International audienceThis paper presents a calibration framework for a precipitation-runoff model for flood prediction in a mesoscale Alpine basin with discharges strongly influenced by hydraulic works. The developed methodology addresses two classical hydrological calibration challenges: computational limitations to run optimization algorithms for distributed hourly models and the absence of concomitant meteorological and natural discharge time series. The presented processes-oriented, multi-signal approach is based on hydrological data from a variety of sources and for different periods, corresponding to various spatial scales. The model parameters are calibrated by sequentially minimizing differences between observed and simulated values for different hydrological signals and signatures such as: (a) the phase of precipitations, (b) the time evolution of point-scale snow heights, (c) the mean inter-annual cycle of daily discharges, and (d) timing of snowmelt-induced spring runoff. We compare the model performance to a benchmark model obtained by simply using the globally optimal parameter values from the nearest gauged and non perturbed catchment. For prediction of flow seasonality and also extreme events, the calibration methodology outperforms the benchmark

Abstract. In France, nuclear facilities were designed around very low probabilities of failure. Nevertheless, some extreme climatic events have given rise to exceptional observed surges (outliers) much larger than other observations, and have clearly illustrated the potential to underestimate the extreme water levels calculated with the current statistical methods. The objective of the present work is to conduct a comparative study of three approaches to extreme value analysis, including the annual maxima (AM), the peaks-overthreshold (POT) and the r-largest order statistics (r-LOS). These methods are illustrated in a real analysis case study. All data sets were screened for outliers. Non-parametric tests for randomness, homogeneity and stationarity of time series were used. The shape and scale parameter stability plots, the mean excess residual life plot and the stability of the standard errors of return levels were used to select optimal thresholds and r values for the POT and r-LOS method, respectively. The comparison of methods was based on (i) the uncertainty degrees, (ii) the adequacy criteria and tests, and (iii) the visual inspection. It was found that the r-LOS and POT methods have reduced the uncertainty on the distribution parameters and return level estimates and have systematically shown values of the 100 and 500-year return levels smaller than those estimated with the AM method. Results have also shown that none of the compared methods has allowed a good fit at the right tail of the distribution in the presence of outliers. As a perspective, the use of historical information was proposed in order to increase the representativeness of outliers in data sets. Findings are of practical relevance, not only to nuclear energy operators in France, for applications in storm surge hazard analysis and flood management, but also for the optimal planning and design of facilities to withstand extreme environmental conditions, with an appropriate level of risk.

Abstract. Nuclear power plants located in the French Atlantic coast are designed to be protected against extreme environmental conditions. The French authorities remain cautious by adopting a strict policy of nuclear-plants flood prevention. Although coastal nuclear facilities in France are designed to very low probabilities of failure (e.g., 1000-year surge), exceptional surges (outliers induced by exceptional climatic events) have shown that the extreme sea levels estimated with the current statistical approaches could be underestimated. The estimation of extreme surges then requires the use of a statistical analysis approach having a more solid theoretical motivation. This paper deals with extreme-surge frequency estimation using historical information (HI) about events occurred before the systematic record period. It also contributes to addressing the problem of the presence of outliers in data sets. The frequency models presented in the present paper have been quite successful in the field of hydrometeorology and river flooding but they have not been applied to sea level data sets to prevent marine flooding.In this work, we suggest two methods of incorporating the HI: the peaks-over-threshold method with HI (POTH) and the block maxima method with HI (BMH). Two kinds of historical data can be used in the POTH method: classical historical maxima (HMax) data, and over-a-threshold supplementary (OTS) data. In both cases, the data are structured in historical periods and can be used only as complement to the main systematic data. On the other hand, in the BMH method, the basic hypothesis in statistical modeling of HI is that at least one threshold of perception exists for the whole period (historical and systematic) and that during a giving historical period preceding the period of tide gauging, only information about surges above this threshold have been recorded or archived. The two frequency models were applied to a case study from France, at the La Rochelle site where the storm Xynthia induced an outlier, to illustrate their potentials, to compare their performances and especially to analyze the impact of the use of HI on the extreme-surge frequency estimation.

Nuclear power plants in France are designed to withstand natural hazards. Nevertheless, some exceptional storm surges, considered as outliers, are not properly addressed by classical statistical models. Local frequency estimations can be obtained with acceptable uncertainties if the data set containing the extreme storm surge levels is sufficiently complete and not characterized by the presence of an outlier. Otherwise, additional information such as regional and historical storm surges may be used to mitigate the lack of data and the influence of the outlier by increasing its representativeness in the sample. The objective of the present work is to develop a regional frequency model (RFM) using historical storm surges. Here we propose a RFM using historical information (HI). The empirical spatial extremogram is used herein to form a homogenous region of interest centered on a target site. A related issue regards the reconstitution, at the target site and from its neighbors, of missed storm surges with a multiple linear regression (MLR). MLR analysis can be considered conclusive if available observations at neighboring sites are informative enough and the reconstitution results meet some criteria during the cross-validation process. A total of 35 harbors located on the French (Atlantic and English Channel) and British coasts are used as a whole region with the La Rochelle site as a target site. The Peaks-Over-Threshold frequency model is used. The results are compared to those of an existing model based on the index flood method. The use of HI in the developed RFM increases the representativeness of the outlier in the sample. Fitting results at the right tail of the distribution appear more adequate and the 100-year return level is about 30 cm higher. The 100-year return level in the initial fitting has a return period of about 30 years in the updated fitting.

Abstract. Nuclear power plants located in the French Atlantic coast are designed to be protected against extreme environmental conditions. The French authorities remain cautious by adopting a strict policy of nuclear plants flood prevention. Although coastal nuclear facilities in France are designed to very low probabilities of failure (e.g. 1000 year surge), exceptional surges (outliers induced by exceptional climatic events) had shown that the extreme sea levels estimated with the current statistical approaches could be underestimated. The estimation of extreme surges then requires the use of a statistical analysis approach having a more solid theoretical motivation. This paper deals with extreme surge frequency estimation using historical information (HI) about events occurred before the systematic record period. It also contributes to addressing the problem of the presence of outliers in data sets. The frequency models presented in the present paper have been quite successful in the field of hydrometeorology and river flooding but they have not been applied to sea levels data sets to prevent marine flooding. In this work, we suggest two methods of incorporating the HI: the Peaks-Over-Threshold method with HI (POTH) and the Block Maxima method with HI (BMH). Two kinds of historical data can be used in the POTH method: classical Historical Maxima (HMax) data, and Over a Threshold Supplementary (OTS) data. In both cases, the data are structured in historical periods and can be used only as complement to the main systematic data. On the other hand, in the BMH method, the basic hypothesis in statistical modeling of HI is that at least one threshold of perception exists for the whole period (historical and systematic) and that during a giving historical period preceding the period of tide gauging, only information about surges above this threshold have been recorded or archived. The two frequency models were applied to a case study from France, at the La Rochelle site where the storm Xynthia induced an outlier, to illustrate their potentials, to compare their performances and especially to analyze the impact of the use of HI on the extreme surge frequency estimation.