Voici les éléments 1 - 10 sur 111
  • Publication
    Accès libre
    Hydraulics and Turbidity Generation in the Milandre Cave (Switzerland)
    AbstractKarst aquifers may convey significant sediment fluxes, as displayed by the intense turbidity peaks commonly observed at karst springs. The understanding of the origin of the suspended solids discharged at springs is key in assessing spring vulnerability and securing drinking water quality. The mechanisms for turbidity generation and sediment transport in karst are however difficult to investigate because of the general lack of access to the karst conduits. These processes have been examined in the Milandre Cave, which hosts a karst drain of regional importance, for more than 10 years by means of turbidity monitoring both inside and at the outlets of this karst system. Additionally, the composition of the suspended load (particle‐size distribution and Escherichia coli content) has been monitored over the course of a flood event. These data are compared against a numerical simulation of the mean boundary shear stress inside the conduit network. The following conceptual model for sediment transport through the system is derived: during minor flood events, most of the turbidity comes from underground sediment remobilization, while during medium to intense flood events, soil‐derived turbidity also reaches the spring. Hydraulics in the epiphreatic zone is tightly linked with autochthonous turbidity generation (mostly during the flooding and the flushing of conduits). In comparison, allochthonous turbidity is associated with finer particles, higher E. coli, and higher UV fluorescence. This improves the overall understanding of turbidity generation and could help the monitoring and forecast of pollution events at drinking water supplies.
  • Publication
    Accès libre
    Conditioning Multi-Gaussian Groundwater Flow Parameters to Transient Hydraulic Head and Flowrate Data With Iterative Ensemble Smoothers: A Synthetic Case Study
    Over the last decade, data assimilation methods based on the ensemble Kalman filter (EnKF) have been particularly explored in various geoscience fields to solve inverse problems. Although this type of ensemble methods can handle high-dimensional systems, they assume that the errors coming from whether the observations or the numerical model are multivariate Gaussian. To handle existing non-linearities between the observations and the variables to estimate, iterative methods have been proposed. In this paper, we investigate the feasibility of using the ensemble smoother and two iterative variants for the calibration of a synthetic 2D groundwater model inspired by a real nuclear storage problem in France. Using the same set of sparse and transient flow data, we compare the results of each method when employing them to condition an ensemble of multi-Gaussian groundwater flow parameter fields. In particular, we explore the benefit of transforming the state observations to improve the parameter identification performed by one of the two iterative algorithms tested. Despite the favorable case of a multi-Gaussian parameter distribution addressed, we show the importance of defining an ensemble size of at least 200 to obtain sufficiently accurate parameter and uncertainty estimates for the groundwater flow inverse problem considered.
  • Publication
    Accès libre
    Numerical simulations and uncertainty analysis for assessing spatial and temporal dynamics in alluvial river-aquifer systems: an application in the context of the 3rd Rhône River Correction
    (2017)
    Gianni, Guillaume
    ;
    ;
    La caractérisation de l’interaction entre les rivières et les eaux souterraines pour des aquifères alluviaux est essentielle afin d’anticiper l’impact des stratégies de gestion de rivière, comme le sont les restaurations de rivières, sur l’ensemble de la ressource en eau et sur sa dynamique. De ce fait, le développement de méthodes et d’approches de calibration permettant une meilleure caractérisation et estimation des paramètres hydrauliques des aquifères et des lits de rivière est nécessaire. De plus, l’analyse d’hypothèses communes, telles que la constance des propriétés hydrauliques des lits de rivière, l’isotropie de la conductivité hydraulique de l’aquifère ou encore la calibration en régime permanent, est indispensable afin d’identifier des biais potentiels dans les prévisions et leurs incertitudes.
    Le chapitre 1 introduit les concepts pertinents, les méthodes, le cadre de recherche, c.-à-d. la 3e Correction du Rhône, ainsi que le but de la thèse doctorale.
    Le chapitre 2 présente une méthode qui permet d’évaluer le caractère transitoire des propriétés hydrauliques de lits de rivière. La méthode est basée sur l’inversion d’une convolution numérique entre les variations de niveau de la rivière et la réponse unitaire, en terme de variation de la surface piézométrique, de l’aquifère. L’estimation du caractère transitoire des propriétés hydrauliques du lit de la rivière est obtenue par des calibrations utilisant des séries temporelles successives de charges hydrauliques. Une analyse synthétique a démontré la fiabilité de la méthode et son application à des données réelles de terrain a permis d’indiquer l’influence d’événements climatiques sur le caractère transitoire des propriétés hydrauliques.
    Le chapitre 3 analyse les conséquences sur l’estimation de l’incertitude de la prévision de l’hypothèse communément faite sur l’isotropie de la conductivité hydraulique de l’aquifère. Il est démontré que les présupposés sur l’isotropie de l’aquifère peuvent causer une sous-estimation de l’incertitude des prévisions sur l’élévation de la surface piézométrique. De plus, il est démontré que la calibration transitoire, en comparaison à celle en régime permanent, permet une meilleure estimation des paramètres hydrauliques de l’aquifère et du lit de la rivière et ainsi de réduire l’incertitude de la prévision.
    Le chapitre 4 présente les prévisions sur l’élévation de la surface piézométrique et leurs incertitudes dans la région de Sion (Suisse) dans le cadre des modifications projetées par la 3e Correction du Rhône. Ce faisant, l’incertitude de la prévision liée à la calibration du modèle numérique hydrogéologique est complétée par la prise en compte, via la modélisation de scénarios, de l’incertitude sur les futures propriétés hydrauliques et géomorphologiques du lit du Rhône. Les résultats montrent que bien que le processus de calibration réduit de manière importante l’incertitude de la prévision, l’incertitude sur l’élévation future de la surface piézométrique, liée au potentielles variations des propriétés hydrauliques du lit du Rhône reste importante.
    Le chapitre 5 résume les résultats des études présentées, fournis des recommandations quant aux approches de modélisations hydrogéologiques d’une manière générale et dans le cadre de la 3e Correction du Rhône., Characterizing river-groundwater interactions in alluvial aquifers is essential when forecasting the impact of river management strategies, such as river restorations, on the overall water resources distribution and dynamics. Therefore, the development of methods and calibration approaches that allow for better identification of the spatial and temporal characteristics of the hydraulic properties of the aquifer and the riverbed are required. Moreover, the analysis of commonly made assumptions, such as constant hydraulic properties of the streambed, isotropy of the aquifer hydraulic conductivity and steady-state calibration, is important in order to identify potential biases in predictions and related uncertainties.
    Chapter 1 introduces the relevant concepts and methods as well as the framework, i.e. the 3rd Rhône River Correction, and the purpose of the Ph.D. thesis.
    Chapter 2 presents a method that assesses the temporal variations of the hydraulic properties of the riverbed. The method is based on the inversion of a numerical convolution between an aquifer unit step response and stream stage variations. Calibrations against successive time series of observed water table variations allow to estimate the transience in the riverbed properties. A synthetic analysis demonstrated the robustness of the method and its application to field data pointed out the influence of climatic events on the transience in riverbed hydraulic properties.
    Chapter 3 aims at understanding how simplifications in modeling practice regarding horizontal isotropy of the aquifer hydraulic conductivity affect the estimated uncertainty of predictions. It is demonstrated that assuming isotropy or fixed anisotropy may cause the predictive uncertainty of the water table elevation to be underestimated. Then, by taking into account the uncertainty in aquifer anisotropy, it is shown that calibration against transient data allows to achieve a better estimation of the aquifer and riverbed hydraulic parameters and to reduce the predictive uncertainty of water table elevations.
    Chapter 4 presents the model forecasting and related uncertainty of the water table elevation in the area of Sion (Switzerland) in the framework of the modifications projected by the 3rd Rhône River Correction. Furthermore, the predictive uncertainty related to model calibration is complemented by scenario modeling taking into account the uncertainties in the future state of the Rhône riverbed hydraulic properties and geomorphologies. The results show that although the calibration process can significantly reduce the predictive uncertainty, the uncertainty in the future elevation of the water table, related to potential variations in the hydraulic properties of the Rhône riverbed, remains important.
    Chapter 5 summarizes the results of the studies and provides recommendations and perspectives regarding hydrogeological modeling approaches in general and in the framework of 3rd Rhône River Correction.
  • Publication
    Accès libre
    Hydraulics and sedimentary processes in the karst aquifer of Milandre (Jura Mountains, Switzerland)
    (2017)
    Vuilleumier, Cécile
    ;
    ;
    Cette thèse vise à caractériser les processus sédimentaires s’opérant dans l’aquifère karstique de Milandre (Suisse), à la fois grâce à des observations de terrain à l’intérieur du réseau de conduits et à l’aide de la modélisation numérique. Un modèle de tuyaux, qui reproduit la physique du système, est développé sur la base de mesures de charges hydrauliques, de débits et de vitesses d’écoulement mesurés dans le réseau spéléologique. Les simulations d’écoulement permettent de calculer la contrainte de cisaillement limite moyenne et la vitesse de cisaillement dans les conduits, qui sont utilisées pour évaluer où et quand l’érosion et le dépôt de sédiment sont probables. Les prédictions du modèle sont comparées à des observations de terrain variées. Dans la rivière souterraine de Milandre, une station de surveillance de la sédimentation a été en fonction pendant 11 ans. La turbidité et la composition des sédiments en suspension ont été surveillées à trois sites dans le système souterrain et à l’exutoire pérenne au cours de la même période. De plus, l’évolution de la granulométrie et de la teneur en bactéries fécales de la charge sédimentaire à la source a été analysée au cours d’un événement de crue. Les simulations sont en ligne avec les observations disponibles. Le modèle conceptuel suivant est proposé : lors d’événements de faible intensité (débit maximal aux exutoires d’environ 400 L·s-1), l’essentiel de la turbidité observée aux sources provient de la remobilisation de sédiments karstiques (turbidité autochtone). La turbidité provenant de la surface (allochtone) peut atteindre la zone saturée avec un délai allant jusqu’à 3 jours, mais dans la plupart des cas elle n’est pas détectée à la source. Quand le pic de débit augmente, le délai entre le pic de crue et le pic de turbidité allochtone diminue, alors que l’intensité du pic de turbidité allochtone augmente. Les événements de crue d’intensité modérée à forte induisent donc une réponse mixte, à la fois autochtone et allochtone aux sources, alors que le signal de turbidité est fortement influencé par les processus de remobilisation de sédiment dans le système karstique. Le modèle a mis en évidence le fait que la contrainte de cisaillement limite est maximale durant le remplissage et la vidange des différents niveaux de conduits du réseau karstique. Cet effet conduit à la génération de pics secondaires de turbidité durant la phase de récession de la crue. Ces pics secondaires de turbidité ont été observés tant dans la rivière souterraine qu’aux exutoires du système. À moyen terme, le modèle prédit que les processus d’érosion et d’accumulation de sédiment sont tous les deux fréquents aux abords de la rivière souterraine. Par contre, dans les galeries épiphréatiques les plus hautes, une accumulation nette de sédiment est prévue par le modèle. D’après les observations, les flux sédimentaires sont principalement contrôlés par l’hydrodynamique du système karstique. Cependant, une composante saisonnière apparaît dans les variations de la concentration de sédiment à la source. Ce cycle annuel est attribué à une disponibilité accrue de sédiments de surface durant l’automne. En terme de composition, une augmentation pluriannuelle de la teneur en phyllosilicate dans les sédiments en suspension est observée. La concentration en phyllosilicate apparaît bien corrélée avec la température des eaux souterraines, autant à l’échelle saisonnière qu’à l’échelle pluriannuelle., This thesis aims at characterizing the sedimentary processes taking place in the karst aquifer of Milandre (Switzerland) both by direct observation inside the conduit network and through numerical modeling. A physics based pipe flow model of the downstream part of the karst system is developed on the basis of measurements of hydraulic heads, flow rates and flow velocities performed in the speleological network. The flow simulations allow to compute the mean boundary shear stress and the shear velocity in the conduits, which are used to assess when and where erosion and deposition of sediments are likely to occur. The model predictions are compared to various field observations. In the Milandre cave stream, a sedimentation monitoring station has been in operation for 11 years. The turbidity and the suspended sediment composition have been monitored at three locations in the underground system and at the perennial outlet over the same time period. Furthermore, the evolution of the grain size and fecal bacteria content of the suspended solids discharged at the spring has been analyzed over the course of a flood event. The simulations are in good agreement with the available data. Overall, the following conceptual model of sedimentary fluxes in the Milandre system is proposed: during low intensity flood events (maximum discharge at the outlets of ∼400 L·s-1), the bulk of the turbidity observed at the springs comes from the remobilization of karstic sediments (autochthonous turbidity). Soil derived (allochthonous) turbidity may reach the saturated zone with a delay of up to 3 days, but is often not detected at the spring. As the peak discharge of the event increases, the delay between the flood peak and the allochthonous turbidity peak shortens and the intensity of the allochthonous turbidity peak increases. Moderate to intense flood events thus yield a mixed autochthonous and allochthonous turbidity response at the springs, while the turbidity signal is mostly shaped by the processes of sediment remobilization in the karst system. The model highlights the fact that the mean boundary shear stress reaches a maximum during the flooding and the emptying of the different levels of conduits of the karst network. This leads to the generation of secondary turbidity peaks during flow recession, which are observed both in the cave stream and at the outlets of the system. In the medium term, the model suggests that both the accumulation and the erosion of sediments are frequent along the cave stream. In contrast, the uppermost epiphreatic passages are predicted to act as effective sediment traps. From the observed data, it appears that the sediment fluxes are mainly controlled by the hydrodynamics of the karst system. There is however a seasonal component in the variations of the sediment concentration at the spring. This annual cycle is attributed to an enhanced soil sediment availability during fall. In terms of composition, there is a pluriannual increase in the phyllosilicate content in the suspended sediment. The phyllosilicate concentration was found to be well correlated with groundwater temperature, both on a seasonal and on a pluriannual scale.
  • Publication
    Accès libre
    Rapid identification of transience in streambed conductance by inversion of floodwave responses
    (2016) ;
    Richon, Julien
    ;
    ;
    Vogel, Alexandre
    ;
    Streambed conductance controls the interaction between surface and groundwater. However, the streambed conductance is often subject to transience. Directly measuring hydraulic properties in a river yields only point values, is time-consuming and therefore not suited to detect transience of physical properties. Here, we present a method to continuously monitor transience in streambed conductance. Input data are time series of stream stage and near stream hydraulic head. The method is based on the inversion of floodwave responses. The analytical model consists of three parameters: x, the distance between streambank and an observation well, a, the aquifer diffusivity, and a the retardation coefficient that is inversely proportional to the streambed conductance. Estimation of a is carried out over successive time steps in order to identify transience in streambed conductance. The method is tested using synthetic data and is applied to field data from the Rh^one River and its alluvial aquifer (Switzerland). The synthetic method demonstrated the robustness of the proposed methodology. Application of the method to the field data allowed identifying transience in streambed properties, following flood events in the Rh^one. This method requires transience in the surface water, and the river should not change its width significantly with a rising water level. If these conditions are fulfilled, this method allows for a rapid and effective identification of transience in streambed conductance.
  • Publication
    Accès libre
    Assessing groundwater quality trends in pumping wells using spatially
    When implementing remediation programs to mitigate diffuse-source contamination of aquifers, tools are required to anticipate if the measures are sufficient to meet groundwater quality objectives and, if so, in what time frame. Transfer function methods are an attractive approach, as they are easier to implement than numerical groundwater models. However, transfer function approaches as commonly applied in environmental tracer studies are limited to a homogenous input of solute across the catchment area and a unique transfer compartment. The objective of this study was to develop and test an original approach suitable for the transfer of spatially varying inputs across multiple compartments (e.g. unsaturated and saturated zone). The method makes use of a double convolution equation accounting for transfer across two compartments separately. The modified transfer function approach was applied to the Wohlenschwil aquifer (Switzerland), using a formulation of the exponential model of solute transfer for application to subareas of aquifer catchments. A minimum of information was required: (1) delimitation of the capture zone of the outlet of interest; (2) spatial distribution of historical and future pollution input within the capture zone; (3) contribution of each subarea of the recharge zone to the flow at the outlet; (4) transfer functions of the pollutant in the aquifer. A good fit to historical nitrate concentrations at the pumping well was obtained. This suggests that the modified transfer function approach is suitable to explore the effect of environmental projects on groundwater concentration trends, especially at an early screening stage.
  • Publication
    Accès libre
    Uncertainty propagation and global sensitivity analysis in multi-layered hydrogeological models of flow and lifetime expectancy
    (2015)
    Deman, Grégory
    ;
    ;
    The main focus of this thesis is the uncertainty propagation (UP) and global sensitivity analysis (GSA) in complex hydrogeological numerical models. Various methods are presented with applications on numerical models for the groundwater flow and mean lifetime expectancy (MLE) in the scope of Andra's (French National Radioactive Waste Management Agency) project for the geological disposal of high-level and intermediate-level long-lived radioactive wastes in a highly impermeable layer from Callovo-Oxfordian age (COX) in France.
    A state of the art is provided for the theory of uncertainty propagation and for the methodologies of sensitivity analyses in a broad sense. Methods for UP are provided from 2-levels Factorial Designs to quasi-random samplings. GSA techniques encompass screening methods such as the Morris Measures and the Derivative-based Global Sensitivity Measures (DGSM), and also the so-called Sobol' indices based upon the variance decomposition of the response of interest. Meta-modelling techniques such as polynomial regression and Polynomial Chaos Expansions (PCE) meta-models are employed as surrogate models for UP and GSA purpose at negligible computational-costs. A comparison of GSA techniques upon various complex analytical test-functions was undertaken with the purpose of determining a relevant method to be employed in the context of “screening” out unimportant variables in computer-intensive, high-dimensional models.
    A numerical model of groundwater flow and lifetime expectancy is employed for assessing the effect of uncertain advection-dispersion parameters and their spatial distributions upon the MLE from a target zone inside the domain. The model is a 2-dimensions synthetic cross-section of the eastern region of the Paris Basin (Meuse/Haute-Marne sector). This model was used as an exploratory tool for sensitivity analysis methods applied upon numerous sets of uncertain hydrodynamic and dispersion parameters in 15 layers. The uncertainty characterizing the permeability-porosity values in aquifer formations encompassing the COX have proved to add much of variability to the MLE calculated from the target zone. The model also served at exploring the effect of the spatial variability of permeability-porosity parameters in two main aquifer sequences on the groundwater flow rates and MLE in the model. The variabilities of the output responses are mainly due to the uncertainty upon the means and variances of the permeability-porosity distributions, as well as the longitudinal correlation lengths, in each sequence.
    Then, a 3-dimensions high-definition hydrogeological model representing the Meuse/Haute-Marne sector in the eastern region of the Paris Basin is a more comprehensive numerical model incorporating realistic geometries, fractures, heterogeneities and discontinuities encountered on field. A sensitivity analysis of the MLE from a given zone located in the middle of the COX layer was performed by perturbing the hydraulic conductivities and porosities values in fourteen hydrogeological formations. The uncertain permeability-porosity parameters in the aquifer formations from Bathonian in the Dogger sequence, and Rauracian-Sequanian in the Oxfordian sequence, have strong and rather non-linear effects on the variability of the output response of interest.
    The methodologies for UP and GSA employed in the present thesis have proved to be very efficient when applied to large hydrogeological models of groundwater flow and MLE. In particular, quasi-random sampling methods offer a flexible frame for providing the uncertainty distribution of the output response of interest at low computational costs. Screening techniques provide a fast estimation for the overall contribution of each input variable to the variability of the output. Meta-modelling techniques such as PCE proved highly accurate in individualising the low- and high-order effects of each input variable upon the output response of interest.
  • Publication
    Accès libre
    Recharge quantification and continental freshwater lens dynamics in arid regions: application to the Merti aquifer (eastern Kenya)
    Avec une population de près de 400'000 personnes, le camp de réfugiés de Dadaab est le plus grand au monde. Il est situé dans la région semi-aride de l’est du Kenya, proche de la frontière somalienne. L’unique ressource permanente en eau potable pour les réfugiés et les communautés locales provient de l’aquifère de Merti, qui consiste en une large lentille d’eau douce souterraine de 250 km sur 50 km, et qui est entourée par de l’eau salée. L’augmentation des volumes pompés, des signes d’augmentation de la salinité de l’eau et l’incertitude sur la diminution du niveau de l’eau souterraine ont fait prendre conscience du risque d’épuisement de la ressource. Ces observations ont abouti à la nécessité de mieux caractériser la recharge de l’aquifère ainsi que la dynamique entre la lentille d’eau douce et l’eau salée environnante.
    Dans un premier temps, une nouvelle méthodologie a été développée pour quantifier la recharge concentrée à travers un modèle numérique, basé sur la physique des écoulements, couplant l’eau souterraine et l’eau de surface. Ce modèle a eu pour but de reproduire les surfaces inondées durant les crues et a été calibré à l’aide d’images satellites. Deuxièmement, la dynamique de la lentille d’eau douce a été investiguée par une série de modèles numériques synthétiques. Ces modèles ont permis d’analyser les effets des taux de recharge et leurs mécanismes (pluie, recharge concentrée) sur la géométrie de la lentille et de comparer ces géométries simulées avec la géométrie réelle de la lentille de l’aquifère de Merti. Ces deux approches ont été contre-validées à l’aide d’un réseau de monitoring de l’eau souterraine composé de vingt stations de mesures à haute résolution temporelle (15 min), installé sur toute la lentille en septembre 2013. Finalement, les résultats de ces trois axes de recherche ont été combinés dans un modèle numérique régional.
    L’approche développée dans cette étude a permis de quantifier une recharge concentrée entre 195 et 329 x 106 m3/a. La recharge diffuse sur la lentille d’eau douce est quant à elle estimée entre 12 et 62 x 106 m3/a. Comparée à ces taux de recharge, l’extraction courante de l’eau souterraine (environ 4.8 x 106 m3/a) est considérée comme durable à l’échelle de l’aquifère. Cependant, la recharge totale est environ 50 à 100 fois plus grande que l’inféroflux traversant le niveau exploité, calculé à l’aide du gradient hydraulique de l’aquifère et des valeurs de transmissivité issues des essais de pompage. Cet écart a mené à postuler la présence d’un aquifère multi-couche considérablement plus épais qu’uniquement l’horizon couramment exploité, mais avec une importante incertitude sur son épaisseur et sur le gradient vertical de salinité entre la lentille d’eau douce et l’eau salée sous-jacente.
    Les modèles numériques ont révélé la très grande inertie de l’aquifère et ont également confirmé que la recharge a principalement lieu de manière concentrée lors des évènements de crue, juste en amont de la lentille d’eau douce. La grande inertie de l’aquifère est cohérente avec les faibles variations des niveaux d’eau et de la conductivité électrique (salinité) observées avec le réseau de monitoring.
    En conclusion, ce travail ouvre de nouvelles perspectives pour la quantification de la recharge en milieu aride à semi-aride lors d’évènements de crue. Il a également permis de mettre en avant la nécessité de poursuivre le monitoring de l’aquifère et de mener de nouvelles investigations sur l’épaisseur de l’aquifère afin de confirmer les résultats de la recharge., The Dadaab refugee camp, the largest refugee camp in the world with a population of approximately 400’000 persons, is located in the arid to semi-arid eastern Kenya, close to the Somali border. The only permanent water resource for the refugees and the host communities comes from the Merti aquifer which consists in a large continental freshwater lens of 250 km by 50 km surrounded by salty water. The increasing groundwater abstractions as well as signs of increasing salinity and uncertainty on the water level depletion led to the necessity to better characterise the aquifer recharge and the dynamics between the freshwater lens and the surrounding salty water.
    Firstly, a new methodology was developed for quantifying the concentrated groundwater recharge through a physically-based coupled surface/groundwater numerical model reproducing inundated surfaces during flood events which is calibrated with inundated surfaces derived from satellite images. Secondly, the dynamics of freshwater lenses are investigated with a series of synthetic numerical models. These models aim to analyse the effect of the recharge rates and mechanisms (rainfall, concentrated recharge) on the freshwater lens geometries and to compare these geometries with the observed geometry of the freshwater lens of the Merti aquifer. These two approaches are cross-validated owing to a groundwater monitoring network of twenty high time-resolution devices installed over the whole freshwater lens in September 2013. Finally, results from these three axes are combined in a regional numerical model.
    The approaches developed in this study allowed to quantify a concentrated groundwater recharge to be between 195 and 329 x 106 m3/y. Diffuse recharge contributing to the freshwater lens is estimated to be between 12 and 62 x 106 m3/y. Compared to these recharge rates, the current groundwater extraction (about 4.8 x 106 m3/y), is considered as sustainable on the regional scale. However, this recharge is about 50 to 100 higher than the axial flow estimated with the gradient and the transmissivities obtained with pumping tests. This discrepancy led to postulate the presence of a multi-layer aquifer much thicker than the currently exploited horizon but with uncertainties on its thickness and the vertical salinity gradient.
    The synthetic numerical model revealed a very high inertia of the Merti aquifer and confirmed that the recharge of the aquifer is mainly controlled by concentrated recharge on flood plains in the upstream area of the freshwater lens. The high inertia of the aquifer is consistent with the very small groundwater level and electrical conductivity variations observed with the monitoring network.
    As conclusion, this work opens new perspectives for the quantification of groundwater recharge in arid to semi-arid areas occurring during large scale flood events. However, it also showed the necessity to continue the monitoring of the aquifer and to carry out further investigations on the aquifer thickness if further exploitations are foreseen.