Refining Flood Evacuation ABM with local stakeholders in the Paris Area

With the rising level of the Seine River during hurricane Kirk, the city of Alfortville (Paris area, France) was facing a major concern. If the water level goes over 6,5m high, 97% of the city will be flooded in 2 days and the decision makers will have to manage the evacuation of 45.000 inhabitants. The massive evacuation of the population in case of a major flood in the Paris area remains a major challenge for emergency managers.

This presentation introduces the results of an agent-based model designed to simulate evacuation behaviors in response to different types of flooding across three territories in the Paris metropolitan area. The model, built using the NetLogo environment, is part of the Paris-Area Flood Evacuation (PAFE) project. We constructed a synthetic population using seven socio-demographic variables, calibrated to match census data and spatially distributed in a realistic way across households in each territory. Individual evacuation decisions were informed by a large-scale empirical survey (n = 5,000), with agents’ responses linked to their socio-economic profiles. Finally, the model was refined in collaboration with local experts and decision-makers who have direct experience with past flood events in the region.