Search

Depth–duration–frequency curves estimate the rainfall intensity patterns for various return periods and rainfall durations. An empirical model based on the generalized extreme value distribution is presented for hourly maximum rainfall, and improved by the inclusion of daily maximum rainfall, through the extremal indexes of 24 hourly and daily rainfall data. The model is then divided into two sub-models for the short and long rainfall durations. Three likelihood formulations are proposed to model and compare independence or dependence hypotheses between the different durations. Dependence is modelled using the bivariate extreme logistic distribution. The results are calculated in a Bayesian framework with a Markov Chain Monte Carlo algorithm. The application to a data series from Marseille shows an improvement of the hourly estimations thanks to the combination between hourly and daily data in the model. Moreover, results are significantly different with or without dependence hypotheses: the dependence between 24 and 72 h durations is significant, and the quantile estimates are more severe in the dependence case. [ABSTRACT FROM AUTHOR]

Summary As machine learning has become increasingly popular over the last few decades, so too has the number of machine-learning interfaces for implementing these models. Whilst many R libraries exist for machine learning, very few offer extended support for survival analysis. This is problematic considering its importance in fields like medicine, bioinformatics, economics, engineering and more. mlr3proba provides a comprehensive machine-learning interface for survival analysis and connects with mlr3 's general model tuning and benchmarking facilities to provide a systematic infrastructure for survival modelling and evaluation. Availability and implementation mlr3proba is available under an LGPL-3 licence on CRAN and at https://github.com/mlr-org/mlr3proba , with further documentation at https://mlr3book.mlr-org.com/survival.html. [ABSTRACT FROM AUTHOR]

Flood frequency analysis (FFA), a widely used method to estimate flood hazard, is affected by several sources of uncertainty. Extending flood samples by reanalyzing historical systematic stage records has the potential to reduce sampling uncertainty, but the historical flood discharges derived from this reanalysis are generally affected by large uncertainties. This paper explores whether historical stage records improve design flood estimates through a chain of uncertainty estimation methods for FFA. Uncertainties are estimated and propagated from stage and rating curves to design flood estimates using Monte Carlo procedures. The role of both streamflow and sampling uncertainties in design flood estimation is examined. This procedure is applied to the 205-year long stage series of the Rhône River at Beaucaire, France (95 590 km2). The estimated streamflow 95% uncertainty varies from 30% (XIXth Century) to 5% (1967–2020). The total uncertainty of design flood is significantly reduced when the length of the series increases from 20 to 100 years due to sampling uncertainty reduction. However, the total uncertainty remains stable beyond this sample size: this is because large uncertainties affecting the XIXth Century flood discharges compensate for the reduction in sampling uncertainty. Enlarging the sample size to two centuries leads to including the two largest known floods in 1840 and 1856. In turn, this induces a 15% increase of the 1000-year flood estimates. • Stage and rating curve uncertainties are propagated to design flood estimates. • XIXth Century floods uncertainty contribution on design flood uncertainty is large. • Sampling uncertainty decrease is offset by streamflow uncertainty increase. • Design flood uncertainty is not reduced beyond 100 years sample size. [ABSTRACT FROM AUTHOR]

This paper presents some improvements of a probabilistic quantitative precipitation forecasting method based on analogues, formerly developed on small basins located in South-Eastern France. The scope is extended to large scale basins mainly influenced by frontal systems, considering a case study area related to the Saône river, a large basin in eastern France. For a given target situation, this method consists in searching for the most similar situations observed in a historical meteorological archive. Precipitation amounts observed during analogous situations are then collected to derive an empirical predictive distribution function, i.e. the probabilistic estimation of the precipitation amount expected for the target day. The former version of this forecasting method (Bontron, 2004) has been improved by introducing two innovative variables: temperature, that allows taking seasonal effects into account and vertical velocity, which enables a better characterization of the vertical atmospheric motion. The new algorithm is first applied in a perfect prognosis context (target situations come from a meteorological reanalysis) and then in an operational forecasting context (target situations come from weather forecasts) for a three years period. Results show that this approach yields useful forecasts, with a lower false alarm rate and improved performances from the present day D to day D + 2. [ABSTRACT FROM AUTHOR]

Summary The global and regional influence of the El Niño-Southern Oscillation (ENSO) phenomenon on extreme precipitation was analyzed using a global database comprising over 7000 high quality observation sites. To better quantify possible changes in relatively rare design-relevant precipitation quantiles (e.g. the 1 in 10 year event), a Bayesian regional extreme value model was used, which employed the Southern Oscillation Index (SOI) – a measure of ENSO – as a covariate. Regions found to be influenced by ENSO include parts of North and South America, southern and eastern Asia, South Africa, Australia and Europe. The season experiencing the greatest ENSO effect varies regionally, but in most of the ENSO-affected regions the strongest effect happens in boreal winter, during which time the 10-year precipitation for |SOI| = 20 (corresponding to either a strong El Niño or La Niña episode) can be up to 50% higher or lower than for SOI = 0 (a neutral phase). Importantly, the effect of ENSO on extreme precipitation is asymmetric, with most parts of the world experiencing a significant effect only for a single ENSO phase. This finding has important implications on the current understanding of how ENSO influences extreme precipitation, and will enable a more rigorous theoretical foundation for providing quantitative extreme precipitation intensity predictions at seasonal timescales. We anticipate that incorporating asymmetric impacts of ENSO on extreme precipitation will help lead to better-informed climate-adaptive design of flood-sensitive infrastructure. [ABSTRACT FROM AUTHOR]

Highlights: [•] Development of a rigorous spatio-temporal regional frequency analysis framework. [•] Enables verification of asymmetric ENSO impact on Queensland precipitation using multi-site local rainfall data. [•] Enables predictions of multi-site local rainfall extremes conditional on ENSO, with uncertainties. [•] During a strong La Niña, the most likely 1 in 100year rainfall can be 20–50% higher than normal. [•] Regionalizing the effect of ENSO strongly reduces uncertainty. [ABSTRACT FROM AUTHOR]

This paper empirically investigates the asymptotic behaviour of the flood probability distribution and more precisely the possible occurrence of heavy tail distributions, generally predicted by multiplicative cascades. Since heavy tails considerably increase the frequency of extremes, they have many practical and societal consequences. A French database of 173 daily discharge time series is analyzed. These series correspond to various climatic and hydrological conditions, drainage areas ranging from 10 to 105 km2, and are from 22 to 95 years long. The peaks-over-threshold method has been used with a set of semi-parametric estimators (Hill and Generalized Hill estimators), and parametric estimators (maximum likelihood and L-moments). We discuss the respective interest of the estimators and compare their respective estimates of the shape parameter of the probability distribution of the peaks. We emphasize the influence of the selected number of the highest observations that are used in the estimation procedure and in this respect the particular interest of the semi-parametric estimators. Nevertheless, the various estimators agree on the prevalence of heavy tails and we point out some links between their presence and hydrological and climatic conditions. [ABSTRACT FROM AUTHOR]

The Cévennes–Vivarais Mediterranean Hydrometeorological Observatory (OHM-CV) is a research initiative aimed at improving the understanding and modeling of the Mediterranean intense rain events that frequently result in devastating flash floods in southern France. A primary objective is to bring together the skills of meteorologists and hydrologists, modelers and instrumentalists, researchers and practitioners, to cope with these rather unpredictable events. In line with previously published flash-flood monographs, the present paper aims at documenting the 8–9 September 2002 catastrophic event, which resulted in 24 casualties and an economic damage evaluated at 1.2 billion euros (i.e., about 1 billion U.S. dollars) in the Gard region, France. A description of the synoptic meteorological situation is first given and shows that no particular precursor indicated the imminence of such an extreme event. Then, radar and rain gauge analyses are used to assess the magnitude of the rain event, which was particularly remarkable for its spatial extent with rain amounts greater than 200 mm in 24 h over 5500 km2. The maximum values of 600–700 mm observed locally are among the highest daily records in the region. The preliminary results of the postevent hydrological investigation show that the hydrologic response of the upstream watersheds of the Gard and Vidourle Rivers is consistent with the marked space–time structure of the rain event. It is noteworthy that peak specific discharges were very high over most of the affected areas (5–10 m3 s-1 km-2) and reached locally extraordinary values of more than 20 m3 s-1 km-2. A preliminary analysis indicates contrasting hydrological behaviors that seem to be related to geomorphological factors, notably the influence of karst in part of the region. An overview of the ongoing meteorological and hydrological research projects devoted to this case study within the OHM-CV is finally presented. [ABSTRACT FROM AUTHOR]

Background: With modern methods in biotechnology, the search for biomarkers has advanced to a challenging statistical task exploring high dimensional data sets. Feature selection is a widely researched preprocessing step to handle huge numbers of biomarker candidates and has special importance for the analysis of biomedical data. Such data sets often include many input features not related to the diagnostic or therapeutic target variable. A less researched, but also relevant aspect for medical applications are costs of different biomarker candidates. These costs are often financial costs, but can also refer to other aspects, for example the decision between a painful biopsy marker and a simple urine test. In this paper, we propose extensions to two feature selection methods to control the total amount of such costs: greedy forward selection and genetic algorithms. In comprehensive simulation studies of binary classification tasks, we compare the predictive performance, the run-time and the detection rate of relevant features for the new proposed methods and five baseline alternatives to handle budget constraints., Results: In simulations with a predefined budget constraint, our proposed methods outperform the baseline alternatives, with just minor differences between them. Only in the scenario without an actual budget constraint, our adapted greedy forward selection approach showed a clear drop in performance compared to the other methods. However, introducing a hyperparameter to adapt the benefit-cost trade-off in this method could overcome this weakness., Conclusions: In feature cost scenarios, where a total budget has to be met, common feature selection algorithms are often not suitable to identify well performing subsets for a modelling task. Adaptations of these algorithms such as the ones proposed in this paper can help to tackle this problem.