Search

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.

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.

National audience; Présentation de la méthodologie de comparaison utilisée dans ExtraFlo

International audience

[Departement_IRSTEA]Eaux [TR1_IRSTEA]ARCEAU [Departement_IRSTEA]Eaux [TR1_IRSTEA]ARCEAU; An abundance of methods have been developed over the years to perform the frequency analysis (FA) of extreme environmental variables. Although numerous comparisons between these methods have been implemented, no general comparison framework has been agreed upon so far. The objective of this paper is to build the foundation of a data-based comparison framework, which aims at complementing more standard comparison schemes based on Monte Carlo simulations or statistical testing. This framework is based on the following general principles: (i) emphasis is put on the predictive ability of competing FA implementations, rather than their sole descriptive ability measured by some goodness-of-fit criterion; (ii) predictive ability is quantified by means of reliability indices, describing the consistency between validation data (not used for calibration) and FA predictions; (iii) stability is also quantified, i.e., the ability of a FA implementation to yield similar estimates when calibration data change; and (iv) the necessity to subject uncertainty estimates to the same scrutiny as point estimates is recognized, and a practical approach based on the use of the predictive distribution is proposed for this purpose. This framework is then applied to a case study involving 364 gauging stations in France, where 10 FA implementations are compared. These implementations correspond to the local, regional, and local-regional estimation of Gumbel and generalized extreme value distributions. Results show that reliability and stability indices are able to reveal marked differences between FA implementations. Moreover, the case study also confirms that using the predictive distribution to indirectly scrutinize uncertainty estimates is a viable approach, with distinct FA implementations showing marked differences in the reliability of their uncertainty estimates. The proposed comparison framework therefore constitutes a valuable tool to compare the predictive reliability of competing FA implementations, along with the reliability of their uncertainty estimates.

International audience; Suite à des questionnements récurrents sur la méthode GRADEX, utilisée depuis près de 40 ans, EDF a développé la méthode SCHADEX. C’est une méthode probabiliste de prédétermination des crues extrêmes qui combine, au moyen d’une simulation stochastique, un modèle probabiliste de pluies extrêmes basé sur des types de temps et un modèle hydrologique conceptuel. SCHADEX est devenue fin 2008 la méthode de référence pour les barrages d’EDF, et a été appliquée depuis sur plusieurs dizaines de bassins. La méthode fait l’objet depuis son origine d’actions d’évaluation scientifique au sein notamment de projets nationaux et internationaux d’inter-comparaison. La performance de son modèle probabiliste de pluies extrêmes a été prouvée sur une vaste base de données pluviométriques. Les résultats de la méthode SCHADEX ont été comparés à ceux de GRADEX sur 60 bassins, les deux méthodes étant appliquées avec les mêmes données. Il en ressort que l’écart sur les estimations de débit millénal est faible en moyenne, mais peut être significatif dans certains cas, notamment pour de petits bassins de montagne. La méthode continue à faire l’objet de développements complémentaires, notamment sur l’utilisation de coefficients de forme variables, l’application en bassin non-jaugés ou en contexte non-stationnaire. / Due to recurring questions about the GRADEX method (used for almost 40 years) EDF has developed the SCHADEX method. It is a probabilistic method for extreme floods estimation which combines, within a stochastic simulation framework, a probabilistic model of extreme rainfall based on weather patterns, and a conceptual hydrological model. Late 2008, SCHADEX has become the reference method for EDF dams, and has since been applied to several dozen watersheds. Since its origin, the method has been subject of scientific evaluation activities, in particular within national and international inter-comparison projects. The performance of the probabilistic model of extreme rainfall has been proven on a large dataset of rainfall. The results of the SCHADEX method have been compared on 60 watersheds with those of GRADEX, both methods being applied with the same data. It shows that the difference of 1000-years flood estimations is low on average but can be significant in some cases, especially for small mountain basins. Further developments of the method are still underway, in particular the use of variable peak-to-volume ratios, and the application of SCHADEX in non-gauged watersheds or non-stationary context.

International audience; In flood risk assessment the methods can be divided in two families: deterministic methods and probabilistic methods. In the French hydrological community the probabilistic methods are historically preferred to the deterministic ones. Presently a French research project named EXTRAFLO (RiskNat Program of the French National Research Agency, https://extraflo.cemagref.fr) deals with the design values for extreme rainfall and floods. The object of this project is to carry out a comparison of the main methods used in France for estimating extreme values of rainfall and floods, to obtain a better grasp of their respective fields of application. In this framework we present the results of Action 7 of EXTRAFLO project. Focusing on five French watersheds, we compare the main extreme flood estimation methods used in French background: (i) standard flood frequency analysis (Gumbel and GEV distribution), (ii) regional flood frequency analysis (regional Gumbel and GEV distribution), (iii) flood frequency analysis improved by historical information (Naulet et al., 2005), (iv) simplify probabilistic method based on rainfall information (i.e. Gradex method (CFGB, 1994), Agregee method (Margoum, 1992) and Speed method (Cayla, 1995)), (v) flood frequency analysis by continuous simulation approach and based on rainfall information (i.e. Schadex method (Paquet et al., 2006, Garavaglia et al. 2010), Shyreg method (Lavabre et al., 2003)) and (vi) multifractal approach. The main result of this comparative study is that probabilistic methods based on additional information (i.e. regional, historical and rainfall information) provide better estimations than the standard flood frequency analysis. Another interesting result is that, the differences between the various extreme flood quantile estimations of compared methods increase with return period, staying relatively moderate up to 100-years return levels. Results and discussions are here illustrated throughout with the example of the Ardeche watershed (South of France).

Consider a scenario where one aims to learn models from data being characterized by very large fluctuations that are neither attributable to noise nor outliers. This may be the case, for instance, when examining supermarket ketchup sales, predicting earthquakes and when conducting financial data analysis. In such a situation, the standard central limit theorem does not apply, since the associated Gaussian distribution exponentially suppresses large fluctuations. In this paper, we argue that, in many cases, the incorrect assumption leads to misleading and incorrect data mining results. We illustrate this argument against synthetic data, and show some results against stock market data., International Conference on Knowledge Discovery and Information Retrieval, KDIR 2011, 26 October 2011 through 29 October 2011, Paris

International audience; Extreme value analysis is one of the cornerstones of hazard quantication and risk assessment. Its basic objective is to estimate the distribution of some environmental variable X, e.g. annual maximum of the areal rainfall over some catchment, annual maximum ood, etc. This distribution can be used to estimate the exceedance probability of a given value of X (often expressed in terms of return period), or alternatively, to estimate the p-quantile of X. The estimation of quantiles is of primary importance since they are used to design civil engineering structures (e.g. dams, reservoirs, bridges) or to map hazard-prone areas where restrictions may be enforced (e.g. building restrictions in ood zones). Extreme value analysis has been the subject of extensive research, yielding an abundance of approaches. In Hydrology, several families of methods exist, including (but not limited to): Standard application of extreme value theory (EVT), i.e. estimation of an extreme value distribution based on a sample of block maxima or peaks over a high threshold Climate/Weather-informed application of EVT. This family of methods uses additional meteorological (e.g., weather type, [1]) or climatic (e.g. Interdecadal Pacic Oscillation IPO, [2]) information. Regional approaches, conjointly using data from several sites to perform the inference, which may improve the precision of estimates. Model-based approaches, using a simulation model reproducing the main characteristics of the environmental variable (e.g. [1]). In practice, users and practitioners of extreme value analyses may feel lost facing such an abundance of methods. Consequently, it is necessary to provide them with practical guidelines to choose and implement adequate methods, depending on the conditions of application (e.g. availability of long series, geographical area, type of hydrological regime, etc.). This presentation describes a methodological framework to perform a data-based comparison of competing approaches for predicting extremes. This framework is based on the following principles: The objective is to assess the predictive performance of competing methods (as opposed to standard goodness-of- evaluations). This requires decomposing the available dataset into estimation / validation sub-samples. Reliably quantifying uncertainties is recognized as a primary objective, and the issue of scrutinizing uncertainty estimates is discussed. To this aim, we make use of predictive distributions for extremes, obtained by integrating out parameter uncertainty. Such predictive distributions are standard in a Bayesian context [4] but can also be derived in a frequentist context [5]. Reliability indices are derived in order to compare the performances of competing methods on an objective basis. In a second step, this framework is used to perform a thorough comparison between approaches currently used in France for extreme prediction. The comparison is based on an extensive dataset of long series of rainfall and runo (about 40-50 years of daily data), available for hundreds of sites over France. Results demonstrate the ability of the comparison framework to distinguish between "good" and "bad" approaches, and yield valuable insights into the optimal ambit of each approach.

International audience; Frequency analysis (FA) is one of the cornerstones of hazard quantification and risk assessment. Its basic objective is to estimate the distribution of some environmental variable X, e.g. annual maximum of the areal rainfall over some catchment, annual maximum flood, etc. This distribution can be used to estimate the exceedance probability of a given value of X (often expressed in terms of return period), or alternatively, to estimate the p-quantile of X, i.e. the value having an exceedance probability equal to 1-p. The estimation of quantiles is of primary importance since they are used to design civil engineering structures (e.g. dams, reservoirs, bridges) or to map hazard-prone areas where restrictions may be enforced (e.g. building restrictions in flood zones). FA has been the subject of extensive research, yielding an abundance of approaches. In practice, FA users and practitioners may feel lost facing such an abundance of methods. Consequently, several initiatives aimed at assisting users in realizing their analyses using best-practice methods. In addition to these end-user-oriented guideline documents, a large number of comparisons between competing methods have been reported in the research literature. The French National research project EXTRAFLO aims to perform a thorough comparison between FA approaches currently used in France, based on an extensive dataset of long series of rainfall and runoff. This poster provides a detailed description of the methodology used to perform the comparison, and presents preliminary results of its application to a large runoff dataset. More precisely, the following topics are presented: Difference between descriptive and predictive ability, and the need to decompose the dataset into calibration / validation sub-samples The issue of scrutinizing uncertainty estimates is discussed, and a method based on the concept of predictive distribution is proposed in order to compare the reliability of competing uncertainty estimates. Reliability indices are derived in order to compare the performances of competing methods on an objective basis. This methodological framework is first applied to synthetic data to verify the ability of the comparison framework to distinguish between good and bad FA approaches. Preliminary results based on a large runoff dataset are then discussed.

Frequency analysis (FA) is one of the cornerstones of hazard quantification and risk assessment. Its basic objective is to estimate the distribution of some environmental variable X, e.g. annual maximum of the areal rainfall over some catchment, annual maximum flood, etc. This distribution can be used to estimate the exceedance probability of a given value of X (often expressed in terms of return period), or alternatively, to estimate the p-quantile of X, i.e. the value having an exceedance probability equal to 1-p. The estimation of quantiles is of primary importance since they are used to design civil engineering structures (e.g. dams, reservoirs, bridges) or to map hazard-prone areas where restrictions may be enforced (e.g. building restrictions in flood zones). FA has been the subject of extensive research, yielding an abundance of approaches. In practice, FA users and practitioners may feel lost facing such an abundance of methods. Consequently, several initiatives aimed at assisting users in realizing their analyses using best-practice methods. In addition to these end-user-oriented guideline documents, a large number of comparisons between competing methods have been reported in the research literature. The French National research project EXTRAFLO aims to perform a thorough comparison between FA approaches currently used in France, based on an extensive dataset of long series of rainfall and runoff. This poster provides a detailed description of the methodology used to perform the comparison, and presents preliminary results of its application to large rainfall and runoff datasets. More precisely, the following topics are presented: 1. Presentation of the datasets, including more than 1000 series of daily runoff and more than 2000 series of daily rainfall 2. Decomposition of the datasets into calibration/validation sub-samples 3. The issue of scrutinizing uncertainty estimates is discussed, and a method based on the concept of predictive distribution is proposed in order to compare the reliability of competing uncertainty estimates. 4. Reliability indices are derived in order to compare the performances of competing methods on an objective basis. 5. This methodological framework is applied to the datasets and preliminary results are discussed.

National audience; Climate change is widely considered as a reality in the scientific community. However, its impact on hydrological extremes is still difficult to observe and to anticipate. A national research project has been set up in order to answer the following questions: How to detect changes in hydro-climatic series? What are the observed changes for hydrological extremes in France? How to take into account such possible changes in the methods used for hydrological risks assessment? In terms of methodology, numerous tests for change detection were compared based on Monte Carlo simulations. Results allowed identifying the key criterions to be considered when choosing a test for a given variable. Moreover, regional methods for change detection were also studied. More accurately, the concepts of field significance and regional consistency were investigated. The application of these methods to a data set arising from 195 streamgauge stations did not show any global pattern of change in high and low flows in France. However, some consistent changes were detected at the smaller scale of hydro-climatic regions: In the Pyreneans, a decrease in runoff was observed in high and low flows. In the Alps, less severe droughts were observed, together with earlier snowmelt-related runoffs and decreasing glacier-related volumes. A slight increase in flood intensity was also observed in the North-East. Lastly, the methods used in the context of hydrological risk assessment are revisited in order to account for possible changes in the hydrological regime.; Le changement climatique est une réalité largement reconnue aujourd'hui dans la communauté scientifique. Néanmoins, son impact sur les extrêmes hydrologiques reste délicat à observer et à prévoir. Dans ce contexte, un projet national de recherches a été mis en place, visant à apporter des éléments de réponse aux trois questions suivantes : comment détecter des changements dans les séries hydro-climatiques ? Quelles sont les évolutions observées sur les régimes extrêmes des cours d'eau en France métropolitaine ? Comment prendre en compte ces éventuels changements dans les méthodes de gestion des risques hydrologiques ? Au niveau méthodologique, un grand nombre de tests statistiques dédiés à la détection de changements ont été étudiés et comparés sur la base de simulations Monte Carlo. Les résultats ont permis de clarifier les critères guidant le choix du test à appliquer à une variable donnée. De plus, les méthodes de détection de changements à une échelle régionale ont également été étudiées. Plus particulièrement, nous nous sommes intéressés aux concepts de significativité régionale et de cohérence régionale. L'application de ces outils à 195 stations hydrométriques en France ne montre pas de changement généralisé, ni en crue, ni en étiage, ni en régime. A plus petite échelle, certaines évolutions sont cependant détectées : dans la région pyrénéenne, une tendance vers une baisse des écoulements a été détectée, à la fois en crue et en étiage. Dans les Alpes, des étiages moins sévères ont été relevés, ainsi qu'une précocité accrue de la fonte et une augmentation du volume de fonte écoulé par les bassins au régime glaciaire. Dans le Nord-Est de la France, on remarque une légère augmentation des intensités de crue. Enfin, nous proposons une mise à jour des méthodes utilisées dans le cadre de la gestion des risques hydrologiques pour prendre en compte d'éventuelles évolutions temporelles.

[Departement_IRSTEA]RE [TR1_IRSTEA]RIE / TRANSFEAU; International audience; This paper describes regional methods for assessing field significance and regional consistency for trend detection in hydrological extremes. Four procedures for assessing field significance are compared on the basis of Monte Carlo simulations. Then three regional tests, based on a regional variable, on the regional average Mann-Kendall test, and a new semiparametric approach, are tested. The latter was found to be the most adequate to detect consistent changes within homogeneous hydro-climatic regions. Finally, these procedures are applied to France, using daily discharge data arising from 195 gauging stations. No generalized change was found at the national scale on the basis of the field significance assessment of at-site results. Hydro-climatic regions were then defined, and the semiparametric procedure applied. Most of the regions showed no consistent change, but three exceptions were found: in the northeast flood peaks were found to increase, in the Pyrenees high and low flows showed decreasing trends, and in the Alps, earlier snowmelt-related floods were detected, along with less severe drought and increasing runoff due to glacier melting. The trend affecting floods in the northeast was compared to changes in rainfall, using rainfall-runoff simulation. The results showed flood trends consistent with the observed rainfall.

International audience; The following section presents a synthetic description of the models that were used in the various articles of this Red Book. The following models are covered (descriptions can be found in alphabetical order) : AFFDEF, GR4H, GR4J, HBV, HYDROTEL, IHACRES, MODSPA, MORDOR, NOAH, RRMT, SAC-SMA, SMAR, SWAP, SWB, VIC.

International audience

International audience

International audience