An agent-based model to simulate inhabitants' behavior during a flood event

Flood management is particularly important for many urban territories. In order to assess the interest of different risk mitigation strategies, it is necessar y to consider the human behavior. Different models and approaches are available to model and simulate people behaviors during a flood event. Despite the considerable recent progress of these models, none of them succeeds in answering a l l the required challenges: (1) simulate the flood event, (2) integrate geographical information, (3) take into account complex behaviors for inhabitants considering emotion and partial knowledge, (4) enable different strategies for inhabitants, (5) take into account the degradation of infrastructures and its impact on their functioning, and (6) ensure the genericity and flexibility of the model in order to be able to apply it to any territory. In order to meet these challenges, we provide a new Agent based Model, called SiFlo. In this paper, we present this model and an application to La Ciotat (France).