Search

The multiple flow paths existing in urban environments lead to complex flow fields during urban flooding. Modelling these flow processes with three-dimensional numerical models may be scientifically sound; however, such numerical models are computationally demanding. To ascertain whether urban floods can be modelled with faster tools, this study investigated for the first time the capacity of the 2D shallow water equations (SWE) in modelling the flow patterns within and around urban blocks with openings, i.e., involving flow exchanges between the flows in the streets and within the urban blocks (e.g., through alleys leading to courtyards or through broken windows or doors). Laboratory experiments of idealized urban floods were simulated with two academic 2D SWE models, with their most notable difference being the parameterization of the eddy viscosity. Specifically, the first model had a turbulence closure based on flow depth and friction velocity while the second model had a depth-averaged k-ε turbulence closure. Thirteen urban layouts were considered with steady flow and five with unsteady flow. Both models simulated the flow depths accurately for the steady cases. The discharge distribution in the streets and the flow velocities were predicted with lower accuracy, particularly in layouts with large open spaces. The average deviation of the modelled discharge distribution at the outlets was 2.5% and 7.3% for the first and second model, respectively. For the unsteady cases, only the first model was tested. It predicted well the velocity pattern during the falling limb of a flood wave, while it did not reproduce all recirculation zones in the rising limb. The peak flow depths in the streets and the peak discharges at the outlets were predicted with an average deviation of 6.7% and 8.6%, respectively. Even though some aspects of the flow in an urban setup are 3D, the findings of this study support the modelling of such processes with 2D SWE models.

International audience; The flow patterns in a street are investigated in the case of flow exchanges with the sewer. The experimental facility includes one street and one circular pipe located below eventually connected by one or two drains. Compared to the runs without drain or with drainage, overflow creates a strong (up to 50%) increase of the streamwise velocity in the street, particularly, near the bottom, and a high degree of turbulence close to the drains and downstream. For overflow from one drain, the rearrangement of the flow is marked by an asymmetrical pattern with a main secondary current moving across the street; this latter pattern disappears when two drains are present. In field conditions, such an increase is likely to permit entrainment of people and cars by the flow. The experimental data sets could help to validate detailed CFD models aiming at extending the latter results in various flooding conditions.

International audience; A prime requirement for hydrological applications, such as sediment budgeting or numerical modelling, is that produced Digital Terrain Models (DTMs) accurately represent the shape of landforms, especially for river reaches where data are not homogeneous. DTM error is a function of data point measurement accuracy and density and also of the field survey strategy when limited amounts of data will be acquired. This paper aims to advance the importance of the field survey strategy for the specific, but common cases, where only limited topographic data will be available. This methodology is based on the idea that any feature can be properly described by a set of cross sections and breaklines describing both main and secondary directions of the flow. Then, a longitudinal linear interpolation can be applied to the defined homogeneous zones. This morphologically oriented (MO) method that includes data acquisition strategy and interpolation, was validated using a reference DTM derived from LiDAR measurements. An stimation of the uncertainties also is suggested based on the distance of the nearest point and the local slope using a geographically weighted regression. The proposed MO method is typically applicable to Alpine river reaches characterized by multiple channels that may always be underwater and not navigable such as an alternate bar system with secondary and transverse channels.

International audience; The authors discuss the use of a 1-D model or a 2-D model for the evolution of the bed between the levees while the spreading of water and sediment over the floodplains is sim-ulated using a 2-D model. A 1-D model cannot represent the asymmetrical behaviour of the flow in the bends but makes easier the estimate of the variations of the sediment transport capacity along the river if the bed geometry is complex. For the case of the Agly river floods, results in terms of water elevation and bed evolution are compared for a sim-plified geometry of the bed. The two models (coupled 1-D/2-D or full 2-D) provide quite similar processes in the flood plain but different erosion rates of the main channel bed. Because differences of both results to observations are large, one cannot decide which model provides the more relevant results.

L'etude de la dynamique sedimentaire dans les retenues de barrage reste complexe du fait de la combinaison entre les evenements naturels et la gestion du barrage. Dans ce papier, nous montrons l'in...

Physical modelling of urban flooding usually considers that the flow is concentrated in the streets, assuming that urban blocks and individual buildings are non-porous structures. This assumption is open to criticism, as potential lateral flow exchanges through openings in the walls of an urban block (e.g. gates, windows and doors) are likely to modify the flow pattern in the block and adjacent streets. This paper reports an experimental study conducted on a 5.4 m long and 3.2 m wide physical model, representing a rectangular urban block and four surrounding streets. The focus of the study was on investigating the lateral flow exchange processes between the porous urban block and its adjacent streets during flooding, for a variable urban block porosity. Eight porosity values were studied, keeping the same boundary conditions. A porosity equals zero, i.e. a flow around a non-porous block, served as a reference flow situation. For each porosity value, the impact of the lateral flow exchanges on the spatial distribution of flow depth and velocity in the streets and within the block was assessed, as well as the risk to pedestrians. First, it was found that, in the streets surrounding the block, flow depth and depth-averaged velocity can vary by 12% and 70%, respectively, when modifying block porosity. Within the block, the most impacted flow parameter is the number and size of horizontal secondary flow cells. Second, noticeable differences in flow depth can be observed from either side of an opening between street and block. Third, when moving away from the block, flow parameters are little affected. Last, the risk to pedestrians, which is both related to local flow depth and velocity, is dependent on the block porosity value.

A comprehensive understanding of redthe dynamics of erosion and sedimentation in reservoirs under different management conditions is required to anticipate sedimentation issues and implement effective sediment management strategies. This paper describes a redunique approach combining fluvial geomorphology tools and morphodynamic modeling for analyzing the sediment dynamics of an elongated hydropower reservoir subjected to redmanagement operations: the GA©nissiat Reservoir on the RhA´ne River. redFunctional sub-reaches representative of the reservoir morphodynamics were delineated by adapting natural river segmentation methods to elongated reservoirs. The segmentation revealed the link between the spatial and temporal reservoir changes and the variability of longitudinal flow conditions during reservoir management operations. An innovative modeling strategy, incorporating the reservoir segmentation into two sediment transport codes was implemented to simulate the dynamics of erosion and sedimentation at the reach scale during historic events. One code used a bed-load approach, based on the Exner equation with a transport capacity formula and the other used a suspended load approach based on the advection-dispersion equation. This strategy provided a fair quantification of the dynamics of erosion and sedimentation at the reach scale during different management operations. This study showed that the reservoir morphodynamics is controlled by bed-load transport in upper reaches, graded suspended load transport of sand in middle reaches and suspended load transport of fine sediments in lower reaches. Eventually, it allowed a better understanding of the impact of dam management on sediment dynamics.

Flows in a compound open-channel (two-stage geometry with a main channel and adjacent floodplains) with a longitudinal transition in roughness over the floodplains are experimentally investigated in an 18 m long and 3 m wide flume. Transitions from submerged dense vegetation (meadow) to emergent rigid vegetation (wood) and vice versa are modelled using plastic grass and vertical wooden cylinders. For a given roughness transition, the upstream discharge distribution between main channel and floodplain (called subsections) is also varied, keeping the total flow rate constant. The flows with a roughness transition are compared to flows with a uniformly distributed roughness over the whole length of the flume. Besides the influence of the downstream boundary condition, the longitudinal profiles of water depth are controlled by the upstream discharge distribution. The latter also strongly influences the magnitude of the lateral net mass exchanges between subsections, especially upstream from the roughness transition. Irrespective of flow conditions, the inflection point in the mean velocity profile across the mixing layer is always observed at the interface between subsections. The longitudinal velocity at the main channel/floodplain interface, denoted $$U_{int}$$ , appeared to be a key parameter for characterising the flows. First, the mean velocity profiles across the mixing layer, normalised using $$U_{int}$$ , are superimposed irrespective of downstream position, flow depth, floodplain roughness type and lateral mass transfers. However, the profiles of turbulence quantities do not coincide, indicating that the flows are not fully self-similar and that the eddy viscosity assumption is not valid in this case. Second, the depth-averaged turbulent intensities and Reynolds stresses, when scaled by the depth-averaged velocity $$U_{d,int}$$ exhibit two plateau values, each related to a roughness type, meadow or wood. Lastly, the same results hold when scaling by $$U_{d,int}$$ the depth-averaged lateral flux of momentum due to secondary currents. Turbulence production and magnitude of secondary currents are increased by the presence of emergent rigid elements over the floodplains. The autocorrelation functions show that the length of the coherent structures scales with the mixing layer width for all flow cases. It is suggested that coherent structures tend to a state where the magnitude of velocity fluctuations (of both horizontal vortices and secondary currents) and the spatial extension of the structures are in equilibrium.

International audience; During urban floods, the flow is generally concentrated in streets, but lateral exchanges with building blocks can alter locally the flow pattern. These blocks include buildings, courtyards, parking lots, gardens, into which water enters through different openings, such as windows, doors, gates, fences, etc. In order to study these exchange processes and their influence on flow depth and velocity in the streets during flood events, experiments were conducted on a devoted physical model termed MURI (Urban Model for the study of Inundation Risk), at Irstea Lyon-Villeurbanne centre, France. A 2D numerical model was also used to simulate the experiments, using a constant eddy viscosity and then an eddy viscosity in relation to the flow depth. The present study focusses on two geometric configurations. The first one consists of a straight street flanked by two lateral cavities, and the second one, a straight street flanked by one-side opening. These flows are compared to a reference configuration under quasi-uniform flow conditions (without openings). The results obtained with the numerical model are consistent with the experiments, mainly in terms of flow depths. The experiments and simulations show that a detailed description of the building blocks is required if the flow pattern is estimated locally.

International audience; The risk assessment for extreme floods is a key point to anticipate and to dimension a set of measures to improve the resilience of the territories. Because extreme floods can seldom be observed, the calculation of hazard in the flooded areas comes from an extrapolation of models calibrated on more frequent events. This latter extrapolation can be based on a detailed representation of the processes that may influence the flood features or on a more simplified approach that permits much faster calculations. For the specific cases of the Rhône and Gardons valleys in the Southern France, an estimation of the uncertainty is carried out in order to compare a detailed representation and a simplified approach. called "Additional Depths Method" that guesses the flooded areas of a reference flood are accurately estimated. Then, the topography of the valley above the reference flood level permits to define a set of cross sections in which a uniform flow is calculated for a flow discharge calculated as the difference between the extreme flow and the reference flow. Finally, the limits of the flooded areas are drawn starting from the water elevation interpolated between the values obtained at the cross sections. This simplified approach is encapsulated in a plug-in of the QGIS software. The uncertainty linked to the simplified approach can vary a lot from as low as 10% up to 100%. The plug-in is convenient in the more favourable cases in which the hypothesis of a 1D flow can be kept and the topography agrees with the reference map.

International audience; A multi-technique approach is proposed to study sand dynamics in an engineered piedmont river. Only a few studies focused on such systems and innovative methodological protocols still need to be designed to better understand sand transport in piedmont rivers where bedload dynamics has been largely modified with nowadays a residual sand transport on a fixed gravel matrix. The proposed methodology is based on an analysis of bathymetry and turbidity measurements and on modelling, including the development of sediment rating curves and 2D numerical modelling, and using sediment budgeting for cross-validation. Its application to the Isère-Rhône confluence (France) provided some insights of sand fluxes in this complex river system where few sediment flux data are available. Indeed, a substantial amount of sand sporadically reaches the downstream part of the Isère River because of the presence of a series of dams, and jeopardizes navigation and flood management at the confluence. Based on the analysis of the 2015 flushing event, it was found that the sediment transport capacity was reached during the event whereas sand supply can be considered as null when dam bottom gates are closed. Suspended load of sand was prevailing downstream of the last dam but quickly settled down at the confluence. The sand deposit was eventually evacuated from the confluence during the small floods occurring after the flushing event with a minimum discharge of approximately 500 m3/s in the Isère River and 1000 m3/s in the headrace canal of the Rhône River. The presented methodology can be transferred to other sites with similar issues.

Open channel flows subjected to a longitudinal transition in roughness, from bed friction to emergent cylinder drag and vice versa, are investigated experimentally in an 18 m long laboratory flume. These are compared to uniform flows subject to (1) bed roughness only and (2) an array of emergent vertical cylinders installed on bed roughness. The nearbed region is investigated in detail for uniform flows through the cylinder array. The water column can be divided into two parts: a region of constant velocity and a boundary layer near the channel bed. In the latter region, a local increase in velocity, or velocity bulge, is observed in line of a cylinder row. The velocity bulge may be related to the disorganization of the von Karman vortex street by the bed-induced turbulence, resulting in reduced momentum loss in the cylinder wake. The boundary layer height is found to be independent of water depth and bed roughness (smooth or rough bottom). Strong oscillations of the free surface (seiching) are observed. Oscillation amplitude is dependent on the longitudinal position within the cylinder array and is found to decrease with decreasing array length. When water depth/boundary layer height ratio is close to unity, the disorganization of the von Karman vortex street throughout the water column prevents seiching from occurring. In the case of roughness transition flows, the water depth is found to vary only upstream of the change in roughness. Vertical profiles of velocity and turbulence are self-similar upstream of the transition and collapse with the uniform flow profiles. Downstream of the roughness change, velocity and turbulence vary over a distance of 35 to 50 times the water depth. Roughness transition flows show that seiching is lowered by flow non-uniformity. A 1D momentum equation integrating bed friction and drag force exerted by the cylinder array predicts accurately the water surface profile (0.9% mean relative error). The computed profiles show that, upstream of the transition, flow depth varies over a distance of about 2600 times the uniform water depth of the upstream roughness. The 1D equation is solved analytically for zero bed friction.

International audience; This paper reports two urban floods in France: the September 2000 flood at the city of Marseille caused by heavy rainfall, and the November 2008 flood in Oullins induced by the Yzeron River overflow. A two-dimensional (2D) depth-averaged model was used to simulate flood hazard and to investigate the uncertainty associated to the forcing inputs. Using a rainfall simulator, several distributions of rainfall inputs over the whole domain were produced for the event of September 2000 in Marseille. Results showed these latter distributions changed the average peak water depth more than other input parameters. The Yzeron River water elevation was the main source of uncertainty in the case of Oullins, but the flow pattern in the floodplain was also influenced by the method representing the urban areas. For both cases, average uncertainty remains high (up to 20% for peak water depth or velocity) and local uncertainty is almost much higher.; Cet article traite de crues urbaines en France: la crue de septembre 2000 à Marseille et la crue de novembre 2008 à Oullins. Un modèle 2D a permis de simuler l'aléa d'inondation et étudier l'incertitude associée aux variables d'entrée. En utilisant un simulateur de pluie pour Marseille, on a pu montrer que la répartition de la pluie influait plus que les autres paramètres. A Oullins, le niveau d'eau dans la rivière était le principal paramètre mais le parcours de l'eau dans la zone urbaine était aussi important. Dans les deux cas, l'incertitude restait en moyenne de l'ordre de 20% sur les hauteurs d'eau et les vitesses maximales mais pouvait localement être bien plus forte.

International audience; Sediment load and budgets are a fundamental component of the process-based hydromorphological framework developed by the REFORM project, and are needed to accurately assess the current condition of a river, its sensitivity to change, and its likely future evolutionary trajectory. This paper presents an evaluation of three different methods for estimating both bedload sediment transport and bed-material budget within river channels, using the Middle Loire River as a case study. The first method is based on the stream power concept and does not need any hydraulic calculations. It yields estimates of the sediment transport in the same order of magnitude as measurements but poor results for the bed-material budget in terms of magnitude and tendency. For the second method, hydraulic parameters are computed using the Manning–Strickler equation (or a 1D hydraulic model for steady flow). It provides useful indicators for understanding river dynamics but does not yield significant improvements compared to the first method. The third method uses 1D numerical software for water flow and river bed evolution. It yields the most accurate results for both sediment transport and bed evolution but requires more data and overall more work to construct the model. Guidance is provided on the amount of data required, the competence needed to build the models, and the predictive capability of each of the methods.

International audience; In this paper, the results of a benchmark test launched within the framework of the NSF–PIRE project “Modelling of Flood Hazards and Geomorphic Impacts of Levee Breach and Dam Failure” are presented. Experiments of two-dimensional dam-break flows over a sand bed were conducted at Université catholique de Louvain, Belgium. The water level evolution at eight gauging points was measured as well as the final bed topography. Intense scour occurred close to the failed dam, while significant deposition was observed further downstream. From these experiments, a benchmark was proposed to the scientific community, consisting of blind test simulations, that is, without any prior knowledge of the measurements. Twelve different teams of modellers from eight countries participated in the study. Here, the numerical models used in this test are briefly presented. The results are commented upon, in view of evaluating the modelling capabilities and identifying the challenges that may open pathways for further research.; Les résultats d'un test comparatif de modèles numériques sont présentés. Le test a été effectué dans le cadre du projet NSF-PIRE sur les impacts de crues extrêmes et les expériences de rupture de barrage 2D sur fond sableux utilisées ont été effectuées à l'université Catholique de louvain en Belgique. Les niveaux d'eau en 8 points ainsi que les topographies initiale et finale ont été enregistrés. Des simulations en aveugle de cette érosion intense ont été effectuées par plusieurs équipes de chercheurs. Les résultats sont commentés en terme de capacité et voies de recherche.

An analysis of the long term morphological evolution of the Génissiat reservoir (France) is provided. First, a methodology for bathymetric data processing and reservoir sediment volume budget calculation is described. An estimation of global uncertainties in volume calculation is proposed. The reservoir bathymetric budget for several dam flushing events and interflush periods is presented, showing the global decrease of deposited sediment volume with time. The spatial dynamics of the reservoir subreaches is highlighted and typical patterns in flush and interflush periods are identified.

The literature is rich with studies and models proposed to estimate soil erosion. Among many models, the Universal Soil Loss Equation (USLE) (Wischmeier and Smith, 1978) remains the most widely used model to estimate soil erosion. This paper aims to map the soil loss using this equation under a GIS environment in the Ksob watershed, this latter is a sub-basin of the fifth large basin of Algeria (Hodna) with an area of 26,000 km2 and located in the center of Algeria

International audience; This paper introduces an 1D numerical code RubarBE for hydraulic and mobile-bed simulations. The code's ability to reproduce the downstream fining of a gravel-sand mixture in response to bed aggradation is tested against laboratory experiments. Unlike in most numerical models, grain size distribution in each sediment layer is not represented using a multi-class model, but using the median diameter d50 and a sorting coefficient σ. The comparison of numerical results with experimental data shows that the adaptation length La, classically used for non-equilibrium sediment transport, is an essential parameter of the model to accurately reproduce the evolution of the deposit front. Empirical laws for adjustments of d50 and σ are proposed to reproduce sediment sorting through two grain-size related adaptation lengths (Ld, Lσ ). They are scaled by the length of the reach in morphological equilibrium, which is a useful result for field applications.

Sewer overflow is one of the main causes of urban floods. Using an urban model called MURI recently built by Irstea, a series of experiments are carried out in order to simulate overflowing cases. The first configuration that was studied is a single street of 5.4 m in length, 0.15 m in width with a slope of 0.1% and two street inlets (of which the orifice diameter is 14 mm) that ensure the connection with the sewer network. Inflows of 2 and 3 L/s are used for the street and an inflow of 0.5 L/s for the sewer network that is partially closed in order to obtain overflow at the street inlets. The rising of the weir downstream the street permitted to obtain same water depth and to measure the vertical profiles of velocities. This study allowed us to investigate how the overflow process influence the water depth and the velocity field around the street inlet. The exchange flow discharge can be calculated using either an orifice equation or the head loss in the two pipes connecting the street and the sewage network, both methods requesting calibration, which is made difficult by the complexity of the flow around the street inlets.

Urban floods cause high damages and thus mapping of flood risk is essential for protection and evacuation planning. To obtain such mapping, hydrodynamic modelling is central because it provides water depths and velocities everywhere and along the whole flood period. The case of the 1988 flood in Nîmes is presented as an example of the method to get the basis for evaluating flood risk. Once the hydro-meteorological event is selected, a lot of causes of uncertainty still exist, which are linked to either the flow processes in a complex environment or the simplifications of the 2-D models that are used for saving computation time. In order to gain information regarding the origin of such uncertainties, two series of laboratory experiments are described: one focuses on the error in the flow distribution at crossroad, the other one stresses the flow complexity in the exchange structures between the streets and the sewer network. Beyond the analysis of the causes of uncertainty, obtaining a detailed local assessment of flood risk requires a sensitivity analysis to cope with the associated uncertainty ranges; this analysis may be reduced if a relevant model is used, which means, for instance, that the processes described in the experiments are conveniently modeled.

This study aims to show the ability of 1D model to simulate the wadi flow during the flood. The wadi is a semiarid stream characterized by the non uniform geometry and an intermittent flow. These complexities can be modelled by a robust 1D model such as Rubarbe. In this article we study the 1994 flood of Ksob wadi that caused a large inundation of M’sila town in the center of Algeria. The simulated reach located downstream the Ksob dam, crosses M’sila over a length of 6800 m with a bed slope of 0.6 %. The recorded flow peak of the 1994 flood is about 1430 m3/s and flood duration is 24 days. The flood propagation is simulated through the studied reach using the Rubarbe code (1D model) developed in Irstea (former Cemagref). The results show that the wadi overflows in some sections at the upstream and the center of the reach. The major part of the overflow is observed just around the peak of the flood, but the duration is variable from a few minutes to several hours. The overflow hazard limits are plotted in each section from the intersections of the highest water elevation recorded during the flood with the ground level of the two banks of the same section, taking into account the floodplain slope. According to our investigations, the calculated flooded area is very close to the real one, especially in the central part of the reach., L'étude vise à montrer la capacité d'un modèle 1D à simuler l'écoulement dans un oued. L'article traite le cas de l'oued Ksob qui a entrainé l'inondation de la ville de M'sila en 1994 et détaille les étapes qui ont mené à une carte des zones inondées proche des observations pour cette crue de 1994.

International audience; This paper presents a paleo-hydrological study conducted on sediment flood deposits at two sites located in the Gardon Gorges, Southeast France. The first study site is a valley‑side terrace (referred to as site GE, located 10m above the channel bed) and the second is located in a cave (referred to as site GG, perched 15m above the bed level). Paleo‑flood events were identified by carefully analyzing each slackwater flood deposits. The flood units were then dated using a variety of dating techniques, namely 137Cs, 210Pbex, 14C geochemical analysis of mining‑contaminated slackwater flood sediments, archeological objects and the historic flood record. Hydraulic modeling was also carried out in order to verify the coherence between the stratigraphic records of flood deposits at the sites and the historic flood levels known along the river Gardon. It is demonstrated that all the major flood events of the 20th century can be identified at terrace GE. This type of geomorphological unit can therefore be used as a sound basis for a paleo-hydrological analysis of the Gardon Gorges. Furthermore, the analysis of the stratigraphic record at site GG demonstrates that the site can be used to trace the largest flood events that have occurred within the past 600 years. The study reveals that the frequency of extreme flood events increased during the Little Ice Age; Une étude paléo-hydrologique a été entreprise dans les Gorges du Gardon, où deux coupes sédimentaires ont été étudiées. La première provient d'une terrasse (coupe GE située à 10 m de hauteur au dessus de la rivière) et l'autre, d'une cavité (coupe GG localisée à 15 m au-dessus de la rivière). L'identification et la mise en évidence des différents paléo-événements de crue ont été établies grâce à l'inspection minutieuse de chaque dépôt sédimentaire et d'une étude géochronologique fine (137Cs, 210Pbex, 14C, métaux traces, objets archéologiques, historiques des crues). Nous avons souhaité évaluer la capacité des terrasses fluviatiles du Gardon à enregistrer des paléo-événements en comparant les coupes sédimentaires aux données historiques des hauteurs d'eau et aux résultats d'un modèle hydraulique 1D. Nous montrons que la terrasse GE enregistre la quasi-totalité des évènements du 20ème siècle et qu'à priori, ce type d'objet géomorphologique peut être utilisé pour toute étude paléo-hydrologique sur le Gardon. De plus, l'étude de la coupe GG (cavité perchée à 15 m) permet de reconstituer les évènements intenses au cours des 600 dernières années. Nous montrons une augmentation des crues extrêmes dans le Haut-Languedoc au cours du "Petit âge glaciaire".

The chapter starts by describing the physical processes typical of floods, particularly, the ones that cannot be observed at low river flows. All these latter processes are linked to the various kinds of floods described above. In order to simulate these processes for flood management, a relevant spatial scale should be selected. For each kind of flood and spatial scale, one – dimensional and two - dimensional classical hydrodynamic models are explained. Limits and requirements of these latter approaches permit to introduce the recent evolutions in modelling and in available data, which permits to obtain more realistic and accurate results., Le chapitre commence par une description des phénomènes physiques propres aux inondations, en particulier, ceux qui les distinguent des écoulements ordinaires en rivière. Ces différents phénomènes sont rattachés aux différents types d’inondations qui ont été énumérés auparavant. Pour modéliser ces phénomènes en vue de la gestion des inondations, il est, entre autres, nécessaire de définir l’échelle spatiale pertinente. En fonction de cette échelle et du type d’inondation, on explique les modélisations hydrauliques classiques 1D et 2D, leurs limitations et leurs besoins ainsi que les dernières évolutions des modèles et surtout des données disponibles qui permettent d’obtenir des résultats toujours plus réalistes et parfois plus précis.

International audience; This laboratory study aims at investigating the longitudinal development of a mixing layer in a compound open-channel (two-stage geometry with a main channel and adjacent floodplains). The floodplains are covered with two roughness types: either a bed roughness representing a submerged dense meadow or emergent roughness elements (cylinders) representing an alluvial forest. The theoretical background used for plane mixing layers is adapted to the highly three-dimensional mixing layer that develops at the main channel/floodplain interface. The mixing layer width is divided into two parts on either side of the interface. For the wooded floodplain, the mixing layer width on the floodplain side levels off downstream much more rapidly than for the grassed floodplain. The lateral profiles of normalised velocity and turbulence quantities are found to be self-similar in the longitudinal direction for a fixed elevation. However, shallowness effects prevented the normalised lateral profiles of velocity and turbulence quantities from coinciding at different elevations. The respective contributions of lateral Reynolds stresses and secondary currents to the lateral exchange of momentum are estimated. At the main channel/floodplain interface, the momentum exchange is driven by Reynolds stresses. In the main channel, both Reynolds stresses and secondary currents contribute to the lateral flux of momentum. Secondary currents are stronger with emergent macro-roughness elements than with bed-roughness only on the floodplains. Large-scale turbulent coherent structures are investigated based on two-point space-time correlations of velocity. These structures are found to span the entire floodplain flow depth, and their convection velocity is close to the depth-averaged longitudinal velocity at the interface. The coherent fluctuations of the longitudinal and lateral velocities have different Strouhal number values, similar to those found in plane mixing layers.

International audience; Urban flood flow characteristics are usually computed using two-dimensional numerical models. How such modelling can be implemented in dense urban areas with obstacles is investigated in this paper. A strategy for representing the effect of urban obstacles in various flow conditions is proposed. The comparison between the available laboratory measurements and the model results show that if the water depth is high enough and the flow remains subcritical, two-dimensional modelling with constant eddy viscosity represents the effect of the obstacles on the flow distribution accurately, even with a coarse mesh. In contrast, if the water depth is low and/or the flow becomes supercritical, the description of the flow is not sufficient and it generates errors in the flow distribution at the crossroads.

International audience; In an urban flood, interaction between surface runoff and flow into the sewer system can take place. Considering this latter interaction is necessary to understand the phenomenon of urban flooding. A laboratory Model to Study the Flood Risk (MURI) is being built, which includes one lower floor for the pipes representing the sewage network and a second floor for the flow along the streets. Before the experimental study, a numerical investigation of MURI is carried out. The main objective is to obtain the main features for a few representative scenarios of the MURI experiments and to establish an experimental protocol taking into account the capacity of the measuring devices. For this purpose, three numerical models have been successively developed: a 2D model for the second floor using Rubar20 software, a 1D model for the pipe network using Rubar3 software and a coupled 1D / 2D model to simulate both latter flows and the flow interactions between the two floors. Supercritical flow conditions partly appear on the second floor, which implies that results are uncertain because it is difficult to locate transitions accurately, particularly if the mesh is not refined. Opening the exchange structures between floors strongly changes the flow distribution and locally the water depths. One can expect that the numerical model will be a useful tool to plan the experimental runs

This paper summarizes the major hydrologic studies of the sediment transport in Algeria through a case study of Soubella watershed in Hodna basin (Algeria). Since the 70s, the Algerian Agency of Water Resources (ANRH) has installed a large number of gauging stations, which allowed carrying out many studies about sediment transport. The major contributions aimed principally to model and estimate the suspended sediment transport. Through these studies, the sediment-rating curve is the large discussed model; it is the best significant model for all the basins presented here below. Wadi Soubella watershed (183.5 km2) is characterized by an aggressive, irregular and violent rainfall. The mean annual precipitation is 288.5 mm (associated with a coefficient of variation Cv = 33% observed during 27 hydrological years 1973–2003). The results show a high specific suspended sediment yields (126 t km−2 year−1) associated with a higher interannual variability (Cv = 61%). Over the 20 years of this study, total sediment flux is mainly observed during two months: September (45.12%) and October (24.13%). Thus, the early autumn floods are responsible for most of the sediment transport. In Soubella watershed, the sediment-rating curve explains more than 75% of the variance for the whole data (pairs Q-C) at the instantaneous, monthly and seasonally scale and only 70% for the annual scale. A significant regressive relationship (logarithmic) between the sediment-rating curve coefficients (a, b) is obtained at the monthly scale for wadi Soubella and other wadis in Algeria.

The Authors thank the Discussers for extending the discussion on the selection of most appropriate 2D models to simulate urban floods. Regarding classical 2D models, the Authors used a constant tur...

Comprehensive measurements of bedload sediment transport through a section of the Danube River, located approximately 70km downstream from Bratislava, Slovakia, are used to assess the accuracy of bedload formulae implemented in 1D modelling. Depending on water discharge and water level, significant variations in the distribution of bedload across the section were observed. It appeared that, whatever the water discharge, the bed shear stress tau is always close to the estimated critical bed shear stress for the initiation of sediment transport . The discussion focusses on the methods used in 1D models for estimating bedload transport. Though usually done, the evaluation of bedload transport using the mean cross-sectional bed shear stress yields unsatisfactory results. It is necessary to use an additional model to distribute the bed shear stress across the section and calculate bedload locally. Bedload predictors also need to be accurate for tau close to tau. From that point of view, bedload formulae based on an exponential decrease of bedload transport close to tau appear to be more appropriate than models based on excess bed shear stress. A discussion on the bedload formula capability to reproduce grain sorting is also provided.

International audience; An experimental study of the characteristics of dividing critical flows in a 90° open-channel junction formed by three horizontal equal-width branches is presented, conducted for various inflow discharges and downstream boundary conditions. Outflow discharges and flow depths were also measured. Four main flow patterns are identified considering the location and length of the hydraulic jumps that develop across the main and lateral channels. A relationship between the discharge division ratio and the tailwater Froude number is found, reproducing well the experimental data. Practical applications of this relationship include the design of open channel and pipe networks and one-dimensional numerical modelling of flood propagation in urban areas. The study shows also that a two-dimensional depth-averaged numerical model hardly gives better predictions of the discharge ratios because of the presence of three-dimensional features at the junction.

During severe urban flooding, the surface flow can become supercritical near some crossroads if the slope of the streets reaches a critical steepness. The present study aims to validate the capacity of a code solving the two-dimensional shallow water equations to simulate the flows in a four-branch supercritical cross junction. An experimental study showed that five main flow patterns can be observed and are determined by the upstream flow characteristics as well as the slope of the channels. Computed and measured flows were compared on five detailed water depth fields and on more than 200 measured flow rate distributions. It appears that the flow rate distribution to the downstream branches and the flow characteristics, computed by the code using a set of reference parameters, compare well with the experimental data. Nevertheless, some discrepancies appear concerning the prediction of the location and of the thickness of the oblique jumps mainly because jumps are set on one cell in the numerical model. F...

International audience; This paper reports on the numerical modelling of flash flood propagation in urban areas after an excessive rain event or dam/dyke break wave. A two-dimensional (2-D) depth-averaged shallow-water model is used, with a refined grid of quadrilaterals and triangles for representing the urban area topography. The 2-D shallow-water equations are solved using the explicit second-order scheme that is adapted from MUSCL approach. Four applications are reported here to demonstrate the potential benefits and limits of 2-D modelling: (i) laboratory experimental dam-break wave in presence of an isolated building; (ii) flash flood over a physical model of the urbanized Toce river valley in Italy; (iii) extreme flood in October 1988 in the dense city of Nîmes (France); and (iv) dam-break flood in October 1982 in the town of Sumacárcel (Spain). Computed flow depths and velocities compare favourably with recorded data although for the experimental study on dam-break wave some discrepancies are observed around buildings, where the flow is strongly 3-D in character. The numerical simulations show that the flow depths and flood wave celerity are significantly affected by the presence of buildings in comparison with the original floodplain. Further, this study confirms the importance of topography and roughness coefficient for flood propagation simulation.; Cet article relate la modélisation numérique de la propagation de crues éclair en zone urbaine suite à de fortes pluies ou une onde de rupture de barrage. Les équations de Saint Venant bidimensionnelles sont résolues sur un maillage de quadrilatères et triangles représentant la topographie de la zone urbaine. Le schéma numérique est du second ordre du type MUSCL. Quatre applications sont décrites : une expérience en laboratoire d'onde de rupture de barrage autour d'un bâtiment isolé ; crue éclair d'un modèle physique de la vallée du Toce en Italie ; crue éclair d'octobre 1988 à Nîmes (France) ; onde de rupture de barrage dans la ville de Sumacarcel (Espagne) en octobre 1982. Les hauteurs d'eau et vitesses calculées sont en accord avec les données relevées bien que l'expérience de l'onde de rupture de barrage montre des différences autour des bâtiments où l'écoulement est fortement 3D. La modélisation numérique montre que les hauteurs d'eau et vitesses sont fortement modifiées par la présence de bâtiments. En outre, cette étude confirme l'importance de la topographie du terrain naturel et des coefficients de frottement.

Flood hazards (flow depth and velocity) must be accurately assessed in high-risk areas during extreme flood events. However, the prediction of the very high flows is not an easy task due to the lack of field data and to the strong link between flow resistance and the land occupation of the floodplain. Confinement and inhomogeneity in lateral and longitudinal directions of hydraulic roughness strongly vary with return period T. The physical processes are complex, some still largely unexplored, and the assumptions linked to numerical modelling cannot be validated without field data. The FlowRes project (2015-2018), funded by the French National Research Agency (ANR), aims at improving the flood hazard assessment in floodplains in: 1) investigating in laboratory the hydrodynamic structure associated with extreme flood flows for various land occupations and flow discharge magnitudes; 2) assessing if the existing numerical modelling practices used for T ~ 100 years are still valid for extreme events with T > 1000 years, relying on the experimental data and on one field case. This paper reports some results obtained during the first year of the project. ISSN:2267-1242

International audience; Continuous sedimentation of fine sediments in dam reservoirs triggers considerable economic and ecological impacts worldwide. The definition of appropriate reservoir management procedures requires accurate tools for predicting the sediment dynamics both in the reservoir and downstream reaches. In this article, a one-dimensional numerical model aiming to reproduce suspended sediment dynamics is presented. To take into account water intake locations and vertical distribution of suspended sediments in areas of larger water depths close to dams, an additional module is developed. The model is applied to a reach of the French Upper Rhône River, including two reservoirs. The calibrated model provides accurate flux estimates and reproduces erosion and deposition patterns observed in the reservoir for several scenarios. The additional module allows to improve the estimation of the sand load released by low-level outlets. Eventually, the model could be used either to provide insights on the outcome of past flushing operations or as a predictive tool to prepare future operations.

Facing a huge deposition of sediments at the confluence of the Isere and Rhône rivers during flush operations, EDF and CNR supported scientific studies to better understand the sediment dynamics.; L'étude vise à la mise en place d'un protocole de suivi complémentaire des flux sédimentaires apportés par la Basse-Isère au Rhône, notamment durant les opérations de chasse. Il a comme objectif : - d’une part d’améliorer les connaissances scientifiques sur le fonctionnement hydro-sédimentaire de la zone de confluence Isère-Rhône afin de permettre de modéliser de façon plus objective le fonctionnement de ce secteur ; - d’autre part de définir et tester des dispositifs permettant de suivre le transport sédimentaire pour envisager un pilotage en temps réel des aménagements.

International audience; Although various numerical methods were used to simulate real floods occurring in cities, the validation of the models was never accurate because of the lack of data about location and event description and about observation for validation. In order to check the capacities of our 2-dimensional shallow water equations model to simulate an urban flood, we then decided to simulate numerically an experimental event with well known characteristics and accurate flow measurements. The physical model presented in (Ishigaki T. et al., 2003) represents the flooding of the city center of Kyoto in Japan due to an overflow from the Kamo river. The 2-dimensional numerical simulation of this event was then set up and the experimental and computed data were compared. It appears that the event was calculated quite fairly in terms of flow depth and flow rates in the streets and in terms of timing. However, some discrepancies appear between the measurements and the numerical results, mostly due to some topographical local uncertainties and to the capacities of the equations to model the complex flows in the crossroads.; Bien que divers codes de calculs 2D aient déjà été utilisés afin de simuler les inondations urbaines dans des zones urbanisées, aucune validation précise de ces modèles numériques n`est disponible. La méthode de validation proposée dans cet article consiste à simuler une inondation préalablement étudiée expérimentalement par (Ishigaki T. et al., 2003) et qui représente l`inondation du centre ville de Kyoto au Japon. La simulation numérique de cet événement expérimental et la comparaison des données calculées avec les données mesurées permet en effet de vérifier les capacités du code de calcul 2D à simuler un tel événement. Il apparaît que la dynamique de l`inondation ainsi que la plupart des détails de l`écoulement sont bien reproduits numériquement, même si une incertitude subsiste concernant notamment des incertitudes topographiques locales.

Un code résolvant les équations de Saint venant 2D par un schéma du second ordre explicite est utilisé pour simuler la crue extrême d'octobre 1988 dans le quartier Richelieu de la ville française de Nîmes. Un calcul de référence utilisant une description détaillée du réseau de rues et des sections en travers des rues mais négligeant le réseau d'assainissement fournit une altitude moyenne du pic 0,13 m plus basse que les laisses de crue avec un écart type de 0,53 m. l'analyse de sensibilité des paramètres topographiques et numériques montre que globalement les résultats gardent le même niveau de précision, ce qui reflète la stabilité de la méthode de calcul et le lissage des résultats. Toutefois, localement des modifications d'écoulement dues à la modification des valeurs des paramètres changent fortement les niveaux d'eau, particulièrement quand le régime est modifié. En outre, la répartition des écoulements dans la partie val du quartier peut être altéré. Finalement, une seconde crue (2002) a été simulée avec le modèle calé donnant des résultats similaires à 1988. Donc, l'article montre qu'après calage, un modèle 2D peut être utilisé pour aider à la planification de mesures de prévention dans une zone urbaine dense. / A code solving the 2-D shallow water equations by an explicit second-order scheme is used to simulate the severe October 1988 flood in the Richelieu urban locality of the French city of Nîmes. A reference calculation using a detailed description of the street network and of the cross sections of the streets, considering impervious residence blocks and neglecting the flow interaction with the sewer network provides a mean peak water elevation 0.13 m lower than the measured flood marks with a standard deviation between the measured and computed water depths of 0.53 m. Sensitivity analysis of various topographical and numerical parameters shows that globally, the results keep the same level of accuracy, which reflects both the stability of the calculation method and the smoothening of results. However, the local flow modifications due to change of parameter values can drastically modify the local water depths, especially when the local flow regime is modified. Furthermore, the flow distribution to the downstream parts of the city can be altered depending on the set of parameters used. Finally, a second event, the 2002 flood, was simulated with the calibrated model providing results similar to 1988 flood calculation. Thus, the article shows that, after calibration, a 2-D model can be used to help planning mitigation measures in a dense urban area.

Lors des crues extrêmes en ville, une forte part des écoulements reste en surface. Pour simuler ces inondations, deux modèles sont présentés : le logiciel REM2 U unidimensionnel a pour objectif de simuler la propagation des débits de crue dans l'ensemble d'un réseau de rues alors que le logiciel Rubar 20 bidimensionnel vise à fournir plus d'information sur ces écoulements. Des calculs avec ces deux logiciels ont été menés sur la crue d'octobre 1988 dans un quartier de Nîmes. Lors de cet événement, les hauteurs d'eau maximales ont dépassé deux mètres en certains points et les vitesses 2 m/s ce qui entraînait des passages en régime torrentiel. A partir des données rassemblées sur les sections en travers des rues, des maillages de calcul limités au réseau de rues ont été construits pour les deux logiciels afin de permettre un calcul détaillé. La comparaison des résultats avec les laisses de crue montre des situations très contrastées d'un point à un autre pour une hauteur d'eau maximale moyenne sur l'ensemble de la zone inondée correctement simulée. L'écart sur cette hauteur est, en moyenne, de 1 m ce qui provient des incertitudes sur les observations, sur la topographie et sur les conditions aux limites, des approximations lors de la modélisation et de particularités locales non décrites. Entre les deux logiciels, l'évolution des hauteurs et des vitesses est généralement très proche bien que, comme pour la comparaison avec les laisses de crue, des différences locales importantes sont observées., The hydraulic models that are used to simulate floods in rural areas are not adapted to model floods through urban areas, because of details that may deviate flows and create strong discontinuities in the water levels, and because of the possible water flow running in the sewage network. However, such modelling is strongly required because damage is often concentrated in urban areas. Thus, it is necessary to develop models specifically dedicated to such floods. In the southern part of France, rains may have a high intensity but floods generally last a few hours. During extreme events such as the October 1988 flood in the city of Nîmes, most of the flow remained on the ground with high water depths and high velocities, and the role of sewage network can be neglected. A 1-D model and a 2-D model were used to calculate such flows, which may become supercritical. On the catchments of the streams which cross the city of Nîmes, the rainfall was estimated as 80 mm in one hour and 250 mm in six hours in October 1988, although some uncertainties remain. The return period can be estimated between 150 and 250 years. The zone selected to test the models was an area 1.2 km long and less than 1 km wide in the north-eastern part of the city. It includes a southern part with a high density of houses. The slope from the North (upstream) to the South (downstream) was more than 1 % on average and was decreasing from North to South. Various topographical and hydrological data were obtained from the local Authorities. The basic data were composed of 258 cross sections of 69 streets with 11 to 19 points for each cross section. Observations of the limits of the flooded areas and of the peak water levels at more than 80 points can be used to validate the calculation results. The inputs consisted of two discharge hydrographs, estimated from a rainfall-discharge model from rains with a return period of 100 years, which may result in an underestimate of these inputs. These two hydrographs correspond to the two main structures that cross the railway embankment, which constitutes an impervious upstream boundary of the modelled area. Whereas the western and eastern boundaries are well delimitated by hills above maximum water levels, the downstream southern boundary is somewhat more questionable because of possibilities of backwater and inflows from neighbouring areas.The 1-D software REM2U solved the Saint Venant equations on a meshed network. At crossroads, continuities of discharge and of water heads were set. The hydraulic jump was modelled by a numerical diffusion applied wherever high water levels were found. The Lax Wendroff numerical scheme was implemented. It included a prediction step and a correction step, which implied precise solving of these very unsteady and hyperbolic problems. The software was validated on numerous test cases (Al Mikdad, 2000) which proved the adaptation to problems of calculations in a network of streets.The 2-D software Rubar 20 solves 2-D shallow water equations by an explicit second-order Van Leer type finite volume scheme on a computational grid made from triangles and quadrilaterals (Paquier, 1998). The discontinuities (hydraulic jumps for instance) are treated as ordinary points through the solving of Riemann problems. For the Nîmes case, the grid was built from the cross sections of the streets. Four grids were built with respectively 4, 5, 7 or 11 points for every cross section and these points correspond to the main characteristics of the cross section: the walls of the buildings, the sidewalks, the gutters and the middle point. The simplest crossroads were described from the crossings of the lines corresponding to these points, which provide respectively 16, 25, 49 or 121 computational cells. The space step was about 25 metres along the streets but went as low as 0.1 m in the crossroads; due to the explicit scheme, which implies that the Courant number was limited to 1, the time step was very small and a long computational time was required.The computations were performed with a uniform Strickler coefficient of 40 m1/3/s. Both 1-D and 2-D models provided results that agreed well with observed water levels. The limits of the flooded area were also quite well simulated. However, locally, the differences between calculated and observed maximum water depths were high, resulting in an average deviation of about 1 metre. The reasons for such deviations could come from three main causes. First, the uncertainty of topographical data is relatively high, because of the interpolation between measured cross sections without a detailed complementary DEM (digital elevation model). Second, the observed levels were also uncertain and reveal local situations that are not reconstructed by the hydraulic models which provided maximum water levels averaged on one cell which may not coincide with the exact location of the observations. Finally, modelling means a simplification of the processes, which implies cancelling the level variations due to some obstacles, such as cars, which are not simple to identify.In conclusion, both software packages can model a flood, even a flash flood, in an urbanised area. Research is still necessary to develop methods to fully use urban databases in order to define details more precisely. The improvements to the 1-D software should include a better modelling of storage and of crossroads with an integration of adapted relations for the head losses. 2-D software has a greater potential but the difficulty to build an optimal computational grid means a long computational time, which limits the use of such software to small areas. For both software packages, methods still need to be developed in order to represent exchanges with the sewage network, storage inside buildings and inputs directly coming from rainfall.

The method to build a numerical model intended to predict the formation and the change of sediment deposits upstream from a dam is presented. From information about the inputs of water and sediments coming from the catchment supported by a QdF type hydrological analysis, a horizontal 2-D hydraulic model which couples shallow water equations and one equation for advection and diffusion of sediment concentration is used. Applying this model to Zardezas reservoir in Skikda (Algeria) region shows, on the one hand, the practical difficulties met on such case and, on the other hand, the potentialities of such a method for the management of Algerian reservoirs., La construction d`un modèle numérique destiné à prédire la formation et l`évolution de dépôts de sédiments à l`amont d`un barrage est présentée. A partir d'informations sur les apports en eau et en sédiments en provenance du bassin versant consolidées par une analyse hydrologique en QdF, un modèle hydraulique bidimensionnel horizontal couplant équations de Saint Venant et une équation de convection - diffusion est mis en oeuvre. L'application de ce modèle sur la retenue de Zardezas de la région de Skikda (Algérie) montre, à la fois, les difficultés pratiques rencontrées dans la mise en oeuvre et l'apport possible d'une telle méthode pour la gestion des retenues algériennes.

Modelling floods in urban areas generally requires the modelling of both surface and subsurface flows, along with the exchange discharges between these flows. These latter exchanges are studied on an experimental facility at the Ujigawa Open Laboratory of Kyoto University representing a typical urban drainage system with a single street connected through inlets to a drainage pipe located underneath. An exchange model is developed for the cases in which exchange flows are pressurized, applying a head balance between the surface and subsurface layers and using common head loss formulae. Results of the exchange model agree with experimental measurements and show that, for this experimental setup, the exchanges cannot be modelled by fitting a simple and commonly used orifice-type equation. A hydrodynamic model coupling the one-dimensional (1D) and two-dimensional (2D) shallow water equations for, respectively, the pipe and the street flows and including the analytical exchange model is used to simulate the experimental flows. The numerical model can simulate the global characteristics of both street and pipe flows but shows some discrepancies concerning local perturbations of street flow because of the exchange processes.

International audience; This paper describes an approach to assess channel changes of a long anthropogenised river (the Middle Loire River) over decadal timescales. Channel changes are evaluated along geomorphically homogeneous river reaches. The classic geomorphic parameters (active channel width, bed slope, grain size) are complemented with parameters extracted from a 1D hydraulic model: width–depth ratio, effective bed shear stress and specific stream power calculated for the biennial discharge assimilated to bankfull flow conditions. The delineation of reaches is undertaken by combining visual inspection with the implementation of simple statistical tests to corroborate discontinuities in flow and sediment transport. The 450 km long study area has been divided into 167 homogeneous reaches. The general trends observed over the last fifty years are narrowing of the active channel width and incision of the river bed. However, changes in bed level and active channel width are not consistently correlated. Channel changes at the reach scale are mainly controlled by the presence of former sediment extraction sites. Significant incision is observed at the peak of the in-stream sediment mining period. This was followed by channel recovery when extractions stopped. The 1D numerical model provides a more rigorous manner to derive hydraulic parameters. The effective bed shear stress made dimensionless by its critical value and the width–depth ratio helps to explain observed channel responses more effectively by relating patterns to geological units along the Middle Loire.

[Departement_IRSTEA]Eaux [TR1_IRSTEA]ARCEAU [Departement_IRSTEA]Eaux [TR1_IRSTEA]ARCEAU; International audience; This paper presents a comparison of models for extreme rainfall and flood values. Based on a large set of thousands of rainfall and discharge data, the French ExtraFlo project showed that design estimate of extreme values based on the fitting of a distribution on a limited sample of maximum values is very inaccurate when only a few decades of observation are available. Some interesting alternatives are possible, using a regional approach, or historical data when available, or simulation methods, or sedimentological or geochemical signatures of flood deposits.; Cette communication présente les principaux résultats du projet ExtraFlo sur l'estimation des pluies et crues extrêmes. A partir d'un jeu de données conséquent de plusieurs milliers de stations, une procédure d'inter-comparaison a permis d'évaluer les performances d'une trentaine d'implémentations différentes issues de différentes combinaisons (modèle probabiliste, méthode d'estimation, cadre local / régional / mixte local et régional). Le projet ExtraFlo a permis de mettre en évidence le manque de robustesse des approches basées sur le traitement statistique d'une série de quelques dizaines d'années d'observation. Des alternatives intéressantes existent, basées sur l'exploitation des records de pluie et de débit à l'échelle d'une région, de la collecte de données historiques quand cela est possible, de méthodes de simulation basées sur le couplage d'un générateur d'averses avec un modèle hydrologique pour la transformation des pluies en débit, ou des traces morphologiques laissées par les crues anciennes.

Two test cases of embankment breaching are selected in order to compare a simplified model to 2-D calculations. The simplified model is based on an erosion model of the levee completed by the calculation of the head available between upstream and downstream sides of the embankment and an empirical model for wave propagation. The 2-D model is based on shallow water equations ; it can be completed either by the same breach erosion model as in the simplified model or by an equation of sediment transport. The first test case is an experiment in Norway in which the embankment closes the valley. This latter example shows that all the models agree between them and differ from the experimental results because of the uncertainty of the input data and of the measurements. 2-D modelling does not bring any advantage except if sediment transport is simulated on a refined grid in order to obtain the detailed topography at any time. The second case is a breaching of a levee protecting the flood plain during high flows in Agly River. The difficulty stands in providing the hydraulic conditions in the river upstream from the breach and in propagating the flood wave in the complex topography of the flood plain. In such a case, the 2-D model that permits to test various assumptions and link them with physical processes can more easily provide an estimate of the uncertainty of the results., Paquier André. Comparison of Three Methods Assessing the Inundation Caused by the Erosion of an Embankment. In: SimHydro 2014. New Trends in Simulation. 11-13 June 2014 Ecole Polytech’ Nice (France) 2014.