Search

International audience; Niamey, the capital of Niger, is particularly prone to floods, since it is on the banks of the Niger River, which in its middle basin has two flood peaks: one in summer (the red flood) and one in winter (the black flood). In 2020, the Niger River in Niamey reached its all-time highest levels following an abundant rainy season. On the other hand, the floods in Niamey have been particularly frequent in the last decade, a symptom of a change in hydroclimatic behaviour already observed since the end of the great droughts of the 1970s and 1980s and which is identified with the name of Sahelian Paradox. This study, starting from the analysis of the 2020 flood and from the update of the rating curve of the Niamey hydrometric station, analyses the rainfall–runoff relationship on the Sahelian basins of the Medium Niger River Basin (MNRB) that are at the origin of the local flood. The comparative analysis of runoffs, annual maximum flows (AMAX) and runoff coefficients with various rainfall indices calculated on gridded datasets allowed to hydroclimatically characterise the last decade as a different period from the wet one before the drought, the dry one and the post-drought one. Compared to the last one, the current period is characterised by a sustained increase in hydrological indicators (AMAX +27%) consistent with the increase in both the accumulation of precipitation (+11%) and the number (+51%) and magnitude (+54%) of extreme events in the MNRB. Furthermore, a greater concentration of rainfall and extremes (+78%) in August contributes to reinforcing the red flood’s positive anomalies (+2.23 st.dev in 2020). The study indicates that under these conditions the frequency of extreme hydrological events in Niamey will tend to increase further also because of the concurrence of drivers such as river-bed silting and levee effects. Consequently, the study concludes with the need for a comprehensive flood-risk assessment on the Niamey city that considers both recent hydroclimatic trends and urbanisation dynamics in flood zones hence defining the most appropriate risk-reduction strategies.

River discharge time series are established using "rating curves", which are models with stage as input and discharge as output. Unfortunately, many hydrometric stations have an unstable stage-discharge relation, particularly because of the change in the river bed during floods. These "rating shifts" pose a problem both for the establishment of long-term hydrological series (retrospective analysis) and for the delivery of real-time data, for example for flood forecasting, with quantified and reliable uncertainties. The existing methods for updating the rating curve are based on a statistical analysis of past calibration data (the gaugings) or on empirical rules. This thesis aims at developing some original methods for the automatic detection of rating shifts and the estimation of their magnitude in both retrospective and real time: a method of segmentation of the residuals between the gaugings and a base rating curve, an analysis of the stage recessions and a method for detecting potential shifts from an indicator available in real time (e.g. cumulative sediment transport). The probabilistic approach allows on the one hand to take into account the uncertainty of the information on the shift and on the other hand to quantify the uncertainties of the calculated streamflow. The combination of the three methods has been applied to the Ardèche at Meyras, France, which shows net shifts after each major flood. Good detection and estimation of the rating shift has been observed retrospectively and in real time. The method is generic and, after further validation, operationally applicable to other sites., Les séries temporelles de débit des rivières sont établies à l'aide de "courbes de tarage", qui sont des modèles avec les hauteurs d’eau en entrée et les débits en sortie. Malheureusement, de nombreuses stations hydrométriques ont une relation hauteur-débit instable, notamment à cause de l’évolution du lit de la rivière lors des crues. Ces "détarages" posent problème à la fois pour l'établissement des séries hydrologiques de long-terme (analyse rétrospective) et pour la fourniture de données en temps réel, par exemple pour la prévision des inondations, avec des incertitudes quantifiées et fiables. Les méthodes existantes pour la mise à jour de la courbe de tarage sont basées sur une analyse statistique des données de calibration (jaugeages) passées ou sur des règles empiriques. Cette thèse a permis de développer des méthodes originales pour la détection automatique des détarages et l'estimation de leur amplitude en rétrospectif et en temps réel : une méthode de segmentation des résidus entre les jaugeages et une courbe de référence, une analyse des récessions du limnigramme et une détection de détarages potentiels à partir d'un indicateur disponible en temps réel (par exemple, transport sédimentaire cumulé). L'approche probabiliste permet d'une part de prendre en compte l'incertitude des informations sur les détarages et d'autre part de quantifier les incertitudes sur les débits calculés. La combinaison des trois méthodes a été appliquée à la station de l’Ardèche à Meyras, en France, qui présente des détarages nets après chaque crue importante. Une bonne détection et estimation des détarages a été observée en rétrospectif et en temps réel. La méthode est générale et applicable en opérationnel à d’autres stations après validation. La méthode est générique et, après davantage de validation, applicable en opérationnel à d'autres sites.

International audience; River discharge time series are usually established using "rating curves" (hereafter called RCs) that approximate the stage-discharge relation. These models are fitted using occasional stage-discharge measurements, also known as gaugings. However, two major problems affect the RC estimation: firstly, gaugings and RCs are both uncertain; secondly, the stage-discharge relation can be unstable and subjected to sudden or transient changes (also known as "rating shifts"). For the first problem many authors have proposed methods for the uncertainty quantification. As regards the second issue transient changes require a dynamic approach while sudden changes require detecting the times of shift and estimating an RC for each stable period. We propose some tools for the detection of effective or potential sudden shift times in a retrospective analysis. To do this we use two main sources of information: the gaugings and the water level (or "stage") record. The proposed retrospective analysis is based on the segmentation of residuals between gaugings and a reference RC; the segmentation of the recessions of the stage record; a sediment transport proxy model. The proposed method has been applied to two case studies mainly affected by morphological changes yielding quite encouraging if compared to the official shift dates.; Les séries chronologiques de débit de la rivière sont généralement établies à l'aide de "courbes de tarage" (ci-après appelées CdT) qui approximent la relation hauteur-débit. Ces modèles sont calibrés à l'aide de mesures occasionnelles de hauteur-débit, les jaugeages. Cependant, deux problèmes majeurs affectent l'estimation des CdT: premièrement, les jaugeages et les CdT sont incertaines; deuxièmement, la relation hateur-débit peut être instable et soumise à des changements soudains ou transitoires ("détarages"). Pour le premier problème, de nombreux auteurs ont proposé des méthodes de quantification de l'incertitude. En ce qui concerne le deuxième problème, les détarages transitoires nécessitent une approche dynamique tandis que les détarages soudains nécessitent de détecter les temps de détarage et d'estimer une CdT pour chaque période stable. Nous proposons quelques outils pour la détection de temps de détarage soudain effectifs ou potentiels à travers une analyse rétrospective. Nous utilisons deux sources principales d'information: les jaugeages et la hauteur d'eau. L'analyse rétrospective proposée est basée sur: la segmentation des résidus entre les jaugeages et une CdT de référence; la segmentation des récessions du limnigramme; un modèle proxy de transport sédimentaire. La méthode proposée a été appliquée à deux cas d'études principalement affectées par des changements morphologiques, ce qui donne des résultats assez encourageants par rapport aux dates officielles de changement.

International audience; River discharge time series are usually established using "rating curves" (hereafter called RCs) that approximate the stage-discharge relation. These models are fitted using occasional stage-discharge measurements, also known as gaugings. However, two major problems affect the RC estimation: firstly, gaugings and RCs are both uncertain; secondly, the stage-discharge relation can be unstable and subjected to sudden or transient changes (also known as "rating shifts"). For the first problem many authors have proposed methods for the uncertainty quantification. As regards the second issue transient changes require a dynamic approach while sudden changes require detecting the times of shift and estimating an RC for each stable period. We propose some tools for the detection of effective or potential sudden shift times in a retrospective analysis. To do this we use two main sources of information: the gaugings and the water level (or "stage") record. The proposed retrospective analysis is based on the segmentation of residuals between gaugings and a reference RC; the segmentation of the recessions of the stage record; a sediment transport proxy model. The proposed method has been applied to two case studies mainly affected by morphological changes yielding quite encouraging if compared to the official shift dates.; Les séries chronologiques de débit de la rivière sont généralement établies à l'aide de "courbes de tarage" (ci-après appelées CdT) qui approximent la relation hauteur-débit. Ces modèles sont calibrés à l'aide de mesures occasionnelles de hauteur-débit, les jaugeages. Cependant, deux problèmes majeurs affectent l'estimation des CdT: premièrement, les jaugeages et les CdT sont incertaines; deuxièmement, la relation hateur-débit peut être instable et soumise à des changements soudains ou transitoires ("détarages"). Pour le premier problème, de nombreux auteurs ont proposé des méthodes de quantification de l'incertitude. En ce qui concerne le deuxième problème, les détarages transitoires nécessitent une approche dynamique tandis que les détarages soudains nécessitent de détecter les temps de détarage et d'estimer une CdT pour chaque période stable. Nous proposons quelques outils pour la détection de temps de détarage soudain effectifs ou potentiels à travers une analyse rétrospective. Nous utilisons deux sources principales d'information: les jaugeages et la hauteur d'eau. L'analyse rétrospective proposée est basée sur: la segmentation des résidus entre les jaugeages et une CdT de référence; la segmentation des récessions du limnigramme; un modèle proxy de transport sédimentaire. La méthode proposée a été appliquée à deux cas d'études principalement affectées par des changements morphologiques, ce qui donne des résultats assez encourageants par rapport aux dates officielles de changement.

International audience; The present study investigates the spatio-temporal variability of riverine sediment fluxes to the Gulf of Lions, one of the most extensive shelf regions in the Mediterranean Sea. Small coastal rivers compete here with the Rhone River, nowadays the largest Mediterranean river in terms of water discharge. Our scientific objectives were to investigate the major controls of riverine sediment yields (SY) in this area and to quantify the role of the small coastal rivers, largely ignored in previous studies, in the total sediment budgets. Another objective concerned the source identification of the Rhone sediments with regard to the major tributary contributions, and to test whether the sediment fluxes are in equilibrium in the basin. For the calculation of representative long-term fluxes, we used a Simplified Rating Curve Approach (SiRCA) which could be validated by high resolution monitoring and existing literature data. An overall of 13 drainage basins could be distinguished, covering about 86% of the study area (6 coastal rivers, the Rhone River, and 6 of its tributaries). Representative SY range from 19 to 151 t km−2 yr−1 in the investigated drainage basins. Despite their smaller basin areas and more torrential discharge regimes, SY of the coastal rivers were generally lower compared to SY of the Rhone River and its tributaries. Confrontation with the lithological, morphological and hydroclimatic basin characteristics indicate that lithology exerts the dominant control on SY in the study region. In particular, the existence of erodible sedimentary rocks in the headwater regions yields high SY. Peak values of 135 and 151 t km−2 yr−1 were observed for the Isere and Durance tributaries of the Rhone River, where badlands exist. The coastal rivers contribute on average only to slightly more than 5% of the long-term sediment inputs to the Gulf of Lions. During individual years however, their contribution can strongly increase (up to 27% in 2011). Their contribution is generally important during late spring and autumn, when flash-floods are frequent (up to 90% of the monthly sediment discharge). Summing the various tributary contributions of the Rhone River produces a sediment budget that is close to the sediment export at the river mouth. This indicates that at spatial scales, the average sediment fluxes are equilibrated in the basin. About 40% of the Rhone sediments originate from the Isere and Durance tributaries. However, fully closure of the budgets requires high SY in the lowermost basin parts. Omitting in our calculations only a few years with exceptional flooding considerably reduce these values. Trend analyses show at the same time that the sediment fluxes from the Saone and Isere tributaries decreased during the study period. At temporal scales, sediment transport from upstream to downstream was therefore not in equilibrium and the remobilisation of older sediments from the downstream basin parts may have been important.

Discharge time series in rivers and streams are usually based on simple stage-discharge relations calibrated using a set of direct stage-discharge measurements called gaugings. Bayesian inference recently emerged as a most promising framework to build such hydrometric rating curves accurately and to estimate the associated uncertainty. In addition to providing the rigorous statistical framework necessary to uncertainty analysis, the main advantage of the Bayesian analysis of rating curves arises from the quantitative assessment of (i) the hydraulic controls that govern the stage-discharge relation, and of (ii) the individual uncertainties of available gaugings, which often differ according to the discharge measurement procedure and the flow conditions. In this paper, we introduce the BaRatin method for the Bayesian analysis of stationary rating curves and we apply it to three typical cases of hydrometric stations with contrasted flow conditions and variable abundance of hydraulic knowledge and gauging data. The results exemplify that the thorough analysis of hydraulic controls and the quantification of gauging uncertainties are required to obtain reliable and physically sound results.