Search

International audience; Parsimonious and simple lumped rainfall-runoff models are useful tools for research and engineering. From years of experience, the Hydrology Group at Irstea (Antony) developed the GR models suite. Aiming at providing an open source tool, the team started to develop an R package (R Core Team, 2018), named airGR (Coron et al., 2017, 2018), that contains these simulation models (including the GR4J model). The airGR package permits to run six hydrological models dedicated to different time steps (hourly to annual), one snow accumulation and melt model (CemaNeige), one automatic calibration procedure, pre-defined graphical plots and a set of efficiency criteria. It is now possible in airGR to use a composite calibration/evaluation criterion that combines several criteria as a single criterion. The choice of the different criteria and the weight applied to each one can be adjusted according to needs. This possibility can improve the calibration performance of models by considering several objectives. Furthermore, several studies have shown that the use of snow cover area (SCA) observations from satellites, such as MODIS SCA, can improve snow model calibration. The airGR package now proposes an improved version of the semi-distributed CemaNeige degree-day model that better represents the hysteretic relationship between SCA and snow water equivalent (SWE). This modification coupled with the combined calibration of the CemaNeige and GR models on SCA data and on river runoff data significantly improve the simulation of the snow cover area by the model and give more robust calibrated parameters (Riboust et al., 2019). Thanks to its success (about 15,000 downloads so far), the airGR development team carries on its efforts to improve the code, offer new features and improve the documentation. Recently, the development team also released a new package, airGRteaching (Delaigue et al., 2018), which depends on airGR and has expressly been developed for educational purposes, in order to avoid the difficulties that students may have with R programming. The airGRteaching graphical user interface will be demonstrated.

Based on a basin scale rainfall runoff model, we proposed a prediction method of debris-flow occurrence on steep mountain slopes related to hydrological processes such as the rainfall infiltration, the surface flow and the slope stability. For example, in one case that the soil layer is unsaturated and a landslide does not occur in the slope even though the groundwater level rises in the slope soil layer during a rainfall event, it is unlikely for a debris flow to occur on the slope. However, if the soil layer is more unstable due to fully saturation and a surface flow also takes place on the slope, the possibility of debris-flow occurrence gets much higher. According to such a consideration, the slope conditions on hydrological processes during heavy rainfalls were classified into six patterns. For these patterns, the possibility of debris-flow occurrence was investigated qualitatively. Then, SiMHiS (Storm Induced Multi-Hazards Information Simulator) by Yamanoi and Fujita was employed as a rainfall runoff model. A slope stability model has been already installed in SiMHiS. Therefore, this model can simulate the time variations of the safety factors for landslides as well as the saturation degrees and the hydrographs of the surface flow for the slopes. SiMHiS was applied to the sediment disasters due to a heavy rainfall in July 2017 in the Akatani river basin to examine the occurrence patterns of debris flow. Also, the differences in the occurrence patterns were shown for other two rainfall events. Using the simulation result on the safety factor, the saturation degree and the surface flow discharge, it was noted whether debris flows took place or not, and the debris-flow occurrence patterns on the slopes in the basin could be identified.

Mahasarakham International Journal of Engineering Technology, 5, 2, 48-54

Typhoons are the main type of natural disaster in Korea, and accurately predicting typhoon-induced flood flows at gauged and ungauged locations remains an important challenge. Flood flows caused by six typhoons since 2002 (typhoons Rusa, Maemi, Nari, Dienmu, Kompasu and Bolaven) are modeled at the outlets of 24 Geum River catchments using the Probability Distributed Moisture model. The Monte Carlo Analysis Toolbox is applied with the Nash Sutcliffe Efficiency as the criterion for model parameter estimation. Linear regression relationships between the parameters of the Probability Distributed Moisture model and catchment characteristics are developed for the purpose of generalizing the parameter estimates to ungauged locations. These generalized parameter estimates are tested in terms of ability to predict the flood hydrographs over the 24 catchments using a leave-one-out validation approach. We then test the hypothesis that a more complex generalization approach, the Generalized Estimating Equation, which includes properties of the typhoons as well as catchment characteristics as predictors of PDM model parameters, will provide more accurate predictions. The results show that the predictions of Generalized Estimating Equation are comparable to those of the simpler, conventional regression. The simpler approach is therefore recommended for practical applications; however, further refinements of the Generalized Estimating Equation approach may be explored.

International audience; In raingage-poor regions, global rainfall reanalysis products can constitute a valuable source of precipitation information for hydrological applications. Indeed, the spatial and time scales of reanalysis data sets are compatible with those of hydrological models. Unfortunately, global rainfall reanalyses often present large biases which limit their use without a preliminary adjustment. The ECMWF ERA-Interim reanalysis provides long rainfall time series from 1979. Over Africa, there is enough overlap with ground measurements to allow for calibrating an error-adjustment strategy. This poster presents a climatic adjustment of the ERA-interim rainfall reanalysis based on Tractebel's historical raingage dataset. The method consists in building a seasonal intensity-dependent error correction curve using all data points where ground measurements are available. One correction curve is estimated for each month and each region. The important hypothesis of the method is that the reanalysis error relative to rainfall intensity is stable in the time. We applied here the method over Africa, and validate it first at the raingage scale and then at the catchment scale.

Degree-day snow models have the advantage of requiring limited data for running and for calibration, which is of the utmost importance for real-time hydrological forecasting or assessment of the impact of climate change on the hydrological regimes of snow-driven catchments. The CemaNeige model is a daily 2-parameter degree-day model that proved to be very efficient for discharge simulation when run together with a daily rainfall-runoff model (usually the GR4J model). In this work, we tried to represent in a more realistic way the snowpack, based on the integration of a SWE-SCA hysteresis. These SWE-SCA relationships aim at describing the heterogeneity of snow patterns both in space and time in the catchments. With this improved model, we showed that it is possible to make use of spatial satellite MODIS SCA data to improve the snow representation without deteriorating the discharge. The sensitivity of the relative weights between snow-based and discharge-based numerical criteria was assessed. Robustness of the model (i.e. its ability to be applied on independent periods and catchments) was improved.

International audience; In many conceptual rainfall-runoff models, the water balance differential equations are not explicitly formulated. These differential equations are solved sequentially by splitting the equations into terms that can be solved analytically with a technique called "operator splitting". As a result, only the solutions of the split equations are used to present the different models. This article provides a methodology to make the governing water balance equations of a bucket-type rainfall-runoff model explicit and to solve them continuously. This is done by setting up a comprehensive state-space representation of the model. By representing it in this way, the operator splitting, which makes the structural analysis of the model more complex, could be removed. In this state-space representation, the lag functions (unit hydrographs), which are frequent in rainfall-runoff models and make the resolution of the representation difficult, are first replaced by a so-called "Nash cascade" and then solved with a robust numerical integration technique. To illustrate this methodology, the GR4J model is taken as an example. The substitution of the unit hydrographs with a Nash cascade, even if it modifies the model behaviour when solved using operator splitting, does not modify it when the state-space representation is solved using an implicit integration technique. Indeed, the flow time series simulated by the new representation of the model are very similar to those simulated by the classic model. The use of a robust numerical technique that approximates a continuous-time model also improves the lag parameter consistency across time steps and provides a more time-consistent model with time-independent parameters.

International audience; Log-transformed discharge is often used to calculate performance criteria to better focus on low flows. This prior transformation limits the heteroscedasticity of model residuals and was largely applied in criteria based on squared residuals, like the Nash-Sutcliffe efficiency (NSE). In the recent years, NSE has been shown to have mathematical limitations and the Kling-Gupta efficiency (KGE) was proposed as an alternative to provide more balance between the expected qualities of a model (namely representing the water balance, flow variability and correlation). As in the case of NSE, several authors used the KGE criterion (or its improved version KGE′) with a prior logarithmic transformation on flows. However, we show that the use of this transformation is not adapted to the case of the KGE (or KGE′) criterion and may lead to several numerical issues, potentially resulting in a biased evaluation of model performance. We present the theoretical underpinning aspects of these issues and concrete modelling examples, showing that KGE′ computed on log-transformed flows should be avoided. Alternatives are discussed.

[Departement_IRSTEA]Eaux [TR1_IRSTEA]ARCEAU [ADD1_IRSTEA]Hydrosystèmes et risques naturels; International audience; Small catchments are largely under-represented in gauged catchments samples used to calibrate nationwide flood quantiles estimation methods. For instance, in France, catchments of less than 10 km² represent 3% of the gauged catchments but 40% of the sites of interest. Those small basins can generally be found upstream of a gauging station. This size gap should be fulfilled by regionalisation techniques. Those methods aim to transfer information from gauged basins to ungauged ones, but they rarely address the size gap issue. The work presented her aims to question about what happens inside the gauged catchments. Are the model calibrated at gauged catchments scale still valid at a finer one or is it useful to implement some downsizing method? To address this question we propose several downsizing techniques of the calibrated parameter of the SHYREG method. SHYREG is a flood frequency analysis method providing flood quantiles estimations at a fine spatial resolution throughout the French territory. It is based on the association of a stochastic rainfall generator and an event-based rainfall-runoff conceptual model. The rainfall events are simulated at each cell of a 1 km resolution mesh. Nonetheless, the rainfall-runoff model calibration and regionalisation are traditionally performed in a global way at the catchment scale. SHYREG appears suitable for the presented study because it takes into account the catchment size effect and distributed rainfall information, therefore the calibrated parameter is assumed to represent the sole rainfall-runoff transformation. The proposed downsizing techniques include a semi-distributed calibration and a cell-based regional calibration of the SHYREG rainfall-runoff model parameter. The impact of this different spatial disaggregation methods are evaluated thanks to a split-sample procedure based on the catchments size, the smallest basins being used for validation. A particular focus is given to some well instrumented nested catchments groups in France. Results suggest that the downsizing process implies a better discharge consistency between upstream and downstream sites, especially when simulating flows at a fine resolution along a river channel. Then, more realism is introduced in the spatial distribution of the rainfall-runoff model parameter. Finally, the decrease of the number of degree of freedom implied by the downsizing process does not affect the overall quality of the flood quantiles estimations. This last result illustrates the existence of a spatial equifinality in the distribution of model parameters, hence the necessity to control this distribution.

International audience; The aim of this presentation is to detail a state-space representation of the simple GR4J model. The state-space representation aims to represent GR4J by an ODE system which provides the internal variables of the model at all times. We will present here the choices made to adapt GR4J to the state-space formulation and to numerically solve this system. Modifications of the model’s equations were also made to adapt the model to lower time step in case it would be used for a time step-variable application.

[Departement_IRSTEA]Eaux [TR1_IRSTEA]ARCEAU; International audience; Much has been written on the subject of objective functions to calibrate rainfall–runoff models. Many studies focus on the best choice for low-flow simulations or different multi-objective purposes. Only a few studies, however, investigate objective functions to optimize the simulations of low-flow indices that are important for water management. Here, we test different objective functions, from single objective functions with different discharge transformations or using low-flow indices, to combinations of single objective functions, and we evaluate their robustness and sensitivity to the rainfall–runoff model. We find that the Kling and Gupta efficiency (KGE) applied to a transformation of discharge is inadequate to fulfil all assessment criteria, whereas the mean of the KGE applied to the discharge and the KGE applied to the inverse of the discharge is sufficient. The robustness depends on the climate variability rather than the objective function and the results are insensitive to the model.; De nombreuses études se sont intéressées aux fonctions objectif pour caler des modèles pluie-débit. Parmi elles, un grand nombre se focalise sur le meilleur choix pour la simulation de l’étiage ou différents buts multi-objectif. Cependant, peu d’études s’intéressent au choix d’une fonction objectif pour optimiser la simulation d’indices d’étiage nécessaires pour la bonne gestion de la ressource en eau en période d’étiage. Dans cet article, nous testons différentes fonctions objectif en partant de fonction objectif simple utilisée avec différentes transformations du débit ou utilisant les indices d’étiage jusqu’aux combinaisons de différentes fonctions objectif simples. Nous évaluons ensuite leur robustesse et leur sensibilité au modèle pluie-débit utilisé. Nous trouvons que le KGE appliqué à une transformation du débit n’est pas approprié pour remplir toutes les conditions d’évaluation alors qu’une combinaison comme la moyenne entre le KGE appliqué au débit et le KGE appliqué à l’inverse du débit est suffisante. La robustesse des simulations n’était pas dépendante du choix de la fonction objectif mais de la variabilité climatique au sein des bassins versants et les résultats n’étaient pas sensibles au choix du modèle pluie-débit.

International audience

[Departement_IRSTEA]Eaux [TR1_IRSTEA]ARCEAU [ADD1_IRSTEA]Hydrosystèmes et risques naturels; International audience; Flood frequency analyses (FFAs) are needed for flood risk management. Many methods exist ranging from classical purely statistical approaches to more complex approaches based on process simulation. The results of these methods are associated with uncertainties that are sometimes difficult to estimate due to the complexity of the approaches or the number of parameters, especially for process simulation. This is the case of the simulation-based FFA approach called SHYREG presented in this paper, in which a rainfall generator is coupled with a simple rainfall-runoff model in an attempt to estimate the uncertainties due to the estimation of the seven parameters needed to estimate flood frequencies. The six parameters of the rainfall generator are mean values, so their theoretical distribution is known and can be used to estimate the generator uncertainties. In contrast, the theoretical distribution of the single hydrological model parameter is unknown; consequently, a bootstrap method is applied to estimate the calibration uncertainties. The propagation of uncertainty from the rainfall generator to the hydrological model is also taken into account. This method is applied to 1112 basins throughout France. Uncertainties coming from the SHYREG method and from purely statistical approaches are compared, and the results are discussed according to the length of the recorded observations, basin size and basin location. Uncertainties of the SHYREG method decrease as the basin size increases or as the length of the recorded flow increases. Moreover, the results show that the confidence intervals of the SHYREG method are relatively small despite the complexity of the method and the number of parameters (seven). This is due to the stability of the parameters and takes into account the dependence of uncertainties due to the rainfall model and the hydrological calibration. Indeed, the uncertainties on the flow quantiles are on the same order of magnitude as those associated with the use of a statistical law with two parameters (here generalised extreme value Type I distribution) and clearly lower than those associated with the use of a three-parameter law (here generalised extreme value Type II distribution). For extreme flood quantiles, the uncertainties are mostly due to the rainfall generator because of the progressive saturation of the hydrological model.

International audience; GR is a family of lumped hydrological models designed for flow simulation at various time steps. The models are now available in a flexible R package called airGR(Coronet al., 2017, submitted). The models can easily be implemented on a set of catchments with limited data requirements.

International audience; SHYREG method is a regionalized method for rainfall and flood frequency analysis (FFA). It is based on processes simulation. It couples an hourly rainfall generator with a rainfall-runoff model, simplified enough to be regionalized. The method has been calibrated using all hydro meteorological data available at the national level. In France, that represents about 2800 raingauges of the French Weather Service network and about 1800 stations of the hydrometric National Bank network. Then, the method has been regionalized to provide a rainfall and flow quantiles database. An evaluation of the method was carried out during different thesis works and more recently during the ANR project Extraflo, with the aim of comparing different FFA approaches. The accuracy of the method in estimating rainfall and flow quantiles has been proved, as well as its stability due to a parameterization based on average values. The link with rainfall seems preferable to extrapolation based solely on the flow. Thus, another interest of the method is to take into account extreme flood behaviour with help of rainfall frequency estimation. In addition, the approach is implicitly multi-durational, and only one regionalization meets all the needs in terms hydrological hazards characterisation. For engineering needs and to avoid repeating the method implementation, this method has been applied throughout a 50 meters resolution mesh to provide a complete flood quantiles database over the French territory providing regional information on hydrological hazards. However, it is subject to restrictions related to the nature of the method: the SHYREG flows are "natural", and do not take into account specific cases like the basins highly influenced by presence of hydraulic works, flood expansion areas, high snowmelt or karsts. Information about these restrictions and uncertainty estimation is provided with this database, which can be consulted via web access.

[Departement_IRSTEA]Eaux [TR1_IRSTEA]ARCEAU; International audience; This paper examines catchments that are almost ungauged, i.e., catchments for which only a small number of point flow measurements are available. In these catchments, hydrologists may still need to simulate continuous streamflow time series using a rainfall-runoff model, and the methodology presented here allows using few point measurements for model parameterization. The method combines regional information (parameter sets of neighboring gauged stations) and local information (contributed by the point measurements) within a framework where the relative weight of each source of information is made dependent on the number of point measurements available. This approach is tested with two different hydrological models on a set of 609 catchments in France. The results show that on average a few flow measurements can significantly improve the simulation efficiency, and that 10 measurements can reduce the performance gap between the gauged and ungauged situations by more than 50%. The added value of regional information progressively decreases until being almost insignificant when sufficient flow measurements are available. Model parameters tend to come closer to the values obtained by calibration in fully gauged conditions as the number of point flow measurements increases.

International audience; In absence of streamflow data to calibrate a hydrological model, its parameters are to be inferred by a regionalization method. In this technical note, we discuss a specific class of regionalization methods, those based on spatial proximity, which transfers hydrological information (typically calibrated parameter sets) from neighbor gauged stations to the target ungauged station. The efficiency of any spatialproximity-based regionalization method will depend on the density of the available streamgauging network, and the purpose of this note is to discuss how to assess the robustness of the regionalization method (i.e., its resilience to an increasingly sparse hydrometric network). We compare two options: (i) the random hydrometrical reduction (HRand) method, which consists in sub-sampling the existing gauging network around the target ungauged station, and (ii) the hydrometrical desert method (HDes), which consists in ignoring the closest gauged stations. Our tests suggest that the HDes method should be preferred, because it provides a more realistic view on regionalization performance.

[Departement_IRSTEA]Eaux [TR1_IRSTEA]ARCEAU; International audience; Estimating low flow indices is of paramount importance to manage water resources and risk assessments. These indices are derived from river discharges which are measured at gauged stations. However, the lack of observations at ungauged sites bring the necessity of developing methods to estimate these low flow indices from observed discharges in neighbouring catchments and from catchment characteristics. Different estimation methods exist. Regression or geostatistical methods performed on the low flow indices are the most common types of methods. Another less common method consists in regionalizing rainfall-runoff model parameters, from catchment characteristics or by spatial proximity, to estimate low flow indices from simulated hydrographs. Irstea developed GR2M-LoiEau, a conceptual monthly rainfall-runoff model, combined with a regionalized model of snow storage and melt. GR2M-LoiEau relies on only two parameters, which are regionalized and mapped throughout France. This model allows to cartography monthly reference low flow indices. The inputs data come from SAFRAN, the distributed mesoscale atmospheric analysis system, which provides daily solid and liquid precipitation and temperature data from everywhere in the French territory. To exploit fully these data and to estimate daily low flow indices, a new version of GR-LoiEau has been developed at a daily time step. The aim of this work is to develop and regionalize a GR-LoiEau model that can provide any daily, monthly or annual estimations of low flow indices, yet keeping only a few parameters, which is a major advantage to regionalize them. This work includes two parts. On the one hand, a daily conceptual rainfall-runoff model is developed with only three parameters in order to simulate daily and monthly low flow indices, mean annual runoff and seasonality. On the other hand, different regionalization methods, based on spatial proximity and similarity, are tested to estimate the model parameters and to simulate low flow indices in ungauged sites. The analysis is carried out on 691 French catchments that are representative of various hydro-meteorological behaviours. The results are validated with a cross-validation procedure and are compared with the ones obtained with GR4J, a conceptual rainfall-runoff model, which already provides daily estimations, but involves four parameters that cannot easily be regionalized.

International audience; Due to its mountainous topography and its Mediterranean climate, the Provence-Alpes-Côte d'Azur (PACA) region in Southeastern France is particularly prone to flash floods, debris flows and mass movements (landslides and rockfall). A mapping system for these rainfall induced hazards has been tested by local and regional authorities and Government agencies since 2011 as part of the RHYTMME project. This system allows, thank to radar rainfall estimation and rainfall-runoff modelling, the real-time warning and monitoring of flash floods wherever they may occur in the PACA territory. It is also intended to enable, during intense rainfall events, the localisation of the streams susceptible to generate debris flows and of the slopes the more likely to trigger landslides and/or rockfalls.

International audience; The impact of hydrological hazard on the reliability of hydraulic structures is difficult to take into account from the complexity of the phenomenon.A common dam design practice toward flood risk is to simulate the reservoir level reached for a design hydrograph. However, the most critical hydrological scenarios are highly variable from one catchment to another, and depend on the characteristics of the dam, the reservoir, the spillways, etc. These scenarios are characterised by multiple variables (peak, volume, initial level, etc.), thus making complex the estimation of their return period.The extreme floods estimation methods based on stochastic rainfall-runoff simulation generate a large number of floods with various dynamics. By coupling these methods with the hydraulic simulation of a dam, the distribution of maximum reservoir levels can be computed, from which design values can be extracted.Two simulation methods, SHYPRE and SCHADEX, have been applied to a virtual dam located on the Tech catchment (477 km², France) and coupled with the hydraulic simulation of the dam, allowing to compute the distribution of the reservoir levels up to the 10 000 years return period. Several variables were analysed to study the involved processes: peak flow, 24 and 72h flow, flood hydrographs, initial reservoir levels.These two approaches produce neighbouring estimates of extreme flood and reservoir quantiles, despite significant methodological differences. Design reservoir levels were finally compared to the traditional approach consisting of simulating the abatement of a design hydrograph.The advantage of these simulation methods is that they allow direct study of the structure response frequency distribution. The structure response depends on the dam functioning, its management and its probability of failure, which can be easily modelled. We can then study in an integrated manner, all operating scenarios of structure coupled to possible flood scenarios.; Les méthodes d'estimation de crues extrêmes basées sur la simulation stochastique génèrent un grand nombre de crues de dynamique variable. En couplant ces méthodes à une simulation hydraulique de l'aménagement étudié, on produit une distribution des cotes maximales atteintes dans la retenue, à partir de laquelle on extrait une cote de projet. Cet article présente deux méthodes de simulation développées en France, SHYPRE et SCHADEX, appliquées à un barrage virtuel situé sur le Tech (477 km², Pyrénées-Orientales) et couplées à une simulation hydraulique de l'aménagement, afin d'établir les distributions de cotes maximales atteintes jusqu'à une période de retour 10 000 ans. Ces deux approches sont comparées à une approche plus classique basée sur la détermination d'une crue de projet unique. Les résultats montrent que les deux méthodes par simulation conduisent à des estimations de débit et de cotes assez convergentes, en dépit d'importantes différences méthodologiques. La comparaison à l'approche classique montre que les résultats de cette dernière sont fortement tributaires des hypothèses que l'on peut faire sur l'estimation de la crue de projet. Les méthodes par simulation permettent d'avoir une approche plus exhaustive des configurations auxquelles sont soumis les barrages. Elles permettent ainsi de tester de multiples configurations probables et de fournir une aide à la décision quant au dimensionnement ou à la gestion de l'ouvrage.

International audience; This study deals with the Rimbaud catchment, a sub-catchment within the Réal Collobrier hydrological observatory in south-eastern France, managed by Irstea since 1966. This observatory suffered a wildfire in 1990.Because of the dense network of streamgauges and raingauges available, this site provides a unique opportunity to test and compare two types of analysis, one based on paired catchments and the other a rainfall-runoff model,used to assess the hydrological impact of forest fire. In the present case, more than 20 years of pre-fire and post-fire data are available. We compare the ability of the two approaches to detect gradual changes at a daily time step. This case study illustrates how natural climatic variability (here a long drought which preceded the wildfire) can make the identification of hydrological changes extremely difficult.; Cet article traite du bassin versant du Rimbaud situé dans le Sud de la France. Il fait partie de l'observatoire hydrologique du Réal Collobrier suivi depuis 1966 par Irstea. Cet observatoire a subi un incendie de forêt en 1990. Grâce à son réseau dense de stations hydrométriques et pluviométriques, ce site fournit une occasion unique de comparer la méthode des bassins appariés et la modélisation pluie-débit pour évaluer l'impact hydrologique de l'incendie de forêt. Dans notre cas, plus de 20 années de données avant et après incendie sont disponibles. Nous pouvons donc comparer la capacité des deux méthodes à détecter des changements graduels au pas de temps journalier. L'étude illustre comment la variabilité climatique naturelle (ici une longue sécheresseprécédent l'incendie) rend extrêmement difficile toute interprétation de modification du comportement hydrologique.

[Departement_IRSTEA]Eaux [TR1_IRSTEA]ARCEAU; International audience; Issues with water resource management result from both an increasing demand and climate changes. The situations of low-flows, droughts and more generally lack of water are critically scrutinized. In this context, there is a need for tools to assist water agencies in the prediction and management of reference low-flows at gauged and ungauged catchment locations. IRSTEA developed GR2M-LoiEau, a conceptual distributed rainfall-runoff model, which is combined with a regionalized model of snow storage and melt. GR2M-LoiEau relies on two parameters which are regionalized and mapped throughout France. This model allows to cartography annual and monthly reference low-flows. The input meteorological data come from the distributed mesoscale atmospheric analysis system SAFRAN, which provides daily solid and liquid precipitations and temperatures data from everywhere in the French territory. In order to fully exploit these daily meteorological data to estimate daily statistics on low flows, a new version of GR2M-LoiEau is being developed at a daily time step, yet keeping only a few regionalized parameters. The aim of this study is to design a comprehensive set of tests to allow comparing low-flows obtained with different regionalization methods used to estimate low-flow model parameters. The new version of GR2M-LoiEau being not yet operational, the tests are made with GR4J (Perrin, 2002), a conceptual rainfall-runoff model, which already provides daily estimations, but involves four parameters that cannot easily be regionalized. Many studies showed the good prediction performances of this model. This work includes two parts. On the one hand, good criteria must be identified to evaluate and compare model results, good predictions of the model being expected about low flows and reference low flows, but also annual means and high flows. On the other hand, two methods of regionalization will have to be compared to estimate model parameters. The first one is rough, all the parameters regionalized in one time. The second one is a gradual regionalization method.

In this paper the development, construction and use of a rainfall simulator (RS-TUHH) are described. On a testing area of 1m² different drop forming measures have been analysed before the simulator was constructed. The RS-TUHH uses ending grid heads with a grid space of 6 cm and an underlying drop splitting net. The simulator consists of four parts: the pipe pressure and water distribution module, the aluminum structure, the irrigation system and the drop splitting net. As it is built, the shaped plot area of 6 m² can be divided in to 3 sectors á 2 m². The main intention is to imitate natural rainfall by reproducing parameters like rainfall velocity, intensity and duration, as well as drop size, amount and distribution. Experimental parameters are measured with laser precipitation monitor (by ThiesClima) and rain collectors. The RS-TUHH is able to imitate natural heavy rain events of different intensities from 3 to 200 mm/h on three sizes of plot areas with any duration. Intensity and duration are set by the control module. Drop size distribution and drop amount depend on the used net grid space. Spatial distribution is homogeneous over the plot area. Fall velocities are low compared to natural rain. Bundesministerium für Bildung und Forschung

The great interest in climate change during the past 20 years has led to a quasi unanimous conclusion for scientists: the Earth's climate is changing (IPCC 2013). It is important to know if this global change could lead to an increase in extreme events in order to prevent hydrological risks. In this work, the analysis of the climate change impact on flood was studied by a chain formed by projections (provided by climate models under SRES scenarios) and a stochastic hydrological model. The National Research Institute of Science and Technology for Environment and Agriculture (Irstea) has developed an original method for flood frequency analysis applied on the whole French territory: the SHYREG method (Arnaud et al., 2008). It generates sequentially a lot of rainfall events at an hourly time step for which a rainfall-runoff transformation is performed. The stochastic rainfall generator has three parameters which are estimated by average, not by extreme, values of daily climatic characteristics. A simple model, parametrized with few parameters allows generating flood events. These parameters have been regionalized on the whole French territory in order to estimate rainfall/flood quantiles at the spatial resolution of 1 km². The rainfall model shows a good skill in reproducing extreme rainfall frequency (Carreau et al., 2013) and has been already used in a climate change context to detect trends in extreme rainfall (Cantet et al., 2011). (Boé at al., 2006) propose climate projections on France at a 8km horizontal spatial resolution with daily rainfall available for two periods: reference period (1981-2000) and the end of the 21th century (2081-2100) under three SRES scenarios (B1, A1B, A2). The parameters of the rainfall model can be easily estimated for the different periods and scenarios and so, the sensitivity of flood frequency to the climate change can be studied under some hypothesis. First, the performance of the climatic model to reproduce extreme rainfall has been tested throughout the SHYREG method. Then projections were used to estimate flood frequency in a possible future. In the first case, the rainfall-runoff relation is assumed to be stationary and only rainfall parameters change between two periods. This configuration allows focusing only on the impact of a possible rainfall change on the flood frequency. Then some scenarios of hydrological model' parameters are proposed to underline the consequences of changes in the rainfall-runoff transformation on the flood frequency estimation in the future.

Development of operational flash flood warning systems is one of the challenges in operational hydrology: flash floods are devastating but difficult to monitor and predict due to their nature. To provide flash flood warnings for ungauged basins, Météo-France and Irstea (formally Cemagref) have developed a discharge-threshold flood warning system called AIGA, which combines radar-gauge rainfall grids with a simplified distributed rainfall-runoff model run every 15 minutes at a 1-km2 resolution. Operational since 2005 in the Southern part of France, the AIGA system produces, every 15 minutes, a map of the river network with a color chart indicating the range of the estimated return period of the ongoing flood event. To increase forecast lead time and quantify the forcing input uncertainty, the rainfall-runoff distributed model ingests the 11 precipitation ensemble members from the PEARP ensemble prediction system of Météo-France. Performance of the experimental probabilistic precipitation and flow forecasts is evaluated from a variety of ensemble verification metrics (e.g., Continuous Ranked Probability Skill Score, Relative Operating Characteristic score) for different French basins. We also discuss planned enhancements and challenges to assess other sources of hydrological uncertainty and effectively communicate the uncertainty information to forecasters for better risk-based decision making.

A major difficulty for studies of small hydrological watersheds is the lack of good quality time series of hydrologic data, mainly because the flow rates in small watersheds are not monitored. Another important issue is that available rainfall runoff models are almost always developed focusing on watersheds of medium and large scale. Regionalization studies have become an important tool to attempt to overcome these limitations. Suitable in most of the hydrological studies is the regionalization of rainfall runoff-model parameters by using specific characteristics of a watershed. A tool was developed in a Geographic Information System which automatically gets the physical characteristics of watersheds from a digital elevation model by selecting the outlet and then generating the rainfall runoff model parameters with neural networks. This study uses data from small dams in the semi-arid region of northeastern Brazil. The developed methodology is applied using target watersheds for the parameter estimation. The results show that the developed tool can be very useful for rainfall runoff estimation in small watersheds. Coordenação de Aperfeiçoamento de Pessoal de Nível Superior - CAPES Uma das maiores dificuldades para os estudos de pequenas bacias hidrográficas é a falta de séries históricas de dados hidrológicos, principalmente porque a fluviometria em pequenas bacias hidrográficas não é monitorada. Outra questão importante é que, os modelos chuva-vazão disponíveis quase sempre são desenvolvidos com foco em bacias de médio e grande porte. Estudos de regionalização se tornaram uma ferramenta importante para tentar superar essas limitações. A regionalização de parâmetros de modelos chuva-vazão, usando características específicas das bacias hidrográficas, é adequada para a maioria dos estudos hidrológicos. Uma ferramenta foi desenvolvida em um Sistema de Informação Geográfica, que determina automaticamente as características físicas das bacias hidrográficas a partir de um modelo digital de elevação, e, em seguida, gera os parâmetros do modelo chuva-vazão com redes neurais artificiais. Este trabalho usa dados de pequenas barragens na região semiárida do Nordeste do Brasil. A metodologia desenvolvida é aplicada e avaliada utilizando também o método da bacia meta para a estimativa dos parâmetros. Os resultados mostram que a ferramenta desenvolvida pode ser muito útil para o estudo da relação chuva-vazão em bacias hidrográficas de pequeno porte.

We address in this study the routing of distributed hydrological information to the outlet of watersheds, in the fields of flood forecasting on ungauged watersheds in the French Mediterranean area. Flood forecasting can benefit of rainfall data provided in real-time by radar networks. This data used as an input to rainfall runoff models gives access to flood anticipation on small ungauged watersheds prone to flash floods. Within the framework of the AIGA method, developed by IRSTEA (formerly Cemagref) and Météo-France (the French National Weather Service) to provide floods alerts, a rainfall-runoff model is implemented at the spatial resolution of the radar data, thus providing a map of the 1 km² pixel contributions to the runoff. We have focused on the routing function of the rainfall-runoff model at the 1 km² cell scale, this scale being the first step of the runoff routing from the production area to the outlet of the catchment. We have then produced streamflow hindcasts for selected observed events using: - two different kinds of routing functions within our rainfall-runoff model, the first being a uniform routing function, the second being distributed routing functions, - two different hindcasts configurations, the first being a research configuration with the calibration of the production function of our rainfall-runoff model, the second being a configuration close to the real-time conditions of an operational flood warning service, i.e. with no possibility to calibrate the model parameters. Results indicate that the AIGA method forecasting ability, assessed via a multicriteria analysis evaluating the forecasting lead-time, contingency statistics and the similarity between the calculated and observed hydrographs, is enhanced (both in research and operational configurations) when implementing a distributed routing function.

/ La connaissance des débits de crue reste un axe de recherche important en hydrologie pour la conception des aménagements des cours d'eau, le dimensionnement des ouvrages de franchissement et la protection des zones urbaines. La méthode SHYREG est une méthode de prédétermination des crues basées sur la simulation des processus. Elle résulte du couplage d'un générateur de pluies horaires et d'un modèle hydrologique, tout deux régionalisés. Le développement opérationnel de la méthode a fait appel à des hypothèses successives de simplification qui ont été revues dans ce travail dont l'objectif était d'apporter des améliorations sur la modélisation pluie-débit utilisée afin de répondre aux problèmes suivants : - quelle est la validité du domaine géographique d'application de la méthode ? - quelle est la validité du domaine d'extrapolation fréquentielle de la méthode ? Les différentes hypothèses dans la méthode ont toutes une interaction les unes avec les autres, ce qui rend leur étude particulièrement difficile. Une hiérarchisation des hypothèses a été faite, en fonction de leur caractère plutôt général (indépendant des bassins versants et soumis à un calage global), ou plutôt local (dépendant du bassin et soumis à un calage local). Dans ce contexte, on traite des points suivants: - L'amélioration de la fonction d'agglomération des débits qui permet de passer du pixel à l'exutoire d'un bassin. La reformulation de cette règle d'abattement a permis d'élargir la gamme de superficie des bassins où la méthode reste applicable. - La prise en compte d'un débit de base dans la modélisation hydrologique des crues a permis d'améliorer les performances en dehors de sa zone de développement (hors bassins méditerranéens). - L'étude de l'équifinalité de certains paramètres a permis de proposer une extrapolation vers les fréquences rares, en fonction de différents critères. Cette approche « multicritère » est un moyen de faire un choix sur les hypothèses les plus probables d'une méthode dans le cas particulier de l'extrapolation en fréquence, où il existe peu d'information pour valider. Sur l'ensemble de ces points, on s'est appuyé sur un large échantillon de bassins versants regroupant l'ensemble des réponses hydrologiques possibles (bassin à influence méditerranéen, pluvio-nival, remontée de nappe). D'autre part, on dispose de quantiles de débit (pour les fréquences courantes et les fréquences extrêmes) issus de l'application d'autres méthodes que l'on a confronté avec les quantiles Shyreg.

International audience; The real-time vision of a large portion of French rivers is not a coloured one, not even a black and white, but a blind one, du to the large number of basins which are not monitored. The event of June 2010 in Southern France, where 25 people died during the flood of the Nartuby and Argens rivers gives a dramatic example of this blind vision. In such a context a simple insight, aiming at providing flood alerts rather than pretending to represent exactly the physical process at stake in the catchment, seems welcomed. The aim of this study is to assess the ability of a simple conceptual and parsimonious rainfall-runoff model to deliver flood alerts in ungauged catchments. We use a downward approach: the model is calibrated on gauged catchments and its performances are checked on nested catchments which are not used for the calibration. The model is calibrated over 20 catchments of southern France with areas ranging from 150 km² to 2170 km². The rainfall information is provided by a radar network at an hourly time-step for 15 events that have occurred between 2005 and 2008. The "ungauged" catchments used for the model evaluation are 47 nested basins located within the 20 calibration catchments. We derive the model parameters for these ungauged catchments, from the gauged ones. The model used belongs to the GR models family developed at the Cemagref institute. It works on en event basis thank to a production parameter S and two routing parameters: a shape parameter B and a lag parameter C. Our knowledge of the production parameter benefits from a previous regionalisation work based on several characteristics of the catchments such as the description of their drainage network, the ground-water typology and a climatic variable. The model calibration rests upon the production parameter only. We implement two versions of the model: a lumped one and a distributed one. The results of this downward approach show that the model tested here is able to deliver successful flood alerts in ungauged areas. The distributed version performs slightly better than the lumped version. This study also show that fully distributing the routing parameters results in better alerts than using uniform routing parameters in the distributed version. To fully assess the performance of the model, further evaluation should be carried out in an operational mode, i.e. using an initialisation procedure to determine the value of the S parameter at the beginning of each event instead of using a calibrated value. We anticipate this study by checking how well the calibrated S correlate with the parameter of a continuous soil moisture accounting model run at a daily time-step. The correlation being better in the case of the distributed version of the model, this version seems hence quite promising for an operational use.