Search

Anticipating floods is a major challenge for communities at risk of flooding as the entire warning system – responsible for the safety of people and goods - relies on this anticipation. There is an existing monitoring system “Vigicrues” for flood damage for a fifth of the river network in France. But for four-fifths of this network, made of small rivers, no monitoring is available. Yet those rivers are the most affected by flash floods which especially require anticipation for crisis management purposes. This is why at the beginning of 2017, the Vigicrues system for flood monitoring has been completed with a new flood warning system called Vigicrues Flash. This system provides automatic information in real-time on flood severity of ungauged basins for 10 000 French communities.Even if this new system is a real innovation for communities with no monitoring at all, the AIGA method which is used in Vigicrues-Flash has some limits. The first one is that the warnings are only based on the assessment of flood severity. But estimating flood severity is not enough to issue efficient flood warnings. To be able to do so, taking into account potential flood losses is essential. The main goal of this work is to enable an anticipated estimation of flood related damage, especially for ungauged basins. We offer a method to assess the risk of flood related damage based on flood severity assessed by the AIGA method and a territorial vulnerability assessment. This last one has been built on a bottom-up approach developed with crisis managers. Putting together this data has enabled a first assessment of the risk of flood risk damage as a dynamic risk index.By adjusting performance testing used in the meteorology field, we have been able to evaluate our risk index and to compare the results with the AIGA method. In order to do so, we have used existing damage data (CATNAT from the GASPAR database) as well as a specific multisource database (using notably social media data) which has been put together as part of this study (DamaGIS). The evaluation process has been tested for 12 communities in the Alpes-Maritimes, 69 in the Gard and 28 in the Var department. Two types of evaluation have been performed: a first comprehensive one continuously with CATNAT data on the 1988-2016 period; and another one per flood event at a finer scale.Our results show that moving from hazard assessment to risk assessment has significantly increased the relevance of the warnings and mostly at a smaller scale than the community one. Though, there is a better detection of flood related damage as the false alarm rate has been significantly reduced. This PhD work offers promising prospects to improve the current French warning system for floods and enable a more efficient emergency response.; Anticiper les inondations constitue un enjeu majeur pour les communes exposées aux crues car c’est sur cette anticipation que repose l’ensemble de la chaîne d’alerte, garante de la sécurité des personnes et des biens. Si un système de suivi du risque de dommages liés aux crues est disponible pour un cinquième du réseau hydrographique français, les petits cours d’eau composant les quatre cinquièmes restants ne font pas partie du dispositif de suivi temps réel du ministère en charge de l’écologie, appelé « Vigicrues ». Or il s’agit également des cours d’eau les plus concernés par le phénomène de crues rapides, pour lesquelles l’anticipation joue un rôle crucial en gestion de crise. Voilà pourquoi début 2017, Vigicrues a été complété par un service automatique d'avertissement des crues appelé Vigicrues Flash. Ce système permet de fournir en temps réel une information sur l’intensité de la crue des cours d’eau pour 10 000 communes françaises.Même si ce nouveau service constitue un réel progrès pour les communes jusqu’alors dépourvues de système d’anticipation, la méthode AIGA qui constitue le cœur de Vigicrues Flash possède certaines limites. L’une d’entre elles, est le fait que la méthode n’avertit que sur le niveau de rareté de la crue, sans tenir compte des enjeux présents. Or, pour générer un avertissement efficace, il est nécessaire de prendre en compte les conséquences potentielles de cette crue. Cette thèse a donc pour but de permettre l’estimation anticipée des dommages liés aux crues rapides, en particulier sur les bassins non jaugés. Pour cela, nous proposons une méthode d’estimation du risque de dommages fondée d’une part sur la qualification de l’intensité de l’aléa crue par la méthode AIGA et d’autre part sur la prise en compte de la vulnérabilité du territoire. Cette dernière a été construite à partir d’une approche bottom-up innovante, directement auprès de gestionnaires du risque. Le croisement de ces deux types d’informations a permis de fournir une première caractérisation du risque de dommages liés aux inondations sous la forme d’un indice de risque dynamique.En adaptant des tests de performances issus de la météorologie, nous avons pu évaluer notre indice par rapport à la méthode AIGA seule. Des informations sur les dommages déjà existantes (les arrêtés « CATNAT » issus de la BD GASPAR) ou spécifiquement collectées (la BD DamaGIS constituée pour cette thèse à partir d’informations présentes notamment sur les réseaux sociaux) constituent nos données de validation. Notre évaluation a porté sur 12 communes dans les Alpes-Maritimes, 69 dans le Gard et 28 dans le Var, et s’est faite de deux manières complémentaires : d’une part une évaluation en continue et exhaustive à partir des arrêtes CATNAT pris pour nos communes sur toute la période 1998-2016 ; et d’autre part une évaluation événementielle, mais à l’échelle infra-communale.Nos résultats montrent que le passage de la caractérisation de l’aléa à celle du risque améliore nettement la pertinence des avertissements émis, surtout à l’échelle infra-communale. Les dommages y sont mieux détectés, avec un taux moindre de fausse alertes. Cette thèse ouvre donc de réelles perspectives d’amélioration de la chaîne de l’alerte actuelle, permettant de mieux organiser la réponse des services de secours et de gestion de crise face à l’annonce de dommages potentiels liés aux crues rapides.

Introduction The consequences of flash floods can be dramatic in terms of casualties or economic losses. Jonkman (2005), in a global assessment of flood-related casualties, showed that flash floods lead to the highest mortality (number of fatalities divided by the number of affected people). For example, in the recent flash flood that occurred in the French Riviera around Cannes on 3 October 2015, 20 casualties and 650 billion euros of insured damage (source: http://www.ccr.fr/) were reported...

This book has been published by Allenvi (French National Alliance for Environmental Research) to coincide with the 22nd Conference of Parties to the United Nations Framework Convention on Climate Change (COP22) in Marrakesh. It is the outcome of work by academic researchers on both sides of the Mediterranean and provides a remarkable scientific review of the mechanisms of climate change and its impacts on the environment, the economy, health and Mediterranean societies. It will also be valuable in developing responses that draw on “scientific evidence” to address the issues of adaptation, resource conservation, solutions and risk prevention. Reflecting the full complexity of the Mediterranean environment, the book is a major scientific contribution to the climate issue, where various scientific considerations converge to break down the boundaries between disciplines.

The consequences of flash floods can be dramatic in terms of casualties or economic losses. Jonkman (2005), in a global assessment of flood-related casualties, showed that flash floods lead to the highest mortality (number of fatalities divided by the number of affected people). For example, in the recent flash flood that occurred in the French Riviera around Cannes on 3 October 2015, 20 casualties and 650 billion euros of insured damage (source: http://www.ccr.fr/) were reported. Flash flood forecasting systems are critically needed to better organize crisis management and rescue operations.As mentioned in chapter 1.3.4 (Gaume et al. 2016), flash floods are characterized by a rapid increase of river water levels. They often affect small watersheds, generally ungauged. The spatial and temporal variability of rainfall, landscape characteristics and pre-event catchment wetness are important influential factors in flash flood generation, contributing to the large space-time variability of hydrological responses (Borga et al. 2011).Forecasting flash floods is therefore a complex task. It necessitates the monitoring of large areas, where each small watershed of a few square kilometres can possibly be affected. Real-time observation networks and models must run at small temporal and spatial scales, on the order of a few minutes and kilometres. Furthermore, discharge time series are not available for the majority of the possibly affected watersheds, posing a real challenge for model calibration and evaluation. In this context, radar based precipitation products and/or meteorological forecasts with a high resolution (typically 1 km2 grid size) are crucial (Creutin and Borga, 2003). Slight misplacements of the precipitation may for instance lead towarnings attributed to the wrong river network and to inappropriate flood management decisions.