Representing people’s behavior during floods for simulation of human-floods feedbacks through agent based modelling

Representing people’s behavior during floods in agent based modelling is a challenging task for a realistic simulation of human-floods feedbacks. Previous research identified different long-term feedbacks that may lead to complex phenomena such as the so-called coping strategies, levee effects, call effects, adaptation effects, poverty traps, and status quo effect. In this work, we develop a methodology based on results from survey analysis to specify behavioral rules for capturing long-term feedbacks between humans and floods with agent based models. As a conceptual framework, we use the typology of flood behavior composed by the four categories: levee effect, learning effect, status quo, and good students effect, which depend on the frequency and magnitude of floods, as well as on the adaption and resilience of the people. The survey was conducted during 2024 in five regions of Chile, with 1007 respondents. A study case considering three localities along the Carampangue river, in the Central part of Chile, is presented. An agent based model of the study case is developed, considering the period 1970-2020. Results illustrate the capabilities of agent based models to capture human-floods feedbacks.