrhynchus.

The Cyanophyceae, Euglenophyceae, Tribophyceae, Chrysophyceae, Zygophyceae and Dinophyceae were particularly sensitive to Zinc. In presence of abundant inoculums, the sensivity to Zinc was observed up to 25 mg.l^-1 (autumn-20°C, spring-20°C and summer-25°C) while in presence of smaller inoculums the sensivity manifested up to 10 mg.l^-1 (winter-15°C). In general, as Zinc concentration increased, Chlorella vulgaris gradually became the dominant taxon and thus, a decrease in the specific diversity of the assayed algal system made of Zinc-tolerant species was observed.

FR :

Les stations d'épuration d'eaux résiduaires sont une des étapes du cycle du sélénium dans l'environnement et contribuent à sa redistribution dans le milieu naturel. Très peu étudié jusqu'à présent dans ces milieux, le sélénium n'en est pas moins un élément très important du point de vue écotoxicologique, sa teneur dans les boues de stations d'épuration destinées à l'épandage agricole faisant par ailleurs l'objet d'une norme.

Nous avons mis au point des techniques permettant la détermination spécifique de l'élément total dans ce type d'échantillon, par minéralisation classique ou assistée par micro-ondes et dosage par Voltamétrie de Redissolution Cathodique Différentielle Pulsée (DPCSV) et Spectrométrie d'Absorption Atomique ElectroThermique (ETAAS). Le contrôle qualité a été effectué sur deux échantillons certifiés fournis par le Bureau Communautaire de Référence (BCR) : la boue CRM 145 R et la boue CRM 007.

Cependant, lorsqu'on parle de risque toxicologique, il est important de s'intéresser à la détermination des différentes formes sous lesquelles cet élément peut être présent. Nous avons pour cela réalisé des extractions parallèles (spéciation de phases) du sélénium contenu dans les boues afin de déterminer quel pourcentage du sélénium total est réellement et potentiellement disponible pour les végétaux lors d'un épandage sur sol agricole. La spéciation d'espèces a été brièvement abordée dans le but de déterminer les teneurs en Se(IV) et Se(VI), espèces les plus toxiques.

EN :

The great effort undertaken for about twenty years to improve the quality of surface waters has led to the construction of numerous waste water treatment plants, generating an increasing amount of sludge. Waste water and sludge treatment processes represent an important point in the hydrological cycle at which the disposal of substantial quantities of trace elements to the environment may be regulated. From the law on waste recovery and disposal in 1975 to the European guideline about wastes in 1991, the priority has been given to waste recovery and recycling. With increasing pressure to ban all sludge dumping at sea, and considering the prohibitive costs of land-filling and incineration, there is a great tendency to dispose of sludge on land (40% in 1988 to 60% in 1992).

Although numerous studies have demonstrated the intrinsic value of sludge for soil amendment, given its nitrogen, phosphorus and homogeneous organic matter content, evidence has accumulated in recent years that numerous environmental problems can arise because of the presence in sludges of high amounts of certain trace elements (potentially toxic to plants and to human beings and liable to be concentrated along the food chain), among which selenium is particularly interesting.

Selenium presents a complex case, as it is also an essential element for living organisms (including humans). The amendment with sewage sludge is sometimes used to increase the selenium content in crops, and afterwards in cattle, when there is a proven lack of this element in a given place. Nevertheless the boundary between essentiality and toxicity is relatively narrow and is expressed at trace levels. It is thus particularly important to survey the selenium concentrations encountered in sewage sludge, especially as guidelines and regulations concerning these data will probably be strengthened.

Presently, in France, sludge must not contain more than 200 mg Se·kg^-1 dry weight and must not be used on soils containing more than 10 mg Se·kg^-1 dry weight (AFNOR U 44-041 norm). This norm concerns only the total amount of selenium contained in sludge and does not take into account the different species (organic and inorganic Se(-II), Se(0), Se(IV) and Se(VI)) that could be present.

First of all we had to develop methods for the classical and microwave-assisted wet digestion of sewage sludge, and the determination of their total selenium concentration by Differential Pulse Cathodic Stripping Voltammetry (DPCSV) and ElectroThermal Atomic Absorption Spectrometry (ETAAS). Quality assurance involved the analysis of two BCR (Community Bureau of Reference) certified sewage sludge reference materials (CRM 145 R and CRM 007) and the different techniques were then applied to natural samples from a representative French sewage treatment plant located in the city of Tarbes (South-West of France).

The mixture HNO₃-H₂O₂-H₂O led to the best results for the digestion and analysis of certified samples, caused few problems for the analysis by DPCSV and ETAAS, and was therefore retained. The decrease of the digestion duration obtained by the use of microwaves was particularly interesting (from one to three days on a hot plate to less than one hour by the Microdigest 301 (PROLABO, France)), and reproducibility was also acceptable (between 3 and 10%). Concentrations obtained for the sewage sludge from the Tarbes treatment plant were very much lower than those for NF U 44-041: 1.08±0.11 mg Se·kg^-1 dry weight.

However knowledge of speciation, that is to say the determination of the different physicochemical forms of selenium present in a given medium, is necessary when speaking of the toxicological risk represented by an element. The mobility of selenium and its toxicity to the biosphere are related to its association with various sludge or soil constituents as well as to its total concentration. "Soft" or partial extraction techniques are necessary when the aim of the study to determine trace element speciation. The extractants used must separate selenium from the matrix without inducing any loss or change in the partitioning of individual chemical species. In parallel extractions the mechanisms involved for each extractant must correspond to processes occurring in nature and are then associated with special fractions of selenium: soluble, exchangeable, "oxidizable", and "mineral" fractions .

Parallel extractions with three types of extractants were chosen for this study and applied first to CRM 007: warm water (soluble fraction), ammonium phosphate-citric acid (soluble + exchangeable fraction) and sodium hydroxide (soluble + exchangeable + "oxidizable" fraction). The soluble, exchangeable, "oxidizable" and "mineral" fractions represent respectively : 11%, 14%, 39% and 36%. The same procedure was then applied to natural samples from Tarbes giving the following results: 36% soluble, 22% exchangeable, 42% "oxidizable". The sodium hydroxide extraction procedure allowed us to extract the entire Se content of this sludge (1.07±0.03 mg Se·kg^-1 dry weight), showing that all the selenium present is potentially available after agricultural land application. It was then possible in this fraction to deal with the species speciation of selenium by the mean of a separation of inorganic and organic species on an Amberlite CG-400 resin and a specific analysis by DPCSV. Se(IV) and Se(VI) represent respectively between 30 and 40% and between 2 and 20% of total selenium in the sludges from Tarbes.

FR :

Plusieurs méthodes existent pour calculer le débit réservé d'un cours d'eau. Dans la présente étude deux approches sont analysées, soit les approches par méthodes hydrologiques et hydrobiologiques. Cinq méthodes hydrologiques d'évaluation du débit réservé ont été appliquées au ruisseau Catamaran au Nouveau-Brunswick (Canada), ainsi qu'une méthode hydrobiologique. Parmi les méthodes hydrologiques, on retrouve la méthode de Tennant, celle de 25% du débit moyen annuel (DMA), la méthode du débit médian (Q50) mensuel, 90% du débit classé (Q₉₀) et la méthode basée sur l'analyse statistique de fréquence des débits faibles (7Q10). La méthode hydrobiologique utilisée dans la présente étude fut l'application du modèle PHABSIM pour le saumon Atlantique juvénile. Ce modèle est calibré en utilisant les données de vitesse d'écoulement (V), profondeur d'eau (D) et grosseur du substrat (S) pour trois différents débits. L'application des méthodes hydrologiques a démontré que certaines méthodes telle que la méthode Tennant, 25% DMA et la méthode du débit médian, donnent des résultats similaires surtout en période d'étiage. D'autre part, deux méthodes en particulier, soit la méthode de 90% du débit classé et celle basée sur une analyse statistique des débits faibles prédisent un débit réservé très faible en période d'étiage.

Une modélisation de l'habitat physique du ruisseau Catamaran démontre que l'habitat disponible maximal se trouve généralement aux environs du débit moyen. De plus, il a été observé qu'en appliquant les modèles hydrologiques, l'habitat disponible était réduit par rapport à l'habitat maximum prédit par PHABSIM. En effet, l'habitat résultant de l'application de Tennant (30% DMA) et du 25% DMA représente environ 70% de l'habitat disponible maximum. Le débit calculé par la méthode du débit médian correspond à un habitat qui n'est que de 50% de l'habitat disponible maximum, tandis que les méthodes basées sur 90% du débit classé et l'analyse statistique des débits faibles ne représentent plus que des habitats de l'ordre de 20% à 40% de l'habitat disponible maximum. L'application de ces deux dernières méthodes laisse beaucoup de doute sur le niveau de protection des habitats aquatiques qu'elles procurent et il a été jugé utile de ne pas les recommander pour l'évaluation du débit réservé dans la région d'étude. Les autres méthodes (Tennant, 25% DMA et Q₅₀) peuvent être utilisées. Cependant, l'application de la méthode du débit médian, qui peut donner des résultats proches de 50% de l'habitat disponible maximum, doit être appliquée avec précaution.

EN :

Many techniques exist to calculate instream flow requirements. This study considers hydrologically-based techniques and hydrobiological or habitat preference methods. The hydrologically-based techniques use only historical streamflow data, and require little or no field work. Conversely, the habitat preference methods require knowledge of the specific hydraulic conditions of the studied water course and the habitat preferences of the relevant fish species.

Five hydrologically-based methods and one habitat preference method were applied to Catamaran Brook, a small drainage basin in New Brunswick, Canada. The hydrologically-based techniques included the Tennant Method, the 25% Mean Annual Flow (MAF), the median monthly flow (Q₅₀), the 90% flow duration method (Q₉₀) and a low-flow frequency method (7Q10). The habitat preference method studied was the PHABSIM model applied for Atlantic salmon.

The PHABSIM model was calibrated using the hydraulic characteristics of water depth (D), velocity (V) and substrate (S) for three flows. It was then used to calculate the same physical habitat parameters (D,V,S) for other discharges. The hydraulic results were used with habitat preference (suitability curves) to calculate the potential habitat or weighted usable area (WUA).

The application of hydrologically-based in-stream flow techniques showed that methods such as Tennant, 25% MAF and the median monthly flow method provided similar results, especially during low flow periods. The in-stream flow requirement calculated by Tennant Method (30% MAF) was 0.20 m³·s^-1, and the 25% MAF represented a value of 0.16 m³·s^-1. The application of the Q₅₀ approach yielded results of 0.13 m³·s^-1 for the month of August with higher values for other months. In contrast, the 90% flow duration and the low-flow frequency methods established very low discharge for in-stream flow requirements during low-flow periods. The lowest Q₉₀ observed was in September at 0.050 m³·s^-1 whereas the months of August, July, October, February and March all showed results slightly higher than September but still lower than 0.10 m³·s^-1. The method that calculated the lowest in-stream flow value was the 7Q10 (low-flow frequency) Method with a discharge of only 0.037 m³·s^-1. These results (Q₉₀ and 7Q10) represent significantly lower in-stream flow values compared to the Tennant, 25% MAF and the Q₅₀ methods.

The application of a habitat model (PHABSIM) at Catamaran Brook showed that the maximum available habitat, expressed as weighted usable area (WUA), was observed at a discharge close to the mean annual flow. Results also showed more habitat for salmon parr than for fry at maximum available habitat (optimal habitat), and this optimal habitat was at a higher flow for parr than for fry.

The results from hydrologically-based methods were compared to the maximum value derived from the PHABSIM method. This comparative study showed that habitat was reduced by 30% to 80% of the maximum WUA calculated by PHABSIM depending upon which hydrologically-based in-stream flow technique was applied. Habitat (WUA) resulting from the Tennant and the 25% MAF methods derived flows represents approximately 70% of the maximum available habitat. Results from the application of the median monthly flow method (Q₅₀) showed in-stream flow providing over 50% of maximum available habitat. The 90% flow duration method and low-flow frequency method (7Q10) showed habitat values in the range of 20% to 40% of maximum. The application of these latter two methods clearly limits the available habitat for the protection of aquatic resources and they were therefore not recommended for use in in-stream flow studies in this region. Instead, methods such as Tennant and 25% MAF should be used. The Q₅₀ method may be used with caution as its use results in habitat availability slightly over 50% of maximum WUA.

FR :

Le terme "lixiviat" ou "jus de décharge", désigne l'eau qui a percolé à travers les déchets en se chargeant de polluants. Ces effluents pollués doivent être traités.

En raison des exigences croissantes des normes de rejet et de la stabilisation des lixiviats au cours du temps, de nouvelles techniques ont fait leur apparition dans ce domaine. La technologie de l'osmose inverse s'est développée dans de nombreux pays européens. Cependant cette technique sélective et coûteuse, se justifie seulement quand les normes sont drastiques. C'est pourquoi la nanofiltration pourrait constituer une solution intermédiaire.

L'objectif de ce travail est de contribuer à mieux maîtriser cette technique pour l'élimination de la Demande Chimique en Oxygène (DCO) récalcitrante, subsistant après les traitements biologiques classiques.

Nous avons évalué les performances de trois membranes (organiques et minérales) pour l'abaissement de la charge organique, en fonction des conditions hydrodynamiques (vitesse et pression).

Chacune de ces membranes possède un comportement spécifique vis à vis de ces lixiviats stabilisés (adsorption, polarisation de concentration, obstruction des pores).

L'influence d'une coagulation préalable sur les performances d'épuration a également été examinée pour l'une des membranes.

Cette étude constitue une étape préliminaire au dimensionnement d'une installation.

EN :

Landfill leachate is the name given to water that has passed through solid waste and contains organic and mineral contaminants. Therefore this effluent must be treated before discharge to the environment. Because of new norms and the stabilization of leachates with time, new treatment methods have been designed. Thus, reverse osmosis is used in many European countries. But the use of reverse osmosis is only justified when norms are severe, because the treatment is highly selective (salt rejection >99%) and very expensive. In other cases, nanofiltration may be an interesting alternative.

The purpose of the present work is to propose a process for recalcitrant organic matter in order to optimize the technique. Thus, three membranes (organic and mineral) have been used to evaluate their ability to decrease the Chemical Oxygen Demand (COD) of the leachate. This study helps to determine the size of the device. First, physical parameters were investigated. Each time, the hydraulic regime was turbulent (Re > 2500). Higher permeation fluxes were obtained with organic membranes than with the mineral one (80 L·h^-1·m^-2 compared to 25 L·h^-1·m^-2) under the same experimental conditions (10 bar and 3.4 m·s^-1). Tangential flow rates higher than 2.5 m·s^-1 do not influence COD retention; at lower flow rates polarisation concentration may occur. The removal of COD is achieved in the three cases. At 10 bar an acceptable value of less than 120 mg O₂·L^-1 (norm) is obtained. The inorganic membrane (Tech-Sep) gives the best results (COD rejection: 70 % at 10 bar).

Membranes behave differently toward landfill. The organic membrane MP 20 (Weizmann membrane, cut-off 450 Dalton (Da), polyacrylonitrile) shows low adsorption with landfill leachate. The organic membrane MP 31 (Weizmann membrane, cut off 450 Da, polysulfone) gave a high COD retention ratio; the values for irreversible fouling and static adsorption are of the same order ofmagnitude; a strong membrane-foulant interaction must occur, which improves membrane selectivity. The mineral membrane N01A (Tech-Sep membrane, cut off 1000 Da, zirconium oxide), like MP-31, gives high static adsorption with leachate and irreversible fouling as well. The latter phenomenon can be explained by the obstruction of membrane pores by leachate particles, the size of which is near the membrane cut-off point. Fouling and static adsorption contribute to the increase in the membrane rejection rate.

We studied coagulation as a pretreatment to improve performances of the N01A membrane. Experiments have been carried out with Jar-Test and FeCl₃-like coagulants. The optimal amount of coagulant was 1.4 g Fe·L^-1; 60% COD reduction was achieved. The results obtained with the N01A membrane are improved: reduction of COD rises from 78% to 92 %, concentration polarisation is lower, and therefore the flux increases up to 53 L·h^-1·m^-2. This value still remains lower than the organic membrane fluxes (respectively 80 L·h^-1·m^-2 for organic membranes and 25 L·h^-1·m^-2 for N01A). However, coagulation may not be the appropriate pretreatment because the fouling index of the supernatant after coagulation was similar to that of the raw leachate. Permeability measurements after treatment show that internal fouling is still important (25%). In fact, coagulation does not remove molecules with molecular weights around 500 Daltons, and consequently these particles still obstruct the membrane pores. The phenomenon limits the performance (flux) of this membrane.

FR :

L'analyse statistique de séries chronologiques de données hydrométéorologiques est un des outils d'identification de variations climatiques. Cette analyse consiste le plus souvent à la mise en œuvre et à l'interprétation de tests statistiques d'homogénéité des séries. Les séries hydrologiques (données de pluie ou de débit) se caractérisent fréquemment par des effectifs faibles, et ne répondent que rarement aux conditions requises par l'application des tests statistiques dont certains sont paramétriques.

Nous avons cherché à évaluer, en terme de puissance et de robustesse, le comportement de quelques méthodes statistiques largement employées dans les études de variabilité climatique. Ce travail a été mené dans chaque cas étudié au moyen de procédures de simulations type Monte-Carlo de 100 échantillons de 50 valeurs conformes aux caractéristiques souvent rencontrées dans les séries naturelles. La variabilité simulée est celle d'un changement brutal de la moyenne. Les procédures concernées sont le test de corrélation sur le rang, le test de Pettitt, le test de Buishand, la procédure bayésienne de Lee et Heghinian, et la procédure de segmentation des séries hydrométéorologiques de Hubert et Carbonnel. Des séries artificielles soit stationnaires, soit affectées par une rupture de la moyenne, normales, non-normales, autocorrélées, présentant une tendance linéaire ou un changement brutal de la variance ont été générées.

Les conclusions de ce travail doivent être nuancées selon la méthode considérée. D'une manière générale la puissance maximale estimée se situe autour de 50% pour des taux de rupture de la moyenne de l'ordre de 75% de la valeur de l'écart-type. Par ailleurs il apparaît que l'autocorrélation et la présence d'une tendance dans les séries sont les deux caractéristiques qui pénalisent le plus les performances des procédures.

EN :

Statistical analysis of hydrometeorological time series is often used to identify climatic variations. Most often this analysis consists of applying and interpreting statistical tests of time series homogeneity. Hydrological time series (rainfall and runoff data) are often short and do not always comply with the hypotheses of the statistical methods. Through simulation we have investigated the power and the robustness of some tests which are widely used in the studies dealing with climatic variability. In each case studied, one hundred samples of fifty elements have been generated based on the main characteristics of natural rainfall series. A shift in the mean has been used to represent a possible climatic variation. The procedures used are the rank correlation test, Pettitt's test, Buishand's test, Lee and Heghinian's bayesian procedure, and Hubert and Carbonnel's segmentation procedure for hydrometeorological series.

Each simulation of one hundred samples is used to assess the performances of the methods considering a specific characteristic of the series: normality or non-normality, autocorrelation, trend, shift in the variance. First of all, stationary series have been simulated to evaluate the type I error of the tests. Then the series have been simulated with a break in the mean with different levels of amplitude, from 25% to 100% of the standard deviation value. The rank correlation test, Pettitt's test, Buishand's test and the segmentation procedure with a significance level of 1% (significance level of Scheffé's test) reject as heterogeneous less than ten series over one hundred homogeneous simulated series. This result is consistent with the type I error of a statistical test. On the other hand, Lee and Heghinian's bayesian method rejects about 40% of the series. This result means that this latter procedure must only be applied under the hypothesis of heterogeneity. The estimated power of the methods exceeds 40% to 50% when the break in the mean is more than 75% of the standard deviation value.

Independent series have been simulated from normal, log-normal and Pearson distributions to compare the performances of the methods requiring normality. The results show that normality has no significant impact on the performances of these methods. However, the simulations do show that the condition of independence of the successive elements of the series is essential to keep performances constant. Otherwise a trend in the series makes the tests inefficient, except for the rank correlation test for which the alternative is a trend. No method seems to be robust against both negative and positive autoregressive dependencies. The procedures requiring a constant variance are robust when the series keep a constant mean, but seem more or less slightly influenced by a break both in the mean and in the standard deviation.

FR :

L'osmose inverse est l'élément clef de la production d'eau ultrapure servant dans la préparation du liquide de dialyse rénale. Le vieillissement des membranes d'osmose inverse se traduit par une dérive croissante de la conductivité en sortie de membrane et par une augmentation de la teneur en bactéries dans le perméat délivré en sortie d'osmose inverse, entraînant un risque de contamination pour le patient.

Les suivis en conductivité et numérations bactériennes dans la chaîne de production d'eau osmosée de l'Hôpital de Colmar (France) montre un fort dysfonctionnement au niveau des prétraitements et notamment du charbon actif, siège d'une importante prolifération bactérienne (2.3 10³ colonies pour 100 ml en amont du charbon actif , 1.1 10⁵ pour 100 ml en aval). Une étude de corrélations entre entrée et sortie d'osmoseurs montre que la forte teneur en germes en sortie du charbon actif est directement responsable de la présence des bactéries dans le perméat.

Des tests hydrauliques effectués sur les membranes après quatre années d'utilisation intensive (6500 heures par an), permettent d'identifier les mécanismes prépondérants de vieillissement et de colmatage des membranes utilisées à l'hôpital de Colmar (compaction, formation d'un dépôt et altération de structure).

Afin d'identifier les substances colmatantes non retenues par le prétraitement, le film colmatant est analysé par analyse élémentaire, spectrophotométrie infrarouge, diffraction X et zêtamétrie, ce qui nous a permis de mettre en évidence la présence notamment d'argiles, de silice et de substances humiques.

EN :

The use of reverse osmosis to produce ultrapure water for kidney patients is very prevalent at the present time. About 20000 patients are treated every day in France and the cost of treatment is estimated at 10 thousand million French francs. The patients are provided with a permeate which in principle contains neither ions nor bacteria, which can cause disease and irreversible alteration of the upper-arm articulations. After several months of utilization, the presence of a foulant deposit and biofilm proliferation is often noted on the thin layered active surface of the polyamide membranes. In many cases, the membrane may become irreversibly fouled and require replacement.

The aim of this work was to determine the causes of membrane fouling at a reverse osmosis plant in "Hôpital Pasteur" (Colmar, France). First we analyzed the ion-rejection capacity of membranes and the evolution of conductivity over 200 weeks. We observed an increasing drift in conductivity, which could not be halted by cleaning and regeneration procedures. In the second phase of the study, we carried out microbiological studies at many sampling points of the ultrapure water plant in "Hôpital Pasteur". We noted a very important proliferation of bacteria downstream of the pretreatments. The activated charcoal unit is responsible for this problem (2.3 x 10³ UFC / 100 mL upstream from the activated charcoal bed, 1.1 x 10⁵ UFC / 100 mL downstream). We noted an important dysfunction in the pretreatment, essentially attributable to the activated charcoal (bacteria) and the resins (silica).

We performed tests with x-ray diffraction as a rapid method to determine the crystalline forms present. In the foulant deposits, we found quartz and clay. Infrared spectroscopy is useful in determining the type of organic constituents; in our deposits we noted the presence of humic materials. The analysis of those deposits gave essentially carbon and silicon. The foulant deposits were of organic and crystalline origins.

FR :

Une partie de la réponse aux impacts économiques, sociaux et humains des inondations réside en une meilleure gestion de l'occupation des sols. La méthode Inondabilité permet d'apporter une réponse opérationnelle aux acteurs en charge de la gestion et l'aménagement des cours d'eau. Elle permet de mesurer dans la même unité et de comparer les deux facteurs indépendants que sont l'aléa et la vulnérabilité, pour aboutir à une quantification objective du risque. Sa mise en oeuvre sur un bassin versant consiste en une modélisation de l'hydrologie grâce aux modèles Débit-durée-Fréquence, de l'hydraulique ainsi que de l'occupation du sol pour aboutir à une représentation cartographique du risque.

L'originalité de la méthode Inondabilité tient à la quantification de l'aléa et de la vulnérabilité en une même unité, une période de retour qui permet une comparaison objective de deux grandeurs très différentes. De plus, la quantification du risque est estimée à l'aide d'une différence (Risque=∆ (aléa, vulnérabilité)) contrairement aux approches traditionnelles qui privilégient souvent un produit (Risque=Coût X probabilité ), permettant ainsi la définition d'un risque acceptable (∆ négatif).

EN :

In recent years, the incidence of flood damage in France and Europe has demonstrated that we still have some progress to achieve if we are to cope with inundations. To do so, prevention, forecasting and promotion of risk culture or awareness are possible tools. They should be used together because of their complementary nature and their common field of application. In risk prevention, it seems that conceptualizing risk by dividing it between a socio-economic dimension (vulnerability) and a hydrological dimension (hazard) is a productive approach.

The first factor, "vulnerability", takes into account how a certain land use can deal with the phenomena, regardless of its occurrence. It is mostly relevant to the fields of sociology and economics. The second factor, "hazard", summarizes the natural phenomena, the occurrence of which could create human or economic damage. Both of these factors are relevant to hydrology and hydraulic sciences and have been studied for years. A given land area could be considered as "risky" when the probability of occurrence of the hazard is higher than what can be tolerated. Both hazard and vulnerability should be estimated as objectively as possible. The main difficulty in comparing these two factors is to define a similar measure for the two concepts, with a single unit. This is the purpose of the Inondabilité method: to produce a similar reference allowing an objective comparison.

The aim of Inondabilité is to define a method to quantify both hazard and vulnerability in a compatible manner and then to build easy-to-understand, synthetic maps of risks as the basis for a real negotiation in the land-use planning process. Such maps allow the comparison of different evolution scenarios, taking into account the two components of risk, and their possible changes.

Hydrology should give a synthetic description of the flow regime of the studied river. This can be based on flow-duration-frequency (QdF) models which describe the statistical functioning of the catchment and facilitate the calculation of the probability of events. Such synthetic models also yield synthetic hydrographs that are useful for hydraulic modelling and allow the quantification of vulnerability. Hydraulics is specially used to calculate hazard. Using hydrological results, a hydraulic model calculates flooded areas for flows of various return periods, from frequent to extreme. Each land parcel could be affected by the return period of the first flooding discharge corresponding to its hazard level. The same model will then be used to estimate the consequences of hydraulic works.

Describing land use allows us to attribute a reasonable protection target to each land type. Varied areas such as natural (forests, grass lands, …), rural (crops fields, orchards, …) and urban (camping areas, leisure areas, cities …) should be treated differently. Each land-use type is characterized by a target expressed in terms of frequency, duration and depth of an acceptable flood, resulting from local reviews and consultations. Then, an equivalent measure could be calculated, according to flow-duration-frequency models. Its unit is a return period expressed in years.

After these two steps, that is to say hazard and vulnerability analysis, it is possible to build different maps. Hazard maps show the flooded areas for different return period flows. Vulnerability maps show the spatial distribution of this measure, equivalent to the need for protection allocated to each parcel according to its land use. The two variables being expressed in the same units, a return period in units of years, it is easy to compare them for each parcel and to convert this comparison into an easy-to-understand colour code:

- either hazard is higher than vulnerability (the hazard's return period is lower than that of vulnerability): this shows an incompatibility between the desired protection target and the actual hydraulic constraint, and the parcel is colored in red ;

- or hazard is lower than vulnerability (the hazard's return period is higher …), which shows a reasonable safety of the parcel according to its land use, and the parcel is colored in green ;

- or the parcel is away from the maximum flooded area: there is no risk at all, and the parcel is colored in yellow .

The same formalism allows the representation of the results for any kind of solution, structural options such as hydraulic works, or non-structural options such as changes in land-use planning, and their comparison. Synthetic maps are a more efficient way to compare management scenarios and their effect on the distribution of risk all along the river than are classical design flow studies. They should facilitate real negotiation and lead to a more sustainable river management.

The Inondabilité method offers advantages from other commonly used methods. First, it doesn't deal with only one referenced flood but with floods of various frequencies. It provides a good description of the overall regime of the catchment. Second, the definition of the risk as a difference ( Risk=∆ (Hazard, vulnerability)) implies a search of an acceptable solution instead of an optimized solution, usually impossible to reach. Third, the projection of vulnerability and hazard in a 1D space makes it very easy to draw a risk map, and this map is easy to understand for anybody, even non technical personnel.

The originality of Inondabilité method lies in the quantification of vulnerability and hazard in the same unit, a flood return period. However, in the future more economic and social concepts have to be introduced to better fit the complex reality.

FR :

Le travail a consisté à préciser les conditions de formation d'une molécule iodée sapide, l'iodoforme, lors de l'oxydation d'une eau brute par le chlore et à proposer une voie réactionnelle possible.

L'étude de la chloration d'une eau brute en présence d'azote ammoniacal et d'ions iodure conduit à la formation d'iodoforme uniquement pour des taux inférieurs au point de rupture. Les résultats montrent que l'oxydation de l'ion ammonium conduit à la formation de monochloramine dont le pouvoir oxydant totalement disponible pourrait être impliqué dans la formation de iodamines ou de chloroiodamines. Ces réactions sont plus favorables en présence d'iode qu'en présence d'ions iodure. Mais l'action de l'iode seul en présence d'ammoniaque et en absence de monochloramine ne permet pas d'expliquer la production des composés organoiodés observés. Ce sont les précurseurs intermédiaires formés à partir des chloramines qui, par action sur la matière organique naturelle, seraient responsables de la formation d'iodoforme. Dans une moindre mesure, certains composés azotés organiques tels les amines et les acides aminés pourraient prendre part à la production des composés organoiodés lors de la chloration.

EN :

This work consisted of specifying the conditions of iodoform formation during chlorination of a raw water containing iodides. To reach this objective, there was need to spike the studied natural water with potassium iodide (200 µg.L^-1) in order to increase the low natural iodide content. Free and combined chlorine, chlorinated and brominated trihalomethanes (THMs) and iodoform were analyzed.

It was shown that :

- iodoform is formed for chlorine doses prior to the breakpoint, in a region where the formation of the most classical chlorinated and brominated THMs is usually disfavored (Figures 1-4);

- in the presence of chloramines the rate of production of iodoform increases with increasing I- or I2 (Figure 5);

- the direct reaction of I2 with THM precursors to produce iodoform is slow and independent of the presence of ammonia (Table 1).

- Nitrogenated compounds such as amines and amino acids would also take part in the production of organoiodinated compounds during chlorination (Figure 7). However, under water treatment conditions, taking into account the amine and amino acid content of natural waters, this class of compounds will only take a small part in the mechanism of iodoform formation.

Among the possible routes that could account for the observations made in this research, the formation of iodamines or chloroiodamines as intermediates is suggested (Figure 8). From a practical point of view, the removal of ammonia from water by a biological process (nitrification step) would inhibit the iodoform formation potential and allow the application of the final chlorination step. Another alternative would involve replacing the chlorination step by oxidation with chlorine dioxide.