FR :

L'évaluation de la qualité de l'eau de rivières pose un certain nombre de problèmes non encore résolus. Parmi ceux-là, nous en avons retenu un qui porte sur la méthodologie généralement adoptée pour réaliser cette évaluation et qui est à la base de nombreux autres problèmes. Actuellement en France, ainsi que dans de nombreux autres pays européens, la qualité de l'eau est évaluée à partir de grilles de valeurs limites de paramètres physico-chimiques ou biologiques. L'usage de telles grilles sous-entend que les paramètres qui la composent sont indépendants les uns des autres alors qu'il n'en est rien. En effet, il est indispensable de considérer les milieux aquatiques comme étant des systèmes chimiques, physiques et biologiques plus ou moins éloignés de leurs états d'équilibre et pour lesquels les paramètres qui les décrivent, considérés à un instant donné, sont liés les uns des autres par des relations qui évoluent avec leurs états.

En plus du problème précédent, il faut rappeler que toute évaluation de l'état d'une masse d'eau (au sens de la Directive Cadre européenne sur l'Eau (DCE)) nécessite que soit précisé au préalable un état de référence. La diversité, hors anthropisation, des milieux aquatiques pose alors le problème du choix de telles références.

Ce travail qui a pour cadre un cours d'eau du centre de la France (la Vienne) s'inscrit donc en partie dans cette double problématique.

S'appuyant sur un suivi analytique mensuel en huit points, réalisé au cours d'une période de 27 mois entre 1999 et 2002, nous avons tenté de montrer que la méthode bien connue de l'analyse en composante principale (ACP) avec étude des corrélations, pouvait, moyennant quelques préalables, être un outil permettant de préciser avec une bonne objectivité l'état, ainsi que l'évolution spatio-temporelle de ces eaux.

Nous avons également tenté de montrer que l'un de ces préalables était la prise en compte dans l'ACP de variables caractéristiques de la matière organique dissoute (MOD). Les paramètres classiques actuellement utilisés, en particulier par le système français d'évaluation de la qualité de l'eau (SEQ-Eau), qui sont essentiellement minéraux, ne suffisent pas à eux seuls pour expliquer et mettre en évidence les effets liés aux apports anthropiques.

Dans cette étude, nous nous sommes donc attachés de différentes manière à montrer sur ce milieu, que les représentations ACP prenaient en compte les principales observations connues comme liées aux rejets polluants.

EN :

Satisfactory systems for evaluating river water quality have not yet been established. In France, as in many other European countries, water quality is evaluated using physico-chemical or biological parameter grids. This method implicitly assumes that these parameters are independent from each other and that a universal state of reference exists. In fact, when taking into account the great diversity of rivers it is clear that any evaluation of water quality requires that reference conditions be specified for each water system. Although it is one of the objectives of the new European Water Framework Directive, this problem has not yet been resolved. In addition, it is important to note that the state of water is characterised by dependent analytical parameters, and that the relationships linking these parameters evolve with the state of the water; this variable dependence must be taken into account. Lastly, it is important to note that organic matter is generally underestimated by current water quality evaluation systems.

In the present study an attempt was made to evaluate the water quality of the Vienne River situated in western France. This study provided us with the opportunity to verify the accuracy of the current system. A monthly analytical monitoring program at eight study sites was carried out during the period 1999-2001. It allowed us to demonstrate that the well-known method of the principal component analysis (PCA), together with correlation analysis, was an appropriate tool for this purpose. More specifically, we showed that when using parameters characteristic of organic matter, this tool is well adapted for monitoring spatial and temporal changes in water quality. As this method is graphical, it made the choice of a reference point possible.

This study showed that traditional parameters currently used, which are primarily inorganic, were not sufficient to identify anthropogenic contributions. For this reason we used unconventional organic variables, formerly used to study the structure of the dissolved organic matter (DOM) in surface waters, such as:

- the distribution of the various organic fractions, such as humic (or hydrophobic) substances and non humic (or hydrophilic) substances, after filtration through macroporous XAD8 and XAD4 resins, under well-defined conditions;

- the distribution of the apparent molecular weights, by ultrafiltration through membranes with molecular weight cut-offs of 30, 10 and 3 kDaltons;

- fluorescence spectrometry;

- the ratio UV-absorbance at 254 nm/dissolved organic carbon (SUVA).

It was noted in this study that these organic variables, coupled with those classically analysed (Chemical Oxygen Demand, COD; Biological Oxygen Demand, BOD5), were often mainly responsible for the variations in the water quality. Principal Component Analysis was used to extract the main factors associated with water quality variability. The graphical representation obtained underlined that the greatest variability of the points characteristic of the stations could be observed on the first principal component. This component was linked to the organic variables. The second principal component was linked to inorganic variables indicative of the geological substrate. In addition, it was shown that the temporal dispersion of stations in the PCA graphical representation can be used to evaluate (either from the variance of the PCA factors, or from the box plots of these same factors) the disturbances related to the contributions of anthropogenic origin. The proportion of variability in water quality due to hydrological conditions could be also highlighted by PCA.

FR :

Dans cet article, nous présentons des travaux mettant en évidence les capacités de traitement biologique des eaux résiduaires urbaines au sein des milieux poreux de textures différentes. Une étude comparative du développement de la biomasse couplé aux mécanismes généraux de transferts gazeux à travers deux réacteurs biologiques est menée. Des lits d’infiltration percolation sont simulés par des colonnes garnies de sables d’origine et de structures différentes. Ils sont alimentés périodiquement via un automate de commande avec un influent d’une charge de 525 mgDCO/l et de 54 mgNK/l. Les résultats obtenus établissent le fait qu’un développement équilibré de la biomasse incluant les phases de croissance et de régression est intrinsèquement lié à la nature physique du matériau support. A l’aide des carottes prélevées sur les massifs filtrants et des sondes d’oxymétrie introduites à différentes hauteurs des lits d’infiltration, nous montrons que la répartition verticale du biofilm dans les colonnes ainsi que l’oxygénation des milieux poreux lors des périodes de repos sont également corrélées à la structure des supports pourtant de diamètres moyens similaires. L’efficacité de traitement biologique du carbone est supérieure pour un sable d’origine alluvionnaire comparativement à un sable concassé ; la tendance s’inversant significativement lorsqu’il s’agit de la diminution de l’azote.

EN :

In this article, we present work highlighting the capacity of variously textured porous media to biologically process urban waste water. A comparative study was undertaken that coupled biomass development with general gas transfer mechanisms through two biological engines.

Infiltration/percolation beds are biological systems that treat water using a purifying bacterial culture that develops on a mineral support. Used in domestic wastewater treatment, they are regarded as being well suited to rural areas. These infiltration/percolation beds are easy to use and attain a high quality in their output, two factors that constitute assets for small communities.

Infiltration/percolation beds were simulated in the present work by columns lined with sands of varying origin and structure. They were fed automatically, at set intervals, with waste water containing 525 mg/L of dissolved organic carbon (DOC) and 54 mg/L of Kjeldahl nitrogen. A balanced development of the biomass, including the phases of growth and regression, was intrinsically related to the physical nature of the material used as a support. First, during the supply period, balanced growth of the biomass was quickly reached within the crushed sand. Secondly, within round sand, the regression of the biofilm was less significant and more regular over several weeks. In both supports, the regression was well correlated with an exponential decay. Lastly, the frequency of the supply periods, the organic loads involved, and the rest periods imposed are all factors that contribute to a lack of accumulation of living and/or inert organic matter in the columns.

After obtaining a balanced development of the biomass, the abilities of the columns to reduce the concentrations of carbon, Kjeldahl nitrogen and ammonia were evaluated. Samples of effluent were taken downstream, both before the beginning of the supply period and as it came out of the column approximately thirty minutes after beginning the drainage period. The percentage of suspended matter coming out of the columns gave rather precise information on the scrubbing of the solid mass caused by various shearing speeds or by the structure of the base.

Generally, it appeared that :

- The reduction of the overall DOC was higher than 70%, regardless of which type of sand base was used.

- In the water coming out of the column made up of round sand, the overall DOC content was without exception lower than 125 mg/L, as would be expected for an infiltration-percolation process.

- The treatment of carbon in the columns based on stream sand was in general more effective than that obtained in columns with crushed sand.

Concerning the treatment of nitrogen, crushed sand yielded outputs with a reduction in ammonia that was a lot higher than 80% (mean ± SD: 92 ± 4%) and was systematically higher than those obtained with stream sand (mean ± SD: 72 ± 7%). The suspended matter content was extremely low in the effluent; since no suspended matter was introduced into the effluent, the concentration coming out of the column was the direct result of the biofilm becoming detached and/or, to a lesser extent, of the transport of the biomass in the liquid phase. As could be expected, because of the roughness causing more shearing, more particles became detached when the interior solid mass was made up of crushed sand.

Using core samples taken from the filter's solid mass, as well as oxygen probes inserted at various levels into the biological engines, we showed that the vertical distribution of the biofilm in the columns, as well as the oxygenation of the porous media during the rest periods, were all correlated with the structure of the solid supports (note that all sands had similar average diameters). Gaseous exchanges within the filter's solid mass were dependent on both the type of coating and the depth at which they took place. The oxygen probes were inserted at depths of 14 cm and 18 cm respectively, to determine the percentage of oxygen saturation in the liquid phase within the filter during one week while the columns were operating. Two phases were characteristic of the exchanges observed in the upper part of the filter's solid mass. First we noted an instantaneous reduction in the oxygen content of the liquid phase, linked to the arrival of a batch, followed by a reoxygenation of this portion. Next there was a long phase, lasting approximately 94% of the time between batches, during which time the oxygen content in the residual moisture remained constant. In the lower part of the solid mass, variation in oxygen content was different. Partial deoxygenation in the liquid phase of the porous media here was primarily due to the biochemical oxidation of the organic matter (i.e., consumption by the biomass). It was also due, to a lesser extent, to the augmentation in moisture content that occurs after each batch, and leads to a reduction in porosity and a decrease in oxygen transfer. During the final phase of drainage, a balance was created between the open porous space with air, and the space containing stagnant moisture. The diffusion of oxygen in the former and its transfer towards the latter thus compensated for this consumption, and it therefore remained constant. If the medium was left to dry out longer, the diffusion and transfer phenomena increased, and we observed an increase in oxygen content.

With respect to the vertical distribution of biomass in the columns, we found that it was almost homogeneous inside the stream sand, up to a depth of 28 cm (total depth = 40 cm). Generally, the organic matter content at each point of measurement in the column was higher in the stream sand than in the crushed sand. The variation was greater around 28 cm, where there was a more significant quantity of moisture inside the sand.

FR :

L'objectif de la présente étude est de mettre en évidence les potentialités d'un hélophyte : Phragmites australis (Cav.) Steudel à épurer l'effluent du tannage riche en chrome sous un régime d'écoulement vertical et les comparer à celles d'un sol non planté sous climat aride.

Le pilote expérimental est constitué de pots de capacité 120 litres, remplis sur une épaisseur de 15 cm de graviers et de 60 cm de sol (texture : limono - argilo- sableuse). Trois pots sont plantés de jeunes tiges de Phragmites australis (36 tiges /m²). Trois autres pots non plantés sont pris comme témoin. L'alimentation du système se fait par l'eau du tannage au chrome diluée à 50%. L’alimentation des systèmes se fait par bâchées de 10 litres/jour pendant 3 jours successifs suivie d’une période de repos de 4 jours. L'écoulement se fait par translation verticale à travers le substrat.

L'étude du fonctionnement hydrologique des pilotes testés a montré que le débit moyen à la sortie du système planté est trois fois plus élevé que celui du lit non planté, ce qui permet de traiter un volume d’eau usée plus important ou bien de réduire la superficie nécessaire par équivalent habitant. Aucun signe de colmatage n’a été détecté aussi bien dans les pots plantés de Phragmites australis que dans les pots non plantés.

Le suivi des performances épuratoires du système à Phragmites australis montre que le pH à la sortie des deux systèmes pilotes augmente de deux à trois unités par rapport à celui des rejets du tannage. La conductivité électrique des effluents traités augmente respectivement pour les deux systèmes.

L’élimination moyenne de la DCO totale est de 74% pour le système planté et de 60,5% pour le système non planté. La différence entre les performances épuratoires des deux systèmes pour l’abattement de la DCO totale est statistiquement significative (p<0,05).

Le Cr total subit globalement un abattement de 99% pour les deux pilotes qui fournissent des eaux épurées limpides. En effet, la concentration moyenne en chrome total dans les eaux usées brutes appliquées à l'entrée des différents pilotes varie de 534 mg/l à 1000 mg/l avec une concentration moyenne de 780±196 mg/l. Les teneurs de chrome total enregistrées à la sortie du lit à Phragmites varient de 2,2 mg/l à 3 mg/l avec une concentration moyenne de 2±0,5 mg/l. La concentration en chrome total des eaux usées à la sortie du sol non planté varie de 2,5 mg/l à 4 mg/l avec une valeur moyenne de 2,4±1,3 mg/l.

La présence de Phragmites australis dans le système planté permet de maintenir une porosité suffisante pour assurer la percolation des eaux en traitement et réduire les risques de colmatage. Avec des temps de résidence hydraulique trois fois plus réduit, le système planté assure un abattement du chrome total comparable à celui du système non planté.

EN :

The aim of this study was to investigate the potential of a helophytic plant, Phragmites australis (Cav.) Steudel, to remove chromium from concentrated tannery effluent in comparison to unplanted soil, under arid climate conditions. The experiment was conducted from August 2002 to August 2003. Six plots (capacity: 120 L, diameter: 50 cm) were filled to depths of 15 cm and 60 cm with gravel and soil respectively (texture: 3.5% clay, 8.5% silt, and 88% sand). Three plots were planted with Phragmites australis (Cav.) Steudel. Young shoots of plants (36 stems/m²) were taken from local and natural reed stands. Three unplanted plots were used as controls. The soil for the study was from the Tensift River (Marrakech). The experimental plots were supplied exclusively with diluted tannery wastewater, 3 times a week (approximately 10 L each time). The water flowed vertically through the substratum.

To investigate the capacity of the systems to increase pH, to decrease conductivity chemical oxygen demand (COD), and to remove total chromium, concentrations in the inflow and outflow of each plot were determined during the experiment. System hydrology indicated that for planted systems, the mean flow varied from 18 mL/min (in winter) to 49 mL/min (in summer), with an average value of 35 mL/min. For the unplanted system, the mean flow varied from 5 mL/min (in winter) to 15 mL/min (in summer) with an average value of 12 mL/min. Water recovery at the outflow of the planted system (PP) and unplanted system (NPP) was 40% and 60% respectively. This may be related to the phenomenon of evapotranspiration from the PP system, which was more significant than simple evaporation from the NPP system. No sign of clogging was observed in the planted systems due to the plant rhizomes that ensure a better aeration of the soil. The study of the hydrological operation showed that mean flow in the planted system was three times higher than that in the unplanted soil, which makes it possible to treat a larger volume of wastewater and to reduce the required per capita surface for treatment. The presence of Phragmites australis ensured a sufficient porosity for the percolation of water for treatment.

The pH of the effluent at the outlets of the planted and unplanted systems increased by two to three units in comparison with the raw wastewater at the inlet. The average inflow pH was 5.5; at the outflow of the two systems, the pH was 7.30 and 8.20 respectively for the planted and unplanted soil. The electrical conductivity (EC) of the treated wastewater increased for the two systems. The average inflow conductivity was 54±5.5ms/cm. At the outflow of the two systems, the EC was about 75±14 ms/cm and 68±11 ms/cm respectively for the planted and unplanted systems. The EC removal for all systems varied significantly from season to another (p< 0.05).

The planted system was more efficient in removing total COD (74%) than the unplanted one (61%). There were significant differences (ANOVA) between the PP and NPP systems (p<0.05). Total chromium concentrations at the inflow varied from 534 mg/L to 1000 mg/L with an average content of 780±196 mg/L. At the outflow, total chromium concentrations varied from 2.2 mg/L to 3 mg/L with an average content of 2±0.5 mg/L for the PP system and from 2.5 mg/L to 4 mg/L with an average value of 2.4±1.3 mg/L for the NPP system. Total chromium removal of 99% occurred for the two pilots, which provides limpid purified water. Despite the shorter retention time for the planted system, which was three times less than in the unplanted one, total chromium was almost completely removed.

The aquatic macrophyte plant seemed to play a physical role by enhancing the hydrological conductivity of the substratum and consequently the operation and management of these systems. Phragmites australis (Cav.) Steudel has the ability to transfer oxygen from the rhizome towards the roots through an internal lacunar system, which offers the creation of an aerobic zone around the roots. This aerobic zone allows the proliferation of micro-organisms, which are the principal decomposers of the organic matter in the root zone.

The treatment of tannery effluent by reed beds is a clean approach from an ecological point of view and constitutes a viable economic alternative in comparison to purely chemical approaches and is thus better adapted to the complexity of the systems used in the leather industry.

FR :

Les mines qui sont abandonnées avec leurs installations minières et métallurgiques constituent un exemple représentatif du transfert des éléments métalliques dans l’écosystème. Au Maroc, le site minier d’Aouli est drainé par l’oued Moulouya (l’unique oued de la région) d’une longueur de 520 km avec un bassin versant d’une superficie d’environ 50.000 km². A travers ce travail, nous avons déterminé un certain nombre de paramètres chimiques afin de voir la destinée de certains métaux (Pb, Zn, Cu et Cd) issue de ce district minier abandonné.

Les résultats ont montré que dans l’eau, dans les matières en suspension et dans les sédiments, il y a une contamination métallique due à la mine d’Aouli situé au niveau de la Haute Moulouya ; tandis qu’au niveau de la Basse Moulouya, l’affluent oued Za contribue à cette contamination par un apport anthropique qui résulte en une augmentation des teneurs en cadmium, plomb et zinc. L’effet de cette contamination externe est également mis en évidence par les caractères géochimiques d’une station témoin situé en amont de l’oued loin de toute influence minière.

L’examen des rapports des différents éléments (Cd, Pb et Cu) par rapport au Zn dans les sédiments montre qu’ils sont préservés dans les sédiments en aval (Basse Moulouya). Par contre, au niveau de la Haute et la Moyenne Moulouya, on assiste à une diminution des rapports Cu/Zn et Cd/Zn et une augmentation du rapport Pb/Zn au niveau de la station S3; ce qui reflète l’effet de la mine d’Aouli. L’augmentation des rapports au niveau des stations S1 et S4 sont en relation avec la présence d’algues et qui jouent un rôle important dans la modification des teneurs en Zn dans les sédiments.

L’étude de ces rapports a permis de confirmer que dans notre cas aucun effet notable dû aux apports des affluents n’est mis en évidence sur le contrôle des concentrations des éléments dans les sédiments de l’oued Moulouya. Cette particularité permet de suggérer que dans la cas de l’oued Moulouya la dilution physique est plus importante que la mobilisation chimique et permet d’expliquer le déclin en aval des concentrations des éléments métalliques.

EN :

Mining activity started in Morocco in the 9^th century. Several metals were prospected for but lead was preferentially mined owing to its silver content. The mining history of Morocco can thus be summarized by the history of lead prospecting within the country. Strong mining activity resumed in the 1970s and has been maintained since that time, with lead still being the main metal produced. In 1980 its production reached about 170,000 tons of concentrated lead, which corresponds to 3.5% of the world's production (Wadjinni, 1998). In 1975, the Aouli and Mibladen mines were depleted and in 1986 the Zaida mine was closed. This resulted in large mining districts being abandoned with their mining and metallurgic equipment left on the surface together with all the accompanying waste materials, including tailings. This has created an important source of contamination, progressively releasing trace elements into the environment.

The goal of the present study was to characterize the spatial distribution of the heavy metals (Cd, Cu, Pb and Zn) released from the Aouli mine into the Moulouya River. This river, which drains the Aouli, Mibladen and Zaida mines, is a good example of the pollution impact arising from abandoned mining sites. In order to understand the behaviour and the fate of these metals within the only permanent river in this region, contamination levels were measured in three components of this river, namely: water, suspended solids and sediments.

Due to the presence of mountainous areas (Rif, Middle and High Atlas), the Moulouya River basin is characterized by variable relief. The river basin also contains high plateaus (the Horsts chain) and low plains (the Missour-Outat El Haj and the Taourirte-Guercif basins). This large area can thus be subdivided into three zones designated by the upper, middle and lower Moulouya basins (Fig.1).

The Upper Moulouya basin corresponds to the southwest region of the Oranaise Meseta that is bounded by the High Atlas on the Southeast and by the Middle Atlas on the Northwest. They are composed of two separate Paleozoic massifs (the Bou-Mia and the Aouli). The Paleozoic substratum that outcrops into these massifs consist of pelletic and quartzitic rocks intruded by Hecynian granites and overlain non-uniformly by a Mesozoic cover consisting of Triassic evaporite/clastics and Jurassic and Cretaceous carbonates and shales.

The Middle Moulouya basin is separated from the Lower Moulouya basin to the North by the Yacoubat paleo-high that corresponds to an East-West oriented topographic bulge where the river narrows considerably. The outcropped rock consists mainly of Jurassic carbonates and marls, Cretaceous carbonates, marls and sandstones and Tertiary and Quaternary conglomerates, marls and gypsum.

The Lower Moulouya basin corresponds to a vast plain extending from the Yacoubat High in the South to the Jbel Mazgout and the Beni-Snassen in the North. This area is underlain by Paleozoic granites, Triassic marls and basalts, Jurassic and Cretaceous carbonates, marls and sandstones and Tertiary and Quaternary conglomerates, marls and carbonates. The Moulouya river basin is characterised by a variable climate, which changes from a Mediterranean type in its low portion to a sub-Saharian type in its median portion and to a continental type at higher elevations.

Sampling was carried out during low water level periods. In order to obtain good representative sampling, three samples were taken from each station and were used to measure metal contaminant concentrations in water, suspended solids and sediments. The sediment samples were taken from approximately 5 cm below the surface within the river plain, which is characterized by fine-grained low-energy organic matter-rich sediments.

We defined a certain number of pollution parameters within the Moulouya basin, based on the samples taken from eight representative sites. The areal distribution of metal concentrations in water and in suspended solids shows the existence of two main contamination sources: the abandoned Aouli mine and urban waters. Dissolved Zn concentrations varied from 3 to 30 µg·L^-1 and the two highest concentrations occurred at station 3 near the Aouli mine and at station 8 located in oued Za near the town of Taourirt (Table 1). Cadmium, Cu and Pb concentrations showed similar trends with high concentrations occurring at stations S3 and S8 for Cd and Cu and at stations S3 and S5 for Pb (Fig. 2). These high levels were explained by the effect of Aouli mine for station S3, the Tindit mineralization area for station S5 and urban waters from Taourirt for station S8.

Metal concentrations in sediments reflect the downstream attenuation of the Aouli mine effect (station S3). In fact this station is characterized by high metal concentrations, especially Pb and Zn and to a lesser degree Cd and Cu. For Cu, its concentration in sediments was found to be similar to its concentration in the suspended solids. This is explained by the precipitation of Cu(II) as oxides and hydroxides. The areal distribution of Pb and Cd was similar to that of Zn, suggesting that the Aouli mine is also a source of these metals.

By comparing concentration ratios of different metals with respect to that of Zn, we found that all these ratios remain constant downstream from the Aouli mine, with the exception of the Pb/Zn ratio, which increased at station S3 (Fig. 3). This distribution indicates an effect of the Aouli mine (station S3) and to a lesser degree of the Tindit area (station S4). No notable effect of the Moulouya tributaries on the concentrations of metals was detected.

FR :

Les analyses corrélatoire et spectrale des chroniques de la pluie (entrée) et de débits (sortie) journaliers enregistrés au niveau des trois bassins sud méditerranéens Sebdou, Mouilah et Isser durant un seul cycle hydrologique nous ont permis d’obtenir des informations sur le fonctionnement de ces systèmes hydrologiques. Bien que la structure du signal « pluie » semble présenter les mêmes caractéristiques pour les trois bassins, le signal de sortie « débit » indique que l’oued Sebdou réagit différemment par rapport aux oueds Mouilah et Isser.

EN :

The aim of this work was to show that correlation and spectral analyses can be used to understand the functioning of hydrological systems. Accordingly, a study was carried out on three southern Mediterranean basins: Sebdou; Mouilah and Isser; located in the north western of Algeria. (Figure 1). Correlation and spectral analyses of daily rainfall and discharge rates for one hydrological cycle were carried out.

Simple analysis

Simple analysis of rainfall showed that the correlograms (Figure 2) decreased rapidly for the three basins, reaching a value of 0.2 within 1-2 days. This result indicated that rainfall was a quasi-random phenomenon. The variance density spectrum (Figure 3) showed that the rainfall distribution was not monotonous and presented a “Leigh” signal structure.

The simple analysis of discharge rates indicated that the Sebdou system was different. The correlogram (Figure 4a) decreased quickly, characterising independent events without memory and with non-significant amounts of water. However, the Mouilah and Isser correlograms (Figure 4 b,c) decreased slowly. They represent important memory effects with regulation of significant amounts of water. The spectral band (Figure 5) confirmed that the Sebdou system did not modify the input information. The regulation time was about 5 d for Sebdou, 21 and 43 d respectively for Mouilah and Isser.

Cross analysis

The correlograms (Figure 6) show that the Mouilah and Isser rivers have a great buffering ability. The Sebdou River was characterised by a composite response of the surface flow and an important groundwater flow. The amplitude function (Figure 7) indicated that the Sebdou system had good inertia. The lag time (Figure 8) was 9, 2 and 5 days respectively for the Sebdou, Mouilah and Isser rivers. The amplification and attenuation of the input signal (Figure 10) show that the Sebdou basin is the most karstified system. The non-linearity of the relationship between rainfall and discharge was expressed by the coherence coefficient (Figure 9), which was lower than 1.

FR :

Ce travail consiste en l’élaboration d’une méthodologie systématique qui permet de substituer une modélisation hydraulique simplifiée à une modélisation détaillée d’un réseau d’assainissement. L’approche préconisée est basée sur une analyse multi-paramètre du processus du drainage en milieu urbain. Les paramètres adimensionnels retenus dans cette analyse font intervenir les caractéristiques du bassin versant, les caractéristiques du réseau et celles de la pluie. Pour donner à cette approche un cadre plus général, les auteurs ont mené cette analyse sur des réseaux et des pluies synthétiques couvrant un spectre très large de cas concrets. La méthodologie élaborée a fait l’objet d’une étude de validation sur le bassin No1 de l’arrondissement de Verdun (Montréal). La concordance entre les débits mesurés à l’exutoire du bassin et les débits simulés par cette approche est avantageusement satisfaisante. Les modèles de transformation découlant de cette analyse, permettant le passage d’une modélisation globale à une modélisation détaillée constituent une avancée très significative pour une gestion en temps réel et optimisée des réseaux d’assainissement.

EN :

It is possible to simulate the hydraulic functioning of a given network either with a detailed “microscopic” model at the street section scale or with a global “macroscopic” model which generates total flow rates at the outlet of a basin. The microscopic model is useful when one is concerned with the hydraulic performance of individual conduit sections and the precise locations of problematic areas within a network. Macroscopic modeling is mainly useful when one is interested exclusively by the exit flow rates of a basin. This may be the case in interceptor management where the flow rate is a parameter of the global optimization procedure, within the framework of real time management of regulators. In this case, detailed modeling of a network is unnecessary. On the other hand, detailed modeling requires that a voluminous data base be built and maintained, implying expenditures exceeding the resources of small municipalities.

The present work consists in elaborating a systematic method which allows one to substitute simplified hydraulic modeling for detailed modeling of a drainage network. The approach is based on the analysis of the drainage parameters in an urban environment. The non-dimensional groupings retained in this analysis comprise the characteristics a of the basin, of the network and of rainfall. In order to provide a greater generality to this approach, the authors applied the analysis to synthetic networks and rainfalls covering a wide range of specific cases.

From a practical viewpoint, the objective was to link the peak flow rates and the time to peak of the global model to those of the detailed model of a basin. Two transfer functions have thus been proposed in order to obtain the peak flow rate and the time to peak of the detailed model as functions of the same parameters provided by the global model. The parameters retained for the transfer function are the density of the drainage network and the rate of filling of the network.

In order to implement the proposed method and perform the required hydraulic and hydrological simulations, the authors have used the SWMM program (Storm Water Management Model). 180 simulations have thus been completed, 15 for each type of network and for each type of rainfall. For every combination, one computes the peak flow rate and the time to peak as obtained by the global and detailed models of the basin.

The proposed multi-criterion analysis revealed that the total discrepancy between simplified and detailed modeling of a drainage network is very important. Within the range of the assumed densities, the difference can be as great as 50%. This discrepancy between global and detailed modeling is explained by the storage capacity of the different types of networks. Indeed, for a same given drained area, a network provides additional storage with increasing length of the network and therefore with increasing drainage density.

On the basis of the above finding, a global model cannot replace a detailed model without the use of required corrections for the computation of peak flow rates for a given specific case. In order to resolve this difficulty, it is recommended to use the empirical models proposed herein to reduce the discrepancy and thereby obtain the appropriate corrections.

The proposed method has been subjected to a validation program on basin No.1 of the Verdun borough. Agreement between the flow rates measured at the entrance of the Rhéaume pumping station and those simulated by the proposed method is very satisfactory. The transformation models derived from the analysis allow for the transition from a global model to a detailed one and provide a significant improvement in optimized real time management of drainage networks.

FR :

Nous proposons onze nouvelles variables pour classifier, caractériser et analyser les facteurs de variabilité spatiale des régimes hydrologiques des affluents du fleuve Saint-Laurent au Québec. Ces variables se rapportent exclusivement aux débits mensuels et utilisent quatre (volume d’écoulement, période d’occurrence, durée et amplitude de variabilité intra-annuelle des débits) des cinq critères proposés par Richter et al. (1996) pour caractériser écologiquement les régimes hydrologiques.

L’analyse en composantes principales de ces onze variables hydrologiques a permis d’extraire trois composantes principales significatives après rotation d’axes par la méthode varimax. La première composante principale est associée aux débits saisonniers hivernaux et aux mois d’occurrence des débits mensuels maximums et minimums. La seconde composante est associée aux débits saisonniers printaniers et au rapport entre ces débits et les débits estivaux. Enfin, la dernière composante est associée au coefficient d’immodération (rapport entre les débits mensuels maximums et minimums) et aux débits mensuels minimums. La variance totale expliquée par ces trois composantes, à part presqu’égale, est d’environ 83%. Sur la base des signes de notes factorielles sur les trois composantes principales, les 72 rivières analysées ont été groupées en huit régimes hydrologiques naturels non contigus dans l’espace. Les caractéristiques de chaque régime hydrologique ont été clairement définies.

Quant aux facteurs environnementaux qui influencent la variabilité spatiale des régimes hydrologiques, il est apparu que les six variables hydrologiques associées aux trois composantes principales sont principalement influencées par la température de l’air ainsi que la superficie couverte par les forêts, les lacs et les marais.

EN :

Several classifications of hydrologic regimes have already been proposed in Quebec. However, these classifications are exclusively based upon the magnitude of discharge (annual and monthly discharge, annual maximum and minimum discharge). This hydrologic parameter isn’t sufficient to describe the ecological hydrologic regime. Thus, Richter et al. (1996) suggested five fundamental characteristics to describe hydrologic regimes that regulate ecological processes in river ecosystems :

1. The magnitude of the water condition at any given time. It is a measure of the availability or suitability of a habitat. It defines such habitat attributes as wetted area or habitat volume, or the position of the water table relative to wetland or riparian plant rooting zones.

2. The timing of occurrence of particular water conditions can determine whether certain life-cycle requirements can influence the degree of stress or mortality associated with extreme water conditions such as flood or droughts.

3. The frequency of occurrence of specific water conditions such as droughts or floods may be tied to reproduction or mortality events for various species, thereby influencing population dynamics.

4. The duration of time over which a particular life-cycle phase can be completed or the degree to which stressful effects such as inundation or drought can accumulate.

5. The rate of change (range) in water conditions may be related to the stranding of certain organisms along the water’s edge, in ponded depressions, or the ability of plant roots to maintain contact with phreatic water supplies.

The application of these characteristics requires a daily discharge time series, but these data are not always available. To overcome this difficulty, we propose eleven new hydrological variables exclusively based upon monthly discharge data. These new variables describe four (magnitude, timing of occurrence, duration of time and the rate of change) of the five characteristics of hydrologic regimes suggested by Richter et al. (1996). The eleven new variables are as follows: seasonal discharge coefficients (%); monthly maximum and minimum discharge coefficients (%); median Julian day of occurrence of maximum monthly discharge; median Julian day of occurrence of monthly minimum discharge; spring and winter seasonal discharge ratios; spring and summer seasonal discharge ratios and monthly maximum and minimum discharge ratios.

We have isolated, using principal component analysis (PCA), three significant principal components after varimax rotation. The first principal component was linked with the magnitude of winter discharge and the timing of monthly maximum and minimum discharge. The second principal component was associated with the magnitude of spring seasonal discharge and the spring and summer seasonal discharge ratio. The third component was linked with the coefficient of immoderation (monthly maximum/minimum discharge ratio) and the magnitude of monthly minimum discharge. The three principal components explain, almost weight for weight, about 83% of the total variance. On the basis of signs of loadings for these three components, 72 rivers were analysed and grouped into eight natural hydrologic regimes that are not geographically contiguous. For example, the first hydrologic regime was characterized by high winter discharge (> 12%), timing of monthly maximum discharge in April, high summer discharge (> 54%), high spring and summer seasonal discharge ratios (> 3.5), high monthly maximum and minimum discharge (> 12) and low monthly minimum discharge (< 3%).

We have also analysed the relation between the six hydrological variables associated with the three principal components and the thirteen environmental factors that can influence the spatial variability of these hydrological variables. The environmental factors were as follows: the drainage area (km2); the length of rivers (km); the mean basin slope (%); the forest surface area (%); the swamp and lake surface area (%); the annual precipitation; the seasonal winter (October to March) precipitation (mm); the number of rainy days in the winter; the mean annual temperature (ºC); the mean winter seasonal temperature (ºC); the mean summer seasonal temperature (ºC) and the number of winter days with temperature > 0ºC. The correlation analysis revealed the following mean results:

- The winter seasonal discharge was influenced by the forest surface area (negative correlation) and both annual and seasonal temperature (positive correlation).

- The timing of the monthly maximum discharge was influenced by the length of rivers (positive correlation), the forest and lake surface area (positive correlation) and both annual and seasonal temperatures (negative correlation).

- The spring seasonal discharge was influenced by the length of rivers (negative correlation), the mean basin slope (positive correlation), the forest surface area (positive correlation), the lake surface area (negative correlation), the annual precipitation (negative correlation) and the winter and summer seasonal temperature (negative correlation).

- The spring and summer seasonal discharge ratio was negatively correlated with the drainage basin, the length of rivers, the mean basin drainage, the annual precipitation and the number of winter days with temperature > 0ºC, but was positively correlated with annual and seasonal temperature.

- The monthly maximum and minimum discharge was positively correlated with forest surface area but negatively correlated with lake surface area, annual and seasonal temperature.

- The monthly minimum discharge was negatively correlated with forest surface area but positively correlated with annual and seasonal discharge.

From this correlation analysis, it appeared that temperature was the only factor that influenced the spatial variability of all hydrological variables, followed by forest and lake surface area. The influence of precipitation on this spatial variability was very weak.