Search

INRAE a mené une réflexion prospective collective sur les risques environnementaux dans une perspective inter- et transdisciplinaire inscrite dans le cadre des sciences de la durabilité. Les risques pour les sociétés, les écosystèmes et les milieux ont été considérés conjointement et de façon holistique. Cet article résume les quatre priorités scientifiques identifiées, exprimées de manière transverse. Une feuille de route pour atteindre ces objectifs est également proposée, incluant les ressources humaines, les infrastructures, l’animation scientifique et les partenariats. Ces conclusions devraient participer à la structuration de la recherche française et internationale, et constituer ainsi une contribution significative à l’effort requis par l’urgence environnementale actuelle.

The value of historical data for flood frequency analysis has been acknowledged and studied for a long time. A specific statistical framework must be used to comply with the censored nature of historical data, for which only floods large enough to induce written records or to trigger flood marks are usually recorded. It is assumed that all floods which exceeded a given perception threshold were recorded as written testimonies or flood marks. Conversely, all years without a flood record in the historical period are assumed to have had a maximum discharge below the perception threshold. This paper proposes a binomial model that explicitly recognizes the uncertain nature of both the perception threshold and the starting date of the historical period. This model is applied to a case study for the Rhône River at Beaucaire, France, where a long (1816–2020) systematic series of annual maximum discharges is available along with a collection of 13 historical floods from documentary evidence over 3 centuries (1500–1815). Results indicate that the inclusion of historical floods reduces the uncertainty of 100- or 1000-year flood quantiles, even when only the number of perception threshold exceedances is known. However, ignoring the uncertainty around the perception threshold leads to a noticeable underestimation of flood quantile uncertainty. A qualitatively similar conclusion is found when ignoring the uncertainty around the historical period length. However, its impact on flood quantile uncertainty appears to be much smaller than that of the perception threshold. [ABSTRACT FROM AUTHOR]

The value of historical data for flood frequency analysis has been acknowledged and studied for a long time. A specific statistical framework must be used to comply with the censored nature of historical data. Indeed, it is assumed that all floods having exceeded a given perception threshold were recorded as written testimonies or flood marks. Conversely, all years without a flood record in the historical period are assumed to have a maximum discharge below the perception threshold. This paper proposes a Binomial model which explicitly recognizes the uncertain nature of both the perception threshold and the starting date of the historical period. This model is applied to a case study for the Rhône River at Beaucaire, France, where a long (1816–2020) systematic series of annual maximum discharges is available along with a collection of 13 historical floods from documentary evidences over three centuries (1500–1815). Results indicate that the inclusion of historical floods reduces the uncertainty of 100- or 1000-year flood quantiles, even when only the number of perception threshold exceedances is known. However, ignoring the uncertainty around the perception threshold leads to a noticeable underestimation of flood quantiles uncertainty. A qualitatively similar conclusion is found when ignoring the uncertainty around the historical period length. However, its impact on flood quantiles uncertainty appears to be much smaller than that of the perception threshold. [ABSTRACT FROM AUTHOR]

Given an increase in data-driven automated decision-making based on machine learning (ML) models, it is imperative that, along with tools to develop and improve such models, there are sufficient capabilities to analyze and assess models with respect to potential biases. We present the package mlr3fairness, a collection of metrics and methods that allow for the assessment of bias in machine learning models. Our package implements a variety of widely used fairness metrics that can be used to audit models for potential biases, along with a set of visualizations that can help to provide additional insights into such biases. mlr3fairness furthermore integrates bias mitigation methods for machine learning models through data pre-processing or post-processing of predictions. These allow practitioners to trade off performance and fairness metrics that are appropriate for their use case. [ABSTRACT FROM AUTHOR]

European flood frequency and intensity change on a multidecadal scale. Floods were more frequent in the 19th (Central Europe) and early 20th century (Western Europe) than during the mid-20th century and again more frequent since the 1970s. The causes of this variability are not well understood and the relation to climate change is unclear. Palaeoclimate studies from the northern Alps suggest that past flood-rich periods coincided with cold periods. In contrast, some studies suggest that more floods might occur in a future, warming world. Here we reconcile the apparent contradiction by addressing and quantifying the contribution of atmospheric processes to multidecadal flood variability. For this, we use long series of annual peak streamflow, daily weather data, reanalyses, and reconstructions. We show that both changes in atmospheric circulation and moisture content affected multidecadal changes of annual peak streamflow in Central and Western Europe over the past two centuries. We find that during the 19th and early 20th century, atmospheric circulation changes led to high peak values of moisture flux convergence. The circulation was more conducive to strong and long-lasting precipitation events than in the mid-20th century. These changes are also partly reflected in the seasonal mean circulation and reproduced in atmospheric model simulations, pointing to a possible role of oceanic variability. For the period after 1980, increasing moisture content in a warming atmosphere led to extremely high moisture flux convergence. Thus, the main atmospheric driver of flood variability changed from atmospheric circulation variability to water vapour increase. [ABSTRACT FROM AUTHOR]