Flooding is of particular concern in low-lying coastal zones that are prone to impacts from multiple flooding drivers, such as coastal (storm surge and waves), fluvial (excessive river discharge), and pluvial (excessive surface runoff). Failure to account for dependence (and its changes over time) between flood drivers, when dependence exists between them, may lead to underestimation of flood risk and under-design of flood defense measures. Characterizing the dependence between compound flooding drivers in space and across seasons (tropical and extra-tropical), and how this dependence changed over time is essential in this context. In this dissertation, compound flooding potential from all relevant flooding drivers is assessed at 35 locations along the contiguous United States (CONUS) coastline. Different dependence measures are derived and analyzed using observations and state-of-the-art re-analysis data sets. In addition, temporal changes in the extremal dependence are assessed, using a sliding time window approach and possible associations with large-scale climate indices are explored. The effects of changes in dependence and marginal distributions over time between coastal and fluvial flooding drivers are investigated in more detail for a selected case study location. To overcome the computational expense of numerical modeling for flood mapping of large sets of events, a framework is introduced based on hybrid statistical modeling and one-dimensional hydraulic modeling combined with a flood inundation tool capable of propagating spatially variable along-river water surface elevations inland. The framework to delineate the flood transition zone is implemented for the Potomac River and different flood scenarios are analyzed to assess how different combinations of coastal water levels and river discharge modulate the flood hazard (specifically flood depth and extent). Overall, the findings provide new insights into characterizing compound flooding potential, its changes in space and time and how incorporating flood driver dependencies affects flood hazard.

L’objectif de la présente thèse est de proposer un cadre conceptuel d’approche desmouvements de population liés aux risques environnementaux. Depuis une vingtained’années, un ensemble de termes a émergé afin de caractériser ces populations: réfugiésenvironnementaux, éco-réfugiés, réfugiés climatiques, éco- migrant pour ne citer que les plusconnus. La caractérisation et le statut accordé à ces populations sont dépendants de la nature,la force et la rapidité de l’évènement environnemental contribuant à leur déplacement. Il estdonc indispensable de définir des critères précis pour éviter tout risque de confusion etprocéder à de véritables évaluations des flux liés à ces catastrophes. Notre analyse sefocalisera sur les risques liés à la hausse du niveau de la mer: risques de submersion etd’inondation sur le littoral et dans les espaces deltaïques. Face à ces risques croissants, il estimpérieux de mettre en place une «stratégie proactive», basée sur le triptyque protection -prévention et anticipation. Cette stratégie exige au-delà de la délimitation des zonesgéographiques concernées, une quantification et qualification des populations potentiellementtouchées. C’est la conjonction de deux formes de vulnérabilité qui est au coeur de notreproblématique: la vulnérabilité face aux risques d’inondations et la vulnérabilité face à lapression démographique croissante. C’est dans cet esprit que nous étudierons deux zonessoumises au risque d’inondation, l’une en France (Estuaire de la Gironde) et l’autre en Grèce(Deltas d’Axios-Loudias-Aliakmonas en Macédoine Centrale) afin d’estimer l’importanceéventuelle des mouvements de population à l’horizon 2025 et 2050. / The present thesis seeks to define a conceptual framework in order to examine the populationmovements related to environmental risks. For about twenty years, several terms haveemerged to describe thedisplaced people such as: environmental refugees, eco-refugees,climate refugees or eco-migrants. The terms and the status attributed to these populations arelargely dependent on the nature, strength and speed of the environmental event contributing totheir displacement. It is therefore necessary to specify the criteria in order to avoid anyconfusion and to be able to assess the physical flows generated by the environmentaldisasters. Our analysis is focused on the risks of submersion and inundation along the coastand in the delta areas associated with rise in the sea levels. In view of these increasing risks, it is absolutely necessary to set up a "proactive strategy", based on the triptych protection -prevention - anticipation. For each type of risk considered, this strategy requires, beyond thedelimitation of the geographical areas concerned, quantification and qualification of thepotentially affected populations. Consequently, the main aspect of our problematic combinestwo types of vulnerability:the firstdue to environmental disasters’ risks such as floods and thesecondto increasing population pressure. Finally, we chose to examine in our study two areashighly subjected to flood risks such as the Gironde estuaryin France and the Axios -Loudias-Aliakmonas Deltas in Central Macedoniain Greece trying to estimate the potential size ofpopulation movements in 2025 and 2050.