Estimating the economic costs of flooding using physical exposure approaches.

Summary The work of this thesis is situated in the framework of the extension of the perimeter of the impacts estimated by the flood model of the Caisse Centrale de Réassurance (CCR). The first chapter allowed us to understand the different impacts of a flood, their interdependencies and to conclude a classification of the impacts. From this, we propose a global model built by chaining the modeling of impacts where the outputs of the different models serve as inputs for the following models. This approach will allow a breakdown of the impacts by item and an understanding of the cascading effects from direct damage to macro-economic effects. Chapter 2 focuses on the modelling of vehicle damage under the CCR specifications. The different requirements are: modeling for the flood hazard alone and for the automobile hazard alone, modeling for a particular event and for the total annual load. The study of this risk, allowed us to describe, implement, identify the problems and possible improvements of different modeling approaches, namely: - a linear regression, the methodology currently used by CCR, - a frequency x severity modeling associated with the extreme value theory, very much used in the insurance world, - a physical modeling on exposure, an approach already adopted at CCR for building damages. We therefore used the CCR flood hazard model and developed the vulnerability and damage modules specific to this risk. As CCR is in charge of the accounting and financial management of the Fonds National de Gestion des Risques en Agriculture on behalf of the State, the work continued in Chapter 3 with the modeling of agricultural risk. From a state of the art on the modeling of the agricultural consequences of floods, a modeling of this risk is proposed and implemented. The model is integrated in the physical modeling approach on exposure developed at JRC. The work focused on the development of vulnerability and damage modules specific to agriculture. The vulnerability model was developed from the graphical parcel register and the damage module was developed from the damage curves for agriculture developed by the national working group Analyse Multi-Critères inondation through the work of IRSTEA. The hazard model used is the model developed by CCR since 2003. The last chapter is thought as a technical base for CCR in order to continue the work of widening the scope of modeled impacts. First of all, a review of the work done (automotive and agricultural risks) and in progress (farm losses due to direct damage) is carried out and placed in the context of the global impact modeling proposed to CCR. For the impacts that have not been addressed, we will present the modeling issues, a short state of the art and the conclusions that we have drawn in terms of modeling.