Volume 18, Number 1, 2005

FR:

Cet article concerne l’absorption et la désorption du SO₂ par des gouttes d’eau de diamètre supérieur à 1mm en chute libre dans un mélange air-SO₂ à faible et moyenne concentrations. Dans ce cas, le transfert résulte du couplage des résistances interne et externe à la goutte. Dans la phase liquide, un modèle local basé sur la vitesse de frottement inter faciale et le diamètre de la goutte permet le calcul du coefficient de transfert interne kl. Le coefficient de transfert externe kg dans la phase gazeuse est déterminé à l’aide d’une expression plus classique

Afin de valider le modèle, des investigations expérimentales sont menées en absorption et en désorption sur une colonne de 2.3 m de hauteur dans laquelle le temps de séjour des gouttes est de l’ordre de la seconde. Le présent modèle simule fort bien l’ensemble de ces expériences réalisées pour différents diamètres de goutte [2.04 ; 4.31] mm et différentes concentrations [100 ; 2000] ppm. Le modèle proposé est aussi comparé avec succès à des résultats expérimentaux de la littérature à faible et moyenne concentrations pour des temps de contact beaucoup plus grands.

Son domaine d’application couvre donc désormais l’absorption et la désorption du SO₂ pour des concentrations comprises entre quelques ppm et quelque %.

EN:

Mass transfer in dispersed media is of interest to fields such as nuclear engineering, process engineering and environmental engineering. It occurs when two phases, not under chemical equilibrium, are in contact. Knowledge of mass transfer mechanisms in the case of gas absorption from and/or into droplets is necessary to understand the scavenging of trace gases in clouds, rain and wet scrubbers. Our studies focus on absorption and desorption phenomena involving free falling water droplets in a mixture of air and gas. For example, acid rain is formed when a drop of rain falls through an atmosphere contaminated with gaseous acid precursors. A similar phenomenon occurs in specific atmospheric scrubbers, where pollution is trapped at the source. In all cases, the transfer of trace gases from the air into the falling droplets is controlled by molecular diffusion and by convection outside and inside the drops.

For droplets, falling inside a soluble gas medium, the main transfer resistance is located in the gas phase. A survey of published studies shows that a number of good numerical models exist, as well as experimental correlations for predictions of the mass transfer coefficient in the gas film. For the liquid phase controlled resistance, Saboni (1991) proposed a model based on local scales, interfacial liquid friction velocity and drop diameter. The model was validated experimentally by Amokrane et al. (1994). The experimental study and model validation in the case of sulfur dioxide absorption by water droplets falling through air with a high gas concentration (few %) has been described previously in detail by Amokrane et al. (1994).

The purpose of the present article was to extend our previous model to predict SO₂ absorption and desorption by droplets (1-5 mm) falling in air with a low gas concentration. In the liquid phase, a model based on local scales, interfacial liquid friction velocity and droplet size diameter was used. In the continuous gas phase a more classical model was applied. To support the model, two types of experiments were carried out. The first type was adapted to measure the absorption of gas by droplets of known diameter. A second set of experiments gave the desorption rate from droplets with an initial concentration of sulfur dioxide falling through SO₂ -free air. Absorption occurred during the fall through a 2.3 m long column for various gas concentrations and for various droplet diameters. A sketch of the experimental equipment is presented schematically in Figure 1. It consists of a cylindrical column 2.3 m in height and 0.104 m in diameter. Before each experiment, a gas mixture with the desired SO₂ concentration in air, ranging between 100 and 2000 ppm, was introduced into the column. The SO₂ concentration was set at the desired value by regulating the volumetric flow rates of sulfur dioxide and air with calibrated rotameters. The gas concentration in the column was measured continuously by a chemical cell analyzer. The air temperature and humidity were continually measured at the top, in the middle and at the bottom of the column. They ranged from 18°C to 20°C and from 40% to 50%, respectively. Droplets were generated using a specific injector consisting of a demineralized water tank at the base of which identical thin needles were placed. In the case of the smallest droplets, seven needles, 300 µm in diameter, were used. For the largest droplets, one needle of about 1 mm was used. The artificial rain was started by exerting an overpressure in the tank and it was stopped by exerting a depression. This device allowed the generation of almost identical water drops at a controlled rate. Droplets fell with zero initial velocity. Their diameters were determined by collecting a known number of droplets and weighing them on a precision balance. The droplets were collected in a special glass cup placed at the bottom of the rain shaft. This collector initially contained a known volume of kerosene. The presence of this organic compound allowed the creation of a film to prevent additional absorption of SO₂ during the experiment and natural desorption of sulfur after the experiment. An experiment consisted of dropping 10 to 20 mL of rain. This amount is enough to precisely measure the sulfur concentration.

For reversible desorption, experimentation was undertaken directly in a lab atmosphere. For these experiments, the 4.31 mm diameter droplets free fall occurred over 16.3 m. Three intermediate levels were also examined with falling times varying from 0.7 to 2.4 s. The ambient temperature was measured in the surrounding area of both the injector and the collector and the maximum variation was 2°C. Various initial sulfurous acid concentrations were obtained as a result of various contact times of demineralized water with air-SO₂ mixtures. Initial concentrations ranged from 0.5 10^-3 mol·L^-1 to 1.8 10^-3 mol·L^-1. In this case, the collector initially contained a known volume of hydrogen peroxide to immediately convert sulfurous acid into sulfuric acid. This avoided additional desorption of sulfurous acid during and after the experiments. In this case, the presence of the organic film was not necessary.

The results achieved with the theoretical model were compared to the experimental results. The present model was successful in correlating the experimental results carried out for various droplet diameters ranging between 2.04 and 4.31 mm, and gas concentrations ranging between 100 and 2000 ppm. The model also compared successfully with experimental results from the literature in the case of much longer contact times. The applicability of the model thus covers the absorption and desorption of SO₂ for concentrations ranging between ppm to a few %.

FR:

Le comportement hydraulique des déversoirs d’orage latéraux est le plus souvent marqué par une évolution discontinue de la ligne d’eau caractérisée par le ressaut hydraulique (Torrentiel/fluvial) et, dans certains cas, par une évolution rapidement variée causée par un écoulement également transcritique mais dans le sens fluvial/Torrentiel. Concernant les modèles actuels, ils ne permettent pas de simuler le comportement en transitoire de ce type d’ouvrage. Compte tenu de cela, nous avons modélisé l’ouvrage par les équations de Barré de Saint Venant écrite sous forme conservative en régime transitoire et couplée au modèle de déversoir de Hager. Le caractère conservatif de ces équations permet de transcrire dans un seul système d’équations les écoulements graduellement et rapidement variés. Afin de résoudre ces équations, nous avons utilisé le schéma numérique UPWIND à « capture de choc » du second ordre du type TVD (Total Variation Diminishing) utilisant le solveur de Roe. Dans l’objectif de valider notre démarche, nous avons créé un pilote de déversoir d’orage sur le site d’Obernai (laboratoire d'hydraulique du lycée agricole). Nous avons fait varier : le diamètre de la conduite aval par rapport au diamètre de la conduite amont, la longueur du déversoir, les pentes des conduites amont et aval ainsi que la hauteur de crête. La comparaison des débits déversés entre les modèles physique et numérique a montré que l’erreur observée rapportée au débit amont n’excédait jamais 13 % avec une majorité des cas entre ±5%.

EN:

In recent years, French and European legislation has introduced regulations about wastewater discharge into natural environments and particularly about combined sewer overflows. As a consequence, it has become essential to control the hydraulic behaviour of these structures and to estimate the pollution loads released at this level. The side weir is the regulation structure that permits the hydraulic regulation of the waste water carried by the sewer system. When the upstream flow intensity exceeds a value referred to as reference flow, the side weir directly rejects part of the waste water to the natural environment. The hydraulic behaviour of the sewer side weir was shown to involve a discontinuous evolution of the water depth, characterized by a hydraulic jump (transition from supercritical to subcritical flow) and also by a rapidly varying transcritical evolution (subcritical to supercritical).

Initially, the side weir flow was determined with the use of empirical relations. Using formulae of Engels, Coleman and Smith, Balmaceda and Gonzales or Dominguez, it was possible to calculate the outflow according to the water level at the upstream and/or the downstream region of the weir. These relations were applicable only for certain flow regimes and in certain cases in which the geometry of the side weir was specified. Subsequently, a more physical approach, initiated by Ackers, was based on the assumption of constant energy along the side weir. This approach made it possible to focus not only on an assessment of the side channel flow, but also on the water profile at the crest. Unfortunately, as the study of El Kashab shows, this method falls short in certain cases because the equations are inappropriate. For example, in the case of hydraulics the constant energy approach was not applicable. Finally, the method that is currently used is based on a momentum equation, which makes it possible to establish equations for shallow water. This approach seemed the most appropriate in the case of the side weir. The numerical solution of these equations was always based on an algorithm that describes all the possible cases according to the flow regime and the hydraulic conditions in the side weir. One must know the flow regime a priori. These models don’t properly simulate the transitory behaviour of these kinds of works.

In this article we propose hydraulic modelling of a sewer side weir that integrates the geometrical characteristics of the flow (height and length of the crest, variation of width along the crest), and avoids the need for a priori knowledge of flow conditions in the side weir. The model also takes into account hydraulic discontinuities (hydraulic jump, transitions from free surface to pressurised flow) and the transitory character of the flow. The numerical results were compared with measured values obtained from a test bench.

For the 1D approach, the solution was found using the 1D shallow water equations written in a conservative form for a transitory situation. The conservative characters of the equations permit us to consider gradually and rapidly varied flows in a single system of equations. In order to account for the lateral overflow, we used the Hager relation, which involves the intensity and direction of the lateral velocity vector and also the influence of width variation of the side weir. The shallow water equations system couldn’t be analytically solved. As a consequence, numerous numerical methods have been developed such as characteristic methods or finite difference methods. Unfortunately, these methods are inadequate when discontinuities such as hydraulic jumps or flow regime transitions (froude number close to 1) appear. To solve these problems, numerical (shock capturing) schemes were developed, based on a finite volume formulation. Godunov was the initiator of this type of finite volume numerical scheme. Eleuterio improved the precision and the ability of these numerical schemes to converge. The result was a combination of total variation diminishing (TVD) interpolation with an appropriate Riemann solver. We used a second order TVD Upwind shock capturing numerical method associated with the Roe Riemann solver.

In order to validate the numerical model, we have built a sewer side weir physical test bench at the Obernai site. The variable parameters were the downstream pipe diameter in comparison with the upstream pipe diameter, the side weir length, slopes of the upstream and downstream pipes and the height of the crest. The tested cases permitted us to sweep a slope ranging from 0.5 ‰ to 1 % for the upstream and downstream pipes.

Globally, 114 configurations have been tested with the 1D numerical model. The operating curve represents a criterion for characterising the operating of the storm overflow. As long as the upstream flow does not reach the reference value, there is no overflow. As soon as the upstream flow exceeds the value, the downstream flow remains close to the reference value. Because comparisons were made in relation with the upstream flow, the criterion used for judging the modelling performance was the absolute error value in relation to the upstream flow. It is important to note that the results need to be weighted as experimental measurements have a margin of error of approximately 5%.

Comparisons between numerical and experimental results permit the following conclusions:

1. The distribution of the number of errors was very close to the normal Gaussian distribution curve, with a slight shift towards the positive side. This indicates that the errors were random and therefore very close to experimental values.

2. In general, the maximum errors varied from -10 to 13%, with the majority of cases occurring between –5 and +5%. This shows that the performance of the tool was very useful in cases that are as complex as side weirs. However, the model has a slight tendency to overestimate the overflow rate as compared to experimental measurements.

FR:

De par leurs nombreuses propriétés physico- chimiques, les organoétains sont très utilisés dans l’industrie et en agriculture et entrent dans la composition de nombreux produits domestiques. Ils sont cependant extrêmement toxiques et la Communauté Européenne les a classés parmi les substances prioritaires dans le domaine de l’eau.

Un suivi des organoétains a été réalisé sur onze rivières du bassin Adour-Garonne et sur l’estuaire de l’Adour. Ces composés y sont systématiquement présents, les butylétains et les octylétains étant les espèces les plus fréquemment détectées. Les concentrations varient de la limite de détection (0.2-0.5 ng(Sn)/l en moyenne) à 50 ng(Sn)/l dans les eaux, et de 15 à 300 µg(Sn)/kg dans les sédiments dulcicoles. Des pics de contamination ont été observés en fin de printemps et d’été, dans plusieurs rivières. Ils correspondent à la présence des mono- butyl- et -phénylétains principalement, leurs concentrations pouvant atteindre 700 à 900 ng(Sn)/l d’eau. Les rivières les plus contaminées sont la Garonne, le Gave de Pau, l’Adour, la Charente et le Thoré. Dans les matières en suspension de l’estuaire de l’Adour les concentrations atteignent quelques mg(Sn)/kg. L’ensemble des données recueillies a permis de mieux comprendre les origines et le devenir des organoétains dans le cycle hydrologique.

EN:

Because of their physico-chemical properties, organotin compounds (OTC) are widely used in industry and are present in a significant number of agricultural pesticides and domestic products. They are highly toxic and the European Community has listed them as priority pollutants in the aquatic environment. Organotins have been monitored in the Adour- Garonne basin and the Adour estuary. They are systematically present in the rivers, with butyl- and octyltins being the species most frequently detected. These species, especially octyltins, probably come from the continuous leaching of plastic tubes. The OTC concentrations ranged from just over detection limits (≥ 0.2-0.5 ng (Sn)/L) to 50 ng (Sn)/L in water and from 15 to 300 µg (Sn)/kg in freshwater sediment. Important seasonal variations were also observed. Thus, at the end of spring and summer, very high monophenyltin (MPhT) concentrations of up to 700-900 ng (Sn)/L were found in the dissolved phase. This phenomenon could be partly attributed to specific triphenyltin (TPhT)-based agricultural treatments, MPhT being one of the TPhT degradation products. High monobutyltin (MBT) concentrations of up to 150 ng (Sn)/L were also detected during the same period. This latter compound comes from leaching of plastics and from tri- and di-butyltin (TBT, DBT) degradation. It represents 80 to 100% of the butyl species found in sediments.

Considering OTC concentrations, speciation and toxicity, the most contaminated rivers appeared to be the Garonne, Gave de Pau, Adour, Charente and Thoré. Urban activities have significant influence on the levels of OTC contamination for most of the rivers, demonstrating continuous OTC inputs from domestic and industrial treatment plants. This is especially the case for the Charente and Thoré rivers, where some specific industrial activities devoted to leather and wood are present close to the sampling points.

A statistical study was performed on the different physico-chemical parameters (temperature, water flow rate, dissolved oxygen concentration) and OTC concentrations. A significant positive correlation between water flow rate and organotin concentrations in the dissolved phase was observed. This correlation was very important when only sampling points far from potential OTC sources were considered, the octyltin concentrations showing the strongest correlation. These observations confirm the presence of a continuous OTC diffusion into aquatic media. A comparison between the present results in Adour-Garonne and OTC monitoring performed in the Rhin- Meuse basin shows that the level of contamination was quite similar in the two basins, especially considering rivers without fluvial traffic. A similar correlation existed between OTC concentrations in the dissolved phase and water flow rate.

Special attention was given to the Adour sub-basin because of its particular geographic position and especially the large built-up area in the estuary. Butyltins remain the main OTC compound present, in terms both of frequency and concentration. According to the different sampling points in this sub-basin, mean OTC concentrations in the estuary did not appear to be really influenced by human activities located upstream, the concentrations in this part reaching 50 ng (Sn)/L in the dissolved phase. In contrast, OTC amounts found in the estuary were considerable higher. The built-up area of the estuary had a strong influence on concentrations, which were 6 to 14 times higher in the city centre than those upstream from the city (in an agricultural region). In addition to the influence of local sources, both a strong dilution effect and significant adsorption/ sedimentation phenomena in the downstream region of the estuary could be important. In the suspended matter of the Adour estuary, organotin concentrations were extremely high, reaching concentrations as high as mg (Sn)/kg. Such concentrations have already been reported for nearby regions of the harbour [Bravo et al. (2004)]. However, in the present case, there should be considerable concern considering the possible environmental consequences. The estuarine sediments appeared obviously contaminated by butyltins, but the concentrations were lower than those that could be expected (2000 µg (Sn)/ kg maximum). This observation could be explained by water flow rates as well as the tide, which could export large amounts of suspended matter outside the estuary. MPhT and TPhT were also detected, especially in sediments from the extreme downstream region of the estuary. Their presence could be attributed to the marina. The different solid/ dissolved partition coefficients were also evaluated. These partition coefficients ranged from over 40x10⁴ for sediments up to 200x10⁴ for suspended matter. Finally, the information on the Adour sub-basin showed that the estuary was more strongly contaminated than the upstream region.

Generally, all these data have contributed to the first evaluation of OTC contamination in the Adour-Garonne basin, and identified organotin sources. The statistical study, comparisons between the different parts of the aquatic environment, and the observation of solid/ liquid distributions lead to a better understanding of the environmental fate of OTCs. Even if differences exist between the level of contamination in freshwater and estuarine environments, the ubiquitous presence of OTC must remain a subject of concern, especially with regard to the high toxicity of organotins. For example, TBT and TPhT have lethal effects on trout and algal species at aquatic concentrations in the µg (Sn)/L range or even below this concentration [TOOBY et al. (1975), WONG et al. (1982)]. Considering this high toxicity, other studies will have to be performed in order to increase the current database concerning OTC in rivers. It is also important to know the conditions that control OTC uptake by biota, and in order to propose effective environmental management strategies.

FR:

L’étude porte sur la détermination de la contribution des espèces O₃, OH° et CO₃°- dans la dégradation de l’atrazine lors de l’ozonation de solutions contenant différentes concentrations en ions hydrogénocarbonate et en carbone organique. Le suivi de la concentration en atrazine et en ozone dissous, et les expressions cinétiques ont permis de calculer les concentrations en radicaux hydroxyle et carbonate au cours des réactions. A partir des données expérimentales obtenues sur des eaux pures additionnées de carbone organique et inorganique, les résultats indiquent que l’élimination du micropolluant résulte de l’action de l’ozone (pour une faible part), des radicaux hydroxyle issus de la décomposition de l’ozone, mais aussi pour une part très significative, des radicaux carbonate. La participation des radicaux CO₃°- diminue lorsque la concentration en carbone organique augmente. Les radicaux carbonate peuvent être responsable de plus de 40 % de la dégradation de l’atrazine lors de l’ozonation en présence de 7 mM d’ions hydrogénocarbonate et 129 µM d’ions glycolate utilisés comme molécule modèle pour l’apport de carbone organique. Les résultats obtenus sur des eaux naturelles confirment les conclusions déduites des expériences sur des eaux de composition connue.

EN:

The inhibiting effect of bicarbonate ions on the oxidation of organic molecules by the hydroxyl radicals is well known. However, the carbonate radicals resulting from the consumption of the OH° radicals by these ions have only rarely been considered to participate in the reactions of organic pollutant removal. In this study, the contribution of O₃, OH° and CO₃°- radicals in the degradation of atrazine during ozonation of aqueous solutions containing various concentrations of bicarbonate ions and organic carbon, was determined.

Experiments were performed in a bubble column fed continuously by ozone gas and an aqueous solution containing atrazine (0.05 µM) as the model molecule. Three sets of experiments were carried out at pH 8:

1. pure water with different concentrations of bicarbonate ions (0.35-70 mM);

2. pure water with different concentrations of bicarbonate ions (0.35 and 7 mM) and glycolate ions (0-129 µM) selected as the organic carbon source;

3. surface and tap waters.

For different contact times in the ozonation reactor, the concentrations of atrazine and dissolved ozone were determined. Inputting these data into kinetic equations enabled us to calculate the concentrations of hydroxyl and carbonate radicals during ozonation. In the absence of organic carbon, the concentration of hydroxyl radicals was determined by assuming steady state conditions (equation II). The concentration of carbonate radicals was deduced from the slope of the evolution of atrazine concentration ([A]₀ -[A])/[A] versus the contact time, and the values of ozone and OH° radical concentration (equation IV). In the presence of organic matter, the concentration of OH° radicals was calculated from the evolution of ([A]₀ -[A])/[A] versus the contact time, and by replacing the carbonate radical concentration by the expression involving the OH° radicals (equations VI-IX). From the steady state assumption, the concentration of carbonate radicals followed the hydroxyl radical evolution.

In the absence of organic carbon, the results confirmed the global inhibiting effect of bicarbonate ions on the removal of atrazine molecules. However, the concentrations of the carbonate radicals were much higher than the OH° radical concentrations (above 4x10^-10 M compared to 2x10^-12 M, respectively). The concentration of these latter radicals decreased as the concentration of bicarbonate ions increased. Under these conditions, the carbonate radicals were mainly responsible for the removal of atrazine.

From the experiments in pure water with given concentrations of bicarbonate and glycolate ions, the hydroxyl radical concentration increased with the concentration of glycolate ions, thus confirming the promoter property of this organic molecule, which favours the removal of atrazine. However, in most cases, the presence of organic carbon was found to be unfavourable to the concentration of carbonate radicals. Therefore, the participation of the carbonate radicals decreased with increasing organic carbon content. Nevertheless, their contribution to atrazine degradation can reach more than 40% during ozonation in the presence of bicarbonate ions (7 mM) and glycolate ions (129 µM) used as an organic carbon source.

The results obtained from the experiments carried out in natural waters agree with the main conclusions from the above experiments. The removal of atrazine involved ozone, hydroxyl radicals as well as carbonate radicals. The contribution of carbonate radicals decreased when the ratio of inorganic carbon to organic carbon decreased. In the surface and tap waters tested, this contribution was found to be 10 to 40 % and 43 % respectively.

FR:

La prévision des crues des petits bassins versants, avec un modèle pluie-débit, est fortement conditionnée par la connaissance de la pluie. Cette information, estimée par des mesures de pluviographes ou de radar, est entachée de nombreuses incertitudes.

Les services français de prévision des crues disposent maintenant d'une version du modèle conceptuel pluie-débit GR3H, adaptée à la prévision opérationnelle. Il utilise une procédure d'optimisation d'un seul paramètre, le niveau initial du réservoir-sol.

On a voulu tester le comportement de ce modèle, face à différentes perturbations du signal pluie de base. L'hydrogramme de notre crue de référence a été préalablement simulé avec GR3H.

On a montré que le modèle est capable d'absorber d'importantes variations du signal pluie, mais seulement si l'origine et la fin de l'épisode pluvieux sont respectés. Sinon, pour compenser un décalage temporel entre pluie et débit, on a combiné plusieurs modèles GR3H à temps de réaction différents, avec une procédure multimodèles simplifiée. Enfin, pour éviter quelques instabilités, on a testé une variante baptisée "multidélais" qui a permis d'apporter un gain supplémentaire sur la qualité de la prévision.

EN:

Flood forecasting in small watersheds (some hundreds of km²) has to take into account rainfall. This is why the lumped conceptual rainfall-runoff GR3H model (Cemagref) has been adapted for the French flood forecasting services for operational use. However, the relevance of forecasting is strongly conditioned by the knowledge of real rainfall on the drainage basin. This information, estimated by rain gauge measurements or meteorological radar, contains numerous quantitative and sometimes temporal uncertainties. In this study, we tested the influence of these uncertainties on the behaviour of the GR3H forecasting model.

In the GR3H model, the input is the hourly rainfall on the watershed and the output is the hourly flow at the outlet. The production function uses one parameter (A), which represents the higher soil reservoir level. The transfer function uses two parameters: B (the maximal capacity of transfer reservoir) and C (the base time of unit hydrographs HU1 and HU2). In a discontinuous event mode, we have to add an additional parameter S0/A, the initial level of soil reservoir A. For each event, it represents the initial hydrological state of the basin. When used as a forecasting model, A, B and C values are fixed. Thus, to adapt the GR3H model for operational forecasting, we used an optimization process to select the S0/A value. At every moment, this process looks for the S0/A value that makes the calculated discharge equal to the known discharge.

To test the impact of rainfall signal perturbations on our forecasting process, we worked on a theoretical flood. Its hydrograph was simulated using the GR3H model from a basic rainfall signal with a constant intensity of 10 mm/h over 12 h. The parameters (A = 400 mm, B = 80 mm, C = 6 h) came from a study of 16 flood events, in a basin of 215 km², located in the French Pyrenees. To initialize the reservoir levels, base runoff was 1 m³ /s (for reservoir B) and S0/A was fixed at 0.65 (for reservoir A). As an operational scenario, we worked without a precipitation forecast (null future rainfall hypothesis); thus, the forecast time was limited to half the C value, i.e. 3 h, due to the parabolic pattern of the unit hydrograph HU1. To quantify forecast performances, we used the persistence index, which compares the studied model with an inert model (i.e., future is equal to present). We tested successively three kinds of perturbations on rainfall signal:

1. the variability (max 50%) of the hourly rainfall intensity, over 50 simulations, preserving the total sum of rainfall;

2. the variability (max 50%) of the total sum of rainfall, over 11 cases, preserving a constant intensity (from 5 to 15 mm/h) and

3. shifting the beginning time of rainfall, over 7 cases from –3h to +3h.

For each kind of perturbation, we considered two forecasting protocols: first a non-operational protocol in which the initial state is known a priori (fixing S0/A); and second as in operational situations, in which the initial state is unknown (optimizing S0/A). We demonstrated that our optimization updating process was rather well adapted to balance the quantitative variability of rainfall. On the other hand, it was not effective to balance an important temporal shift in rainfall. Indeed, in the GR3H model, the temporal parameter (C) is independent from production parameters (A, B and S0/A). To solve this problem we used a multi-model procedure (PMM), i.e., a linear weighting method of results from different forecasting models. The weight of each variable depends on the relevance of the past forecast. We combined three different GR3H models with the same A and B values and different C values (4 h, 6 h and 8 h). This method gave better results but we observed some forecasting instabilities. To solve this problem, we used a multi-time PMM. To improve the 3 h time forecast, we also considered the performances of short-term forecasts (1 h and 2 h). We tested GR3H forecasting over ten French watersheds, using from 12 to 25 events. Results were rather interesting, except when the rainfall signal was not representative of the real spatio-temporal variability (e.g., thunderstorms or basins that were too large). In these cases, semi-distributed models should be useful.

A priori, our conclusions were focused on the GR3H model and our updating procedure. However, we propose that they could be similar for other hydrological global models, which use reservoirs and few parameters, offering some inertia and stability to the system. To conclude, when the GR3H was able to model the hydrological behaviour of a small watershed, forecasts were not strongly influenced by quantitative imprecision in the rainfall signal, as long as this imprecision did not greatly affect the beginning and, mainly, the end of the rainy episode.

FR:

Les débits annuels minimums des rivières déterminent le volume d’habitat minimum disponible pour assurer la survie des espèces aquatiques en période d’étiage. Dans cette étude, nous comparons les impacts de barrages sur les caractéristiques (période d’occurrence, magnitude, amplitude de variation et asymétrie) de ces débits dans trois régimes hydrologiques artificialisés d’une part, et les débits annuels minimums mesurés en aval des barrages aux normes de débits réservés pour protéger les habitats du poisson au Québec, d’autre part. Nous avons analysé 72 stations appartenant aux régimes artificialisés d’Inversion (26 stations), d’Homogénéisation (18 stations) et de Type Naturel (28 stations). Toutes ces stations appartiennent au bassin versant du fleuve Saint-Laurent. La présente analyse est fondée sur la comparaison des débits mesurés en rivières naturelles (75 stations) à ceux mesurés en aval des barrages au moyen des méthodes de proportionnalité et graphique. Il ressort de ces comparaisons les principaux résultats suivants.

En régime artificialisé d’Inversion caractérisé par les débits mensuels maximums en hiver et les débits mensuels minimums au printemps, les impacts des barrages se traduisent par une hausse significative de fréquence des débits annuels minimums au printemps au moment de la fonte des neiges mais une baisse en été, une diminution significative de la magnitude des débits pour les bassins versants de taille < 10 000 km², une hausse de la variabilité inter-annuelle et une forte asymétrie de la distribution.

En régime artificialisé de Type Naturel caractérisé par des débits mensuels maximums au printemps et des débits mensuels minimums en hiver ou en été, on observe une hausse de la fréquence des débits annuels minimums pendant la première moitié de la période froide (de novembre à janvier), une diminution significative de la magnitude pour certaines rivières de taille < 6000 km².

En régime d’Homogénéisation caractérisé par des débits mensuels quasi constats toute l’année, les barrages provoquent une hausse de la fréquence des débits annuels minimums (printemps et automne) mais une baisse en été. Mais contrairement aux deux régimes précédents, l’impact des barrages se manifeste surtout par une hausse de la magnitude des débits annuels minimums pour quelques rivières.

Pour les trois régimes artificialisés et durant les quatre saisons, les débits réservés sont systématiquement supérieurs aux débits annuels minimums lâchés en aval des barrages. L’écart entre les deux types de débits est surtout observé au printemps et en été pour les bassins versants > 10 000 km².

EN:

Annual minimum discharges represent a crucial hydrologic parameter for the health of aquatic ecosystems. They determine the volume of available habitat for aquatic species and influence the concentration of pollutant within the fluvial system during low flows. They are also of importance for instream infrastructures and for the regulation of fluvial transport. For these reasons, the minimum discharges constitute the main hydrologic parameters for which clear regulation have been defined in several countries. In the province of Québec, albeit the large amount of dams on several important fluvial systems, there seems to exist a lack of studies examining their effects on the annual minimum discharges. This paper is aiming at highlighting the effects of dams (1) by examining their effect on the characteristics of annual minimum discharges for artificialised flow regimes in Québec, and (2) by comparing those discharges with recommended instream flows to protect fish habitats.

Firstly, the effect of dams on annual minimum discharges is examined for the three types of artificialised flow regimes found in Québec. From the analysis of seasonal and monthly discharges, ASSANI et al. (2004) documented the three types of artificialised hydrologic regime downstream from dams: the inversion, the homogenization, and the natural type flow regimes. The inversion flow regime presents high monthly discharge values in winter and low monthly discharge values during spring. This type of regime occurs solely on the north shore of the St-Lawrence River and pertains to rivers with large reservoirs feeding in hydropower stations. The homogenization flow regime presents small annual fluctuations of the monthly discharge. The maximum monthly discharges are recorded during spring where- as the minimum monthly discharges frequently occur during fall. This type of regime is often associated with reservoirs created on large streams for which the storage of spring water is less important. This regime is observed mainly on the north shore of the St-Lawrence river. In the natural type flow regime, the maximum monthly discharges take place during spring snowmelt while minimum monthly discharges occur either during summer or winter. The annual natural flow characteristics are thus conserved albeit the existence of the dam. This regime pertains to dams with small reservoirs and it is found on both side of the St-Lawrence River.

Secondly, annual mimimum discharges are compared with minimum instream flows recommended by BELZILE et al. (1997). These ones defined the minimum instream flows based on the different species of fish and their life cycle. Downstream from dams, the instream flows (Qr) can be estimated using the following relation:

Q_r = e^k.S^a

where S represents the drainage area upstream from the dam; a and k are respectively regional and seasonal parameters. These parameters are associated to the ecohydrological region, to the season as well as to the critical phases of life cycle for the fish species found within the ecohydrological regions.

From the Historical Stream Flow Summary of Environmental Canada, the distribution of discharge from 107 stations were selected and analysed. From those, 72 were located on rivers with dams and 75 on rivers with no regulation. On regulated rivers, 26, 18 and 28 were identified as belonging to the inversed, homogeneous and natural type regimes, respectively. All stations were located in the St-Lawrence drainage area. To highlight the effect of dams, we performed a comparison between the annual minimum discharges for stations on artificialised rivers to those from stations belonging to rivers with no regulation. The comparison is performed according to the size of the drainage basins (proportionality method) and uses a set of parametric and non-parametric statistical tests depending on the type of data. The proportionality method was chosen because of the non-availability of the discharges for the pre-dam periods. According to RICHTER et al. (1996), river flows can be described using several parameters relating to the daily discharges: the magnitude, the frequency, the duration, the timing and the rate of change (amplitude of the variability). The daily discharges required to compute these parameters were not available. The date of occurrence of annual minimum discharges, their magnitude, the interannual variability of the magnitude and the skewness of the distribution could however be obtained from the Historical Stream Flow Summary of Environmental Canada.

The analysis of annual minimum discharges for the three types of artificialised flow regimes highlights several key elements associated with the effect of dams. For the inversion flow regime, the presence of dams increases and decreases significantly the occurrence of annual minimum discharges during spring and summer, respectively. For drainage area smaller than 10 000 km², the magnitude of the annual minimum discharge is decreased significantly. Finally, the between-year variability is increased and the distribution presents a strong skewness. For the natural type flow regime, an increase in annual minimum discharges during the period between November and January can be observed as well as a significant decrease in magnitude for the small fluvial systems (drainage area < 6000 km²). For the homogenization flow regime, the frequency of annual minimum discharge is increased during spring and fall while decreased during summer time. However, in contrast with the two previous artificialised flow regimes, there is an increase in magnitude for the minimum annual discharges.

Finally, for the three types of artificialised flow regime and for the four seasons, the minimum annual discharges released downstream from the dams appear to be systematically smaller than the instream flows recommended by BELZILE et al. (1997). The main differences are observed during spring and summer for drainage basins > 10 000 km².