Large-scale real-time evacuation modeling during urban floods: a coupled agent-based multi-model framework

The escalating frequency of urban pluvial flooding disasters, attributed to climate change and rapid urbanization, has emerged as a formidable global challenge. Meanwhile, the task of large-scale real-time urban evacuation simulation research remains arduous and complex. In response, this paper presents a coupled agent-based multi-model framework (CABMF) that intertwines (1) an agent-based model capturing the decision-making attributes, (2) a transportation model delineating pedestrian and vehicular evacuation movements, and (3) a hydrodynamic model simulating catastrophic floodwater progression. Notably, this framework achieves pioneering real-time evacuation simulation concurrently with the impacts of flood disasters. Applying the CABMF to the evacuation simulation in a flood emergency of the 10-kilometer buffer zone along the Yangtze River and Han River within Wuhan city, along with scenario-based simulations, enables an examination of the influence of transportation modes, shelter arrangements, and flood-affected road networks on evacuation experimental results. Research findings suggest that optimizing the ratio of transportation modes, arranging vertical evacuation shelters, and restricting access to identified high-risk road links contribute to improved evacuation efficiency and reduced disaster risk. The proposed CABMF couples the impacts of flood uniquely and is anticipated to yield substantial insights for large-scale evacuation management in the context of flood disasters.