Volume 10, Number 3, 1997

FR:

L'échantillonnage par valeurs supérieures à un seuil consiste à retenir tous les événements d'une chronique, définis par l'existence d'un maximum local supérieur à un seuil critique. L'étude probabiliste est alors menée par calage de deux lois de probabilité, une sur le processus d'occurrence de ces événements (date des événements), une autre sur la marque des événements (valeur du maximum local), puis par recomposition de ces deux lois pour obtenir la loi de probabilité associée au maximum annuel.

La théorie du renouvellement permet d'étudier le processus d'occurrence d'événements. Les propriétés générales de la loi le plus souvent utilisée, la loi de Poisson (stationnaire ou non), sont présentées, ainsi que des éléments nouveaux concernant la loi Binomiale et la loi Binomiale négative. Ces propriétés sont relatives à la distribution du nombre d'événements sur un intervalle de temps donné, et à la distribution de la durée de retour, définie comme l'intervalle de temps séparant deux occurrences successives d'événements.

Les relations existant entre la loi de probabilité d'une variable et la période de retour de l'événement associé sont ensuite détaillées. Il s'agit d'un rappel de résultats lorsque la variable étudiée est obtenue par sélection d'un ou de plusieurs maximums par an, ou dans le cas d'un processus marqué de Poisson; et d'éléments nouveaux dans le cas d'un processus représenté par une loi Binomiale ou une loi Binomiale négative.

Pour finir, on trouvera les correspondances entre les deux types d'échantillonnage précédents (par maximum annuel ou par valeurs supérieures à un seuil), en terme de période de retour, de distribution et de variance d'échantillonnage.

EN:

The principle of over-threshold sampling is to consider all the events in a time-series that exceed a given threshold. The probabilistic analysis implies estimating two statistical models, one describing the occurrence of events (date of the events), the other describing their magnitude (value of the local maximum). These two models are then combined to obtain the distribution of annual maximum flows.

The theory of renewal processes can be used to study the occurrence of flood events. We present here properties of the well-known Poisson distribution (stationary or non-stationary process), and certain new results for the binomial and negative binomial distributions. These results concern the distribution of the number of events in a given time interval and the distribution of the waiting time, defined as the time span between two successive exceedances of the threshold.

The relationship between the distribution of a variable and its corresponding return period are then studied in more detail. We review the results for the case where the variable of interest is obtained by selection of one or more events per year, or from a Poisson point process. New results are presented for the case of the binomial and negative binomial processes.

Finally, we establish the analytical relationship between the two types of sampling, annual maximum sampling and peaks-over-threshold sampling, in terms of return period, distribution of the annual maximum, and sampling variance.

FR:

Dans le contexte pluviométrique déficitaire de l'Afrique intertropicale, l'hydraulicité des fleuves soudano-sahéliens a beaucoup diminué depuis 25 ans. Les fleuves des régions humides, relativement épargnés jusqu'en 1980, montrent aussi, avec un retard de plusieurs années une hydraulicité qui baisse dans des proportions importantes, alors qu'un retour à la normale du régime des pluies paraît amorcé. Le déficit des apports est pour l'Afrique humide de 16% pour la décennie 80 (365 km3 an-1), contre 7% pour la décennie 70, et pour l'Afrique sèche de 27% (65 km3 an-1) pour la décennie 80 contre 13% pour la décennie 70. Les maximums annuels de crue sont systématiquement plus faibles dans la période récente. En régime équatorial, la crue de printemps est la plus affectée par le déficit hydrologique et la différence entre crue de printemps et crue d'automne a triplé entre les années 1950-1960 et les années 1980. L'ensemble de la région connaît des étiages exceptionnels répétés. En zone soudano-sahélienne, le tarissement s'est considérablement accéléré et montre une vidange des écoulements de base deux fois plus rapide dans la période récente. Le phénomène est également sensible en Afrique humide et traduit un amenuisement des réserves souterraines dû aux effets cumulés des déficits pluviométriques et décalé dans le temps, que l'on retrouve dans la baisse du niveau piezométrique des nappes phréatiques. Un effet mémoire est donc observé sur les écoulements de base, avec pour conséquence un un appauvrissement durable des ressources hydriques. Celui-ci restera indépendant, pour un temps, d'une éventuelle amélioration des conditions climatiques, comme le montrent les observations des années les plus récentes.

EN:

The extent of the current drought phenomenon in Sudano-Sahelian regions no longer needs to be proven. The study of rainfall deficits in intertropical Africa shows a global climatic degradation which has been going on for about twenty five years. These deficits concern also the humid areas of Western and Central Africa and finally all the african rivers draining the flows to the Atlantic Ocean. Various studies have showed that the current phenomenon greatly differed, in intensity and duration, from other 20th century deficit situations (Sircoulon et Olivry, 1986, Lamb,1985, Nicholson et al, 1988). Despite very great fluctuations which could remind pseudocyclic variations, the current data pointed out a global downward trend since 1860 (Olivry, 1987). Moreover, some authors have seen a statistical break in the rainfall conditions in 1970 (Carbonnel et al, 1985). The necessity to obtain common continual data series led to select the period from 1950 to 1990 with a large number of operational rainfall and jauging stations (Mahe, 1993) (figures 2 et 3).

Without expecting a continuation in this trend, or, on the contrary, a return to a humid period, the current drought will obviously have durable consequences on some hydrological parameters. After a relative improvement of rainfall conditions, observed in 1985 and 1986 and even more recently, as shown by the evolution of the Lamb index (1985). African drought and rainfall deficits observed since 25 years have important repercussions on the flow regime of rivers and especially on the annual runoff and the strenght of annual floods. This phenomenon has intensified during the last ten years while, paradoxically, rainfall deficit greatly has diminished. For instance, the upper Niger river ,with the Bani river, (surface basins 250,000 km2) shows an annual runoff deficit of 20% during the 70's and of 46% during the 80's and an annual rainfall deficit respectively of 15 and 20%. In the regional context of Western and Central Africa, the global runoff deficit varies from -7% to -16% for the decades 1970 and 1980 in the humid areas, and from -13% to -27% in the dry areas (deviations calculated in relation to the 1951-1990 mean values) (Olivry,1993 ). The global annual runoff deficit during the 80ís decade reached 430 km3 year-1 for the atlantic african tributaries (figures 4 et 5, table 1).

During the two last decades, flood hydrographs of large river basins have been greatly reduced in volume and duration; maximum values are much smaller. the flood hydrographs showed lower annual maximum discharges and shorter high water seasons than during the preceding period. In the equatorial areas, with two flood periods by year, the datas showed that the spring floods were more affected than the autumn floods by the climatic changes. (figure 6). A frequential study of flood maxima shows a clear break in data series, and emphasizes the danger of flooding estimations which are based upon recent years only. The long term effects of rain shortage are more marked still for the low flow regimes and different studies have showed that, in addition to the immediate effects of the drought, a large reduction of the groundwater storage can explain the persistent fall of the hydrological resource. In the case of recession-depletion period and low flows, the memory effect of the current deficit period is particularly clear. An unusual repetition of exceptionally dry low flows can be observed, especially an abrupt change in the depletion regime, unprecedented in hydrological time series. It also may be observed that a return to more favourable climatic conditions does not involve an immediate return to the earlier hydrosystem . The figure 7 shows two representative curves of the variability of the low flows (absolute minimum daily discharge) between 1950 and 1990 for a Sudano-Sahelian basin (the Bani river) and for a tropico-equatorial basin (the Sangha river, tributary of the Congo river). It is clear that the dry Africa is more affected by drought and rainfall deficits than the humid Africa, but during the last decade the low flows of the Sangha river have been divided by 2 in relation to the low waters observed before 1970 (Laraque et al,1996). The large deficit of the low flows was also observed on the flows of the Ubangi, Congo/ZaÔre, Ogowe, Sanaga, Kwilu, Bandama and Konkure rivers in the humid areas during the last decade (Olivry, 1993) .

In the dry Africa the deficit of low flows is larger still and it has been observed during the last two decades. After the annual flood and the recession, which occur especially as fast as the rainy season ends, there is a classical pattern of flow drop in the rivers in this region. This corresponds to the basin aquifers draining, also called depletion. This depletion is linked to an exponential law : Q_i=Q_o e^-a(ti-to) with the flow Qo on a to day, the flow Qi on a ti day, and (, a coefficient expressed in days-1, according to the physical and geometrical characteristics of the aquifers. The studies show a significant degradation of the depletion.. This phenomenon, resulting from a deterioration of aquifers, was first noted on the Senegal river in 1983 (Olivry, 1987). Since then, it has been observed again and specified, in relation to all the flows in the Sudano-Sahelian region. On a regional scale, it can be considered as a likely model of the progressive decrease of flows, In the Senegal river (218,000 km_ at Bakel), the depletion coefficient has been relatively stable for over seventy years during humid as well as dry periods. The average is 0.0186 day-1 and the variation coefficient is only 0.075 . With the current drought, the alteration in the pattern only becomes significant after 1975. The depletion coefficient rapidly rises, to reach values such as 0.04 day-1 around 1985. The Niger river (120,000 km_ at Koulikoro and data since 1907) shows before 1975 an average coefficient of 0.023 day-1. The current period shows an average coefficient over the first 8 years as being around 0.028 day-1 and then an abrupt increase to values as high as 0.040 day-1 in the beginning of the 90's.The Bani river(102,000 km_ at Douna), tributary of the Niger river, shows a significant evolution (figures 8 et 9). It is easy to give many examples in the Sudano-Sahelian region and the phenomenon is still likely in humid tropical areas, even it is less marked. Thus in the case of the Ubangui river at Bangui (500,000 km_); the average coefficient from 1935 to 1975 was 0.021 day-1. It has been equal to 0.025 day-1 during the last fifteen years. The increase of the depletion coefficient of the Sangha river is lower, but in the equatorial regions, it is often difficult to observe the depletion periods, the flows being affected by rains during a short dry season ( figure 10) . The depletions similarity in the case of different rivers and even of streams which was observed in the flow time series before the current period must be underlined. The ( values of 0.02 day-1 are characteristic of a low water supply deriving from small hillslope aquifers. These aquifers are typical of the intertropical african geomorphology, where there is no large groundwaters. A good relationship appears between the coefficients variations of recession and of drainage ( Olivry, 1976). According to Darcy's law, the deviations from the norm would therefore mainly correspond to the variations in the aquifers width. The hydrogeological datas showed a drop of 10 m and more of the water table above hillslope aquifers in Senegal, Mali and Burkina Faso during the 80ís. Consequently, there is nothing surprising in the fact that the decrease of these small aquifers results in a comparable ´†hyperdepletion†ª in the basins of various areas affected by the same climatic hazards. A return to better sustained discharges during the dry season first requires the aquifers reconstitution. This cannot happen immediately, as shown by the low effect of the last better rainfall on depletion coefficients. There is no exact convergence between the variations of Lamb's rainfall index and the depletion coefficient ones. Through the aquifers, reactions to the low water restitution are slower ; they require a response time of several years and only become effective with the accumulation of similar climatic variations. The groundwater supply deficit in rivers, considered here, which is noticeable in the dry season, is obviously underlying during the flood period and this also explains the low annual runoff and the poor flood maxima we still observe despite some rainfall recovery. The"memory effect" of the droughtness indicates the important part taken by the base flow in the flood hydrographs of large rivers. A specific stream drought, which could be refered to as a "phreatic drought", is to be added to the climatic drought with a pluriannual gap. The hydrogeological laws are such that, under good rainfall conditions, a restoring of the hydrological regime should last as long as it took to bring it to its current degradation.

FR:

Le problème de la sédimentation des barrages en Algérie est connu dans le monde de par de son importance. Devant cet état de fait la mesure de l'envasement des barrages présente un intérêt certain pour la quantification des sédiments et l'exploitation des ouvrages.

L'objectif de cette étude est de contribuer à la quantification de l'alluvionnement afin de mieux cerner le dimensionnement des ouvrages hydrotechniques, d'accroître leur durée de vie et de réduire les coûts de réalisation.

L'étude porte sur un échantillon de dix neuf barrages algériens situés dans des bassins versants répartis sur l'ensemble du territoire. Une première réflexion a conduit à identifier la diversité des caractéristiques hydromorphométriques (indice d'altitude, densité de drainage, indice de pente, coefficient d'écoulement,...) et climatiques (sub-humide, semi-aride, aride). L'évaluation des volumes des sédiments à partir des mesures topo-bathymétriques effectuées à deux dates différentes montre une perte dans la capacité initiale, soit 44 % de perte en volume total, qui arrive à la limite de la durée de vie pour certains barrages.

Ces derniers résultats ont été exploités dans l'objectif de mettre au point un modèle simple d'évaluation du débit solide accumulé dans les cuvettes des barrages algériens. A partir de cette réflexion, nous avons cherché, graphiquement, les liens préférentiels entre la sédimentation, la superficie et les différents facteurs hydromorphométriques. On distingue trois facteurs déterminants : la superficie, l'indice d'altitude et le coefficient d'écoulement. Ces derniers paramètres ont permis de mettre au point un abaque de classification en fonction de la continentalité et le contexte géographique. Cette taxonomie a été confirmée par l'analyse en composantes principales des données hydromorphométriques des systèmes étudiés.

EN:

Deposition of sediments reduces available water reserves by 2 to 5% annually in the man-made reservoirs of the Maghreb countries. The sedimentation problem in the reservoirs in Algeria is known in the world because of its great importance. Measurement of reservoir siltation is important for quantifying deposits and for the management of the hydraulic structures.

The A.N.R.H. (Agence Nationale des Ressources Hydrauliques) first attempted to monitor sediment loads by measuring the transport of suspended particles (solids) at hydrometric stations in a punctual and irregular manner (draw-off by section without measuring the flow velocity), but the real value of a stream's sediment load is not necessarily well reflected by point sampling (TOUAT, 1989; TOUAT 1991). Therefore, small experimental drainage basins (several hectares in surface area) were equipped to study sedimentation in the hydraulic structures solely by slope erosion (RAMDANE, 1989). The specific rate of erosion in catchment areas in northern Algeria is the highest in the Maghreb. Often it is over 2000 T/km_/year in the majority of the Tellian Altas basins, and it reaches 4000 T/km_/year in the Dahra coastal catchments. These latter values are among the highest in the world (DERI and HUSSON, 1976; HEUSCH, 1970). Variation of sediment transport across the wetted cross section was demonstrated by gauging the suspended particles transport (Tss). Measurement of the transport of bed material is not reliable, and often it is estimated as a percentage of the Tss.

The aim of the present study was to improve sedimentation estimates in order to define proper hydrotechnical structures and to increase their lifetime (reducing the costs).

Firstly, we consider information stemming from the topo-bathymetric processing of 19 reservoirs in Algeria. From the temporal point of view, the information is not homogeneous (not synchronous) but has the advantage of integrating all the types of sediments (suspended sediments, bed-load, bank erosion, deposits). The first analysis involved the topo-bathymetric survey of basins done by the A.N.B. (Agence Nationale des Barrages) (DERI, 1977; SOGREAH and A.N.R.H., 1989) whose data were used as the basis for the calibration of our models. These latter data then allow one to predict the lifetime of recent reservoirs. Calibration consists of defining the most influential hydroclimatic or physico-geographic factors affecting the abrasion rate.

We therefore considered 19 Algerian reservoirs located on basins distributed over a large part of the country. A first analysis led to the identification of the variability of the hydromorphometric (elevation index, drainage density, slope index, flow coefficient, etc.) and the climatic (subhumid, semiarid, arid) characteristics of the studied sample. The evolution of the volumes of sediments from two topo-bathymetric measurements, which were carried out on two different dates, shows a variation between 0.017 Mm3/year and 2.490 Mm3/year. This sedimentation resulted in a loss of initial capacity, which caused the termination of life for some reservoirs. The initial capacity of the 19 considered reservoirs was about 1900 Mm3; it was reduced to 1070 Mm3 in 1986. This constitutes a loss of 44% of the initial volume.

These latter results were exploited with the aim of setting up a simple model for estimating sedimentation and classifying the Algerian reservoirs. To this end, we graphed the preferential links between sedimentation, surface area, and the different hydromorphometric factors. Three determining factors were identified: the surface of drainage basin, the elevation index and the flow coefficient. These latter parameters allowed a spatial partitioning of the examined sample into three groups relating to the continentality (elevation index) and to the geographical context: class Ia represents the catchment basins of the high plateau zone (hard and semiarid climate, a low abrasion rate); class Ib corresponds to an intermediate position (with a moderate physico- geographic characteristic, the cross between two climates: semiarid and subhumid); and class II shows the coastal drainage basins that are influenced by the subhumid climate (an important abrasion rate). Essentially, sedimentation within the hydraulic structures depends on the precipitation, the surface area of the catchment, the inflow (flow coefficient) and the relief (elevation index).

In order to validate these first results, we further analyzed the data table corresponding to the hydromorphometric factors and the abrasion rates. The study of such a data table is generally carried out by interpreting the information by means of multidimensional statistical methods (principal components analysis, factorial discriminant analysis, factorial correspondence analysis, etc.). These descriptive methods, which disregard part of the initial information, yield an explanation of the structure of the data table in terms of hydrokinematics (ROLET and SEGUIN, 1986a and b). Principal components analysis, based on multivariate analysis, implies a mode of deductive reasoning (MUDRY, 1991). Application of this approach showed the presence of sub- groups, based on hydromorphometric criteria, which resembled those obtained earlier in the graphical analysis. This method also corroborated the formulated observations in the classification analysis (graphical analysis).

Calibration of the mathematic functions, based on the minimization of errors by the least squares method, on the data series of "abrasion rate surface" according to the observed data of deposits, showed a significant correlation between the observed and calculated values of the abrasion rate. These latter models did not integrate all the determining factors, but they can be useful for predicting the lifetime of recent reservoirs. These last tasks complete the results obtained from the previous models based on the classification (graphical, principal components analysis) and reveal great efficient means.

To validate our models, we extended our application to the other sites presenting neighbouring conditions. We then inserted in our abacus the deposition values for seven Tunisian reservoirs and sixteen Moroccan reservoirs. The three countries (Algeria, Tunisia, Morocco) are situated in the same physico-geographical conditions. The Tunisian and Moroccan reservoirs behave similarly to those in Algeria and present a high abrasion rate.

Another task is the application of the taxonomy already defined (classification abacus model) to recent reservoirs to determine their lifetime. Regarding the obtained results, it turns out that the reservoirs that have a short calculated lifetime must be fitted out first, and a specific protocol for their exploitation must be established.

Finally, we can imagine the use of our abacus to include the bathymetric results of the reservoirs in order to analyze the evolution of the abrasion rate in small catchment areas. It is suggested that further investigation be undertaken in order to better understand the hierarchical organization of the factors that govern sedimentation in the Maghreb.

FR:

Cet article porte sur l'étude de la réaction de nitratation (oxydation de nitrites en nitrates) par voie biologique aérobie avec des bactéries autotrophes nitratantes. Les phénomènes d'inhibition de l'ammoniac sur l'activité nitratante de populations microbiennes issues de boues activées (populations mixtes ou enrichies en bactéries nitratantes) ont été caractérisés.

A l'aide d'une méthodologie faisant appel à la respirométrie, les caractéristiques de la population enrichie ont été définies :

- les conditions optimales de mise en oeuvre sont un pH de 7,8 et une température de 29·C

- les paramètres cinétiques définissant les performances sont QSmax=61 mgN-NO2-/gMVS×h et KS=3,04 mgN-NO2-/l

- cette population présente une bonne tolérance vis-à-vis de NH3, une inhibition de 60% de la respiration des bactéries nitratantes ayant été obtenue pour 11,4 mgN-NH3/l.

Les mêmes effets inhibiteurs ont été observés lors de la mise en culture discontinue de populations mixtes dans les conditions optimales de pH et température. Dans ces conditions de mise en oeuvre, différentes cultures dont les concentrations en biomasse totale ont varié de 0,1 à 2 gMVS/l, en présence de 3 mgN-NH3/l, ont présenté la même vitesse spécifique de nitratation.

Ces phénomènes d'inhibition de la nitratation par NH3 paraissent complexes et fortement dépendant de facteurs environnementaux qui agissent sur la dynamique de croissance de ces bactéries.

EN:

This paper focuses both on nitrification and on inhibition by ammonia of the bacteria responsible for this process in activated sludge (nitrifying bacteria). Nitrification (autotrophic nitrite oxidation to nitrate) is the limiting step in wastewater treatment plants during nitrogen biological treatment (ammonia is known as the major inhibitor).

To assess the effects of environmental conditions on nitrification, respirometric measurements of a nitrifying bacteria-enriched activated sludge were carried out using a chemostat providing a stable nitrite-oxidizing population. The influence of pH, temperature, and the free ammonia concentration [NH3-N] was studied. The maximum activity of the nitrifying bacteria was found at pH 7.8 and at 29 oC, with the 90% range of maximum activity lying between pH values of 7.5 and 8.1. Activation and deactivation energies for nitrification were estimated by Arrhenius analysis to be 12.2 and - 41.5 kcal/mol, respectively. A Q10 value (multiplicative factor of oxygen consumption rate for each 10oC increase) of 2.8 was determined. Saturation or half-velocity coefficient (Ks) and maximum specific rate of nitrite oxidation (QSmax) were calculated by respirometric measurements, and are 3.04 mg NO2--N/l and 61 mg NO2--N/g VSS.h, respectively. In the study of effect of free ammonia, respirometric measurements were performed at ammonia concentrations ranging from 0.1 to 70 mg NH3-N/l. For a concentration of 11.4 mg NH3-N/l, a nitrification inhibition degree of 60% was reached. Contrary to the work of Anthonisen et al. (1976), which showed total inhibition of nitrifying bacteria at a concentration of 1 mg NH3-N/l, our results indicate a residual respiration, even at a concentration of 70 mg NH3-N/l. This displacement of total inhibition could be due to the predominance of nitrifying bacteria in the activated sludge used.

In the second phase of the present research, our results were validated for another mixed bacterial population that was not enriched in nitrifying bacteria. Different concentrations of NH3 were imposed in flask cultures of nitrifying activated sludge, to assess the inhibition thresholds for ammonia. Between 0 and 10 mg NH3-N/l, the degrees of inhibition were inferior to those obtained by respirometric measurements for cultures enriched with nitrifying bacteria, except at 10 mg NH3-N/l, where about 58% inhibition was achieved. The effect of the total biomass concentration of nitrifying activated sludge was also studied. An ammonia concentration of 3 mg NH3-N/l was used in flask cultures with different biomass concentrations (0.1 to 2 g VSS/l). This variable was not significant as similar specific nitrite oxidation rates were measured in all cultures. Inhibitory effects of ammonia on nitrifying bacteria activity seem to be complex and strongly dependent on other environmental factors which modify the dynamics of biomass growth.

FR:

Cet article présente les résultats d'une étude menée sur le traitement d'effluents fortement chargés (DCO > 1,5 kg.m-3) par un procédé à boues activées. L'objectif principal de ce travail est de montrer qu'une alimentation séquentielle d'un réacteur parfaitement agité avec maintien permanent du recyclage des boues et de l'oxygénation permet d'obtenir une bonne épuration tant du point de vue de l'élimination de la pollution carbonée que de la maîtrise de la décantabilité des flocs microbiens. Les essais sont menés sur un pilote de laboratoire constitué d'une colonne à bulles et d'un décanteur, alimenté avec un substrat synthétique. Contrairement à une alimentation continue, la mise en place de cycles d'alimentation permet d'obtenir des flocs de bonne décantabilité de manière constante dans le temps. Ainsi une concentration élevée en biomasse peut être maintenue dans le réacteur. De plus, le suivi d'un cycle 1 h/1 h confirme que la mise en œuvre d'un cycle court n'entraîne pas de variations importantes des performances d'épuration. Par ailleurs, le suivi d'un long cycle (24 h/24 h) montre des variations significatives des concentrations en DCO, polysaccharides dans la phase liquide et exopolysaccharides dans les flocs microbiens. Ceci semble indiquer que des produits microbiens sont relargués des flocs vers la phase liquide pendant l'arrêt de l'alimentation et que des exopolysaccharides sont produits dans les flocs pendant l'alimentation.

EN:

The overall efficiency of an activated sludge wastewater treatment system depends both on the ability of the activated sludge culture to remove the soluble organic pollution and on the efficiency of the separation of the treated water from the biomass. In the treatment of a high-strength wastewater (COD > 1.5 kg .m-3), ultrafiltration or crossflow filtration systems can sometimes substitute for the final clarifier, the separation of the sludge from the effluent water being very good regardless of changes in sludge settleability. Because of the high pollutant concentrations, the biomass must be more concentrated in the bioreactor than in the usual aerobic processes (10 to 20 kg.m-3 rather than 4 kg.m-3), whatever the separation system (settler or filtration system). Thus, with this biomass concentration, the volumetric loading rate can be increased while the organic loading rate is maintained at a low level.

The use of a clarifier is preferable because it is cheaper than ultrafiltration or crossflow filtration systems, which have very high investment and operation costs, on the other hand, and some fouling problems, on the other hand. However, its use requires a very good biomass settleability. Results from the literature show that there is no general relation between sludge settleability and such wastewater treatment operating parameters as dissolved oxygen concentration in the aeration basin, organic loading rates, pH, or low relative influent nitrogen and phosphorus contents. On the other hand, the separation ability of the clarifier depends on the hydraulic configuration of the aerated reactor: biomass settleability can be controlled by completely stirred reactors in series, plug-flow reactors or intermittently-fed systems. Indeed, this spatial or temporal staggering should allow microorganisms to consume their accumulated substrate and thus restore their accumulation capacity. In the case of high-strength wastewater treatment, temporal staggering is better than spatial staggering. Indeed, higher organic loading rates require a higher number of reactors. Thus, the COD concentration significantly increases in the first reactors of the cascade and some toxic or inhibitory problems could appear. For this reason, a temporal staggering was chosen.

Biomass settleability is linked to the extracellular polymers that are the third component of the sludge flocs, after the cells and the water. These polymers are composed largely of microbial exopolysaccharides. Measurements of exopolysaccharides could be used to characterize sludge settleability. A too low level of exopolysaccharides indicates a sludge defloculation problem, whereas a too high level of exopolysaccharides means that sludge settleability could be hindered because of an excessive growth of filamentous organisms.

The purpose of the present work is to show that the use of only one completely stirred reactor, fed in a cyclic way with permanent oxygenation and return sludge flow, leads to good sludge settleability and effluent quality.

The experiments have been performed with a laboratory pilot-plant composed of a 11.3 litre bubble column, which is a completely stirred reactor, and a secondary settling tank of 14.8 litres. The reactor is fed with an equilibrated synthetic substrate consisting mainly of meat extract, saccharose and ethanol (COD:N:P=100:5:1 and 1 kg COD=1 kg meat extract + 0.444 kg saccharose + 0.2 kg ethanol). The feeding cycle is a set of consecutive periods of feeding and starvation. For example, during a 1h/2h feeding cycle, the substrate feeding is continuous during 1 hour and stopped during 2 hours.

The sludge comes from the Nancy-Maxéville wastewater treatment plant and its acclimation to the synthetic substrate begins with an aeration without feeding during 12 hours. Afterwards, the experimental system is fed during 1 hour every 2 hours with the substrate (first concentration of COD=0.1 kg.m-3). Then, the inlet COD is increased by steps of 0.1 kg.m-3 every day until the required concentration is reached. Once the biomass is acclimated to the influent and sufficiently concentrated, the experiment can begin.

In order to show the positive influence of cyclic feeding on wastewater treatment quality, we have carried out one experiment with continuous feeding and two experiments with cyclic feeding (see Table 1). For these experiments, the volumetric loading rate is about 4 kg COD.m-3.d-1. With an appropriate feeding cycle, the settleability of the sludge and the effluent quality remain good: the diluted sludge volume index is 56 cm3.g-1 and 96% of the inlet COD is removed. Thus, the biomass concentration remains high in the reactor (19 kg.m-3). This is not the case with a continuous feeding where the biomass settleability rapidly deteriorates. The principal advantage of sequential feeding is the maintenance of good sludge settleability and a high biomass concentration in the aeration basin. Moreover, during the short feeding cycles, the variations of the concentrations of COD, polysaccharides in the bulk phase and extracellular polysaccharides in the microbial aggregates are very low. This augurs well for system stability.

In contrast, monitoring of a long cycle (24 h of feeding and 24 h of starvation) shows great variations in the concentrations of COD, polysaccharides and extracellular polysaccharides in the system. When the feeding is stopped, the COD and polysaccharide concentrations in the bulk phase increase whereas the exopolysaccharide concentration in sludge flocs decreases.

Reciprocally, after the feeding is begun, the COD and polysaccharide concentrations in the bulk phase decrease, and then remain constant, whereas the exopolysaccharide concentration increases. As 1 kg of measured polysaccharides represents 0.9 kg COD, and as in the bulk phase the COD concentration is equivalent to 4 times the polysaccharide concentration, it seems that microbial products other than polysaccharides are released in the bulk phase during starvation, these products being slowly biodegradable. Extracellular polysaccharides in the microbial aggregates are formed during the feeding phase.

FR:

Les impacts possibles d'un changement de climat induit par l'augmentation de la concentration des gaz à effet de serre sur le bilan hydrique ont été simulés sur un ensemble de bassins hydrographiques situés en Belgique et en Suisse. Le modèle hydrologique conceptuel IRMB à pas de temps journalier a été utilisé à cette fin et les paramètres du modèle ont été optimisés sur chaque bassin. Les bassins ont une taille comprise entre 100 km2 et 1200 km2 et couvrent des régions de plaine aussi bien que de moyenne montagne.

Un même scénario climatique a été adopté pour tous les bassins. Il est principalement caractérisé par une augmentation de la température de près de 3·C et par une légère augmentation des précipitations annuelles. Cette méthodologie a été adoptée afin de montrer les sensibilités respectives des différents termes du bilan hydrique et de les mettre en rapport avec les caractéristiques des bassins étudiés. L'étude s'est focalisée sur l'évolution de l'évapotranspiration et de l'humidité du sol, de l'enneigement, des débits à l'exutoire et des réserves en eau souterraine. Les impacts sont également abordés en termes d'événements extrêmes.

Outre des évolutions qui sont prévisibles pour l'ensemble des bassins étudiés, telles une augmentation de l'évapotranspiration, une diminution légère de l'humidité du sol et une réduction de l'enneigement, les réponses de certains termes du bilan hydrique régis par les caractéristiques du sous-sol des bassins peuvent être sensiblement différentes d'une région à une autre. Ainsi, les bassins caractérisés par une infiltration importante subiraient une évolution favorable de leurs réserves en eau souterraine et des débits de base, alors que les bassins où le ruissellement de surface prédomine verraient une diminution se produire. L'altitude des bassins semble aussi jouer un rôle non négligeable. Tous les bassins de plaine présenteraient une augmentation des débits de crues extrêmes, alors que les bassins de moyenne montagne ne subiraient pour ainsi dire pas ces conséquences négatives.

EN:

By strengthening the so called greenhouse effect, the rise of the atmospheric concentrations of anthropogenic gases, such as CO2, chlorofluorocarbons (CFCs) and methane, will progressively modify the energy budget of the Earth atmosphere and disturb the climate. Temperature at the soil level will rise. Precipitation and air humidity will be modified, inducing a large perturbation of the water cycle and thus of water availability and hydrological extremes. The last International Panel on Climate Change report (IPCC 1994) states that a rise of 0.3 degrees Celsius per decade could be felt in the mean global temperature in the next century. The conclusions of the Second World Climate Conference (1992) pointed out that among the most likely impacts of climate change will be its effects on the hydrological cycle and water management systems. An increase of the incidence of extreme events, such as floods and drought, would cause increased frequency and severity of disasters.

The present paper is a synthesis of several separate published and some unpublished climate change impact studies (Bultot et al., 1988 a and b, 1992, 1994; Gellens, 1991; Schädler et al., 1992; Gellens and Demarée, 1993; Gellens and Roulin, 1996) carried out in Belgium and in Switzerland. All these studies have been done with the same hydrological model and the same climate change scenario. This procedure allows a strict comparison of the sensitivity studies and enables us to identify the common responses and the specific behaviour of the catchments. In this latter case, an attempt to identify the geomorphological origin of the particular responses is possible.

A set of eight catchments in Belgium and belonging to the Scheldt and Meuse river basins has been studied. These individual catchments cover areas from 100 km2 to 1200 km2 and are spread over the country in order to represent the main catchment types. Precipitation ranges from 730 to 1160 mm per year. In addition, three tributaries of the Rhine river in Switzerland have been selected in the low alpine altitude (lower than 1500 m a.s.l.). Precipitation rates are higher for these three catchments and range from 1080 to 1300 mm.

The adopted IRMB (Integrated Runoff Model - Bultot) hydrological model (Bultot et al., 1976 and 1985) is a daily time step conceptual model. It has been designed at the Royal Meteorological Institute of Belgium to simulate the components of the water cycle in medium- sized catchments, i.e. catchments for which the input data, and in particular the precipitation, can be considered as uniform. The main data needed to run the IRMB model are the precipitation and the potential evapotranspiration. This latter variable is assessed by following the procedure described by Bultot et al. (1983) and requires several climatological data, i.e. the net radiation, the air temperature and humidity, the soil temperature at 10, 20 and 50 cm depth and the wind speed at 2 m above the soil. These data are also taken into account in the snow melting- accumulation simulation.

The adopted climate scenario has been constructed from the literature in order to combine the results of various simulations produced by different climate models (Bultot et al., 1988b) in a single set of climate increments. The main characteristics of the climate change scenario is a temperature rise reaching some 3 ¡C, with a higher increase in winter than in summer. Precipitation stress consists of a 50 mm yearly rise distributed as a winter rise (about 10 mm) and a slight summer decrease. Although these scenario increments are small in comparison with the year-to-year variability, they are however large enough to reveal the sensitivity of the water balance to climate perturbations. The detailed description of the algorithms used to apply the monthly increments on the daily time step are presented in Bultot et al. (1988b). After a calibration phase for the present climate conditions, a sensitivity analysis of the water balance of the catchments has been carried out by modifying the input data according to the scenario. This well known if - then - what? method gives the sensitivity of the various terms of the water cycle by comparing their values in the present runs and in the disturbed 2xCO2 runs. For practical reasons, the reference periods of the simulation runs are not the same for all the catchments.

The study focuses on the evolution of evapotranspiration and soil moisture, of snow cover, of streamflow at the outlet and of groundwater storage. The impacts are also studied in terms of extreme events.

For all the catchments, the analysis shows a rise of the evapotranspiration equivalent to some 7 to 10 percent. A small decrease in soil moisture has also been simulated associated with an increase in dry soil days. Due to the temperature rise a strong depletion of the snow cover might be an economically dominant effect in the low alpine regions where winter sport activities represent a large part of the inhabitants' resources. According to the winter precipitation rise, the monthly streamflows in the cold period are also increased under the 2xCO2 conditions.

Besides predictable trends common to all the catchments, the study also shows that some components of the water balance governed by the underground characteristics can present uneven responses. Catchments characterized by strong infiltration could be subject to positive evolution of the groundwater storage and of the baseflow, whereas catchments with predominant surface runoff could exhibit the reverse effect. These effects could be important mainly in summer during the low flow period. The altitude of the catchments also seems to be significant. All the lowland catchments present higher extreme streamflows, whereas catchments in low Alpine regions are spared this negative consequence.

While a large degree of uncertainty remains in the assessment of the climate in the next century, this study gives a first insight into the direction of the expected climate change impacts. It also points out the need to analyse the sensitivity of catchments with a particular attention to their characteristics.