From Flood Depth to Flood Probability: Improving Urban Evacuation Planning for Vulnerable Populations under Climate Change

Urban areas are increasingly exposed to flood risk under climate change, where evacuation planning based on fixed inundation depth thresholds may inadequately capture spatial uncertainty and the needs of vulnerable populations. This study investigates how shifting from a depth-based to a probability-based flood risk perspective can improve urban evacuation and shelter planning. Using a dense urban district in New Taipei City, Taiwan, as a case study, we integrate urban drainage modelling with multi–return-period rainfall analysis to construct probabilistic flood distributions under an end-of-century RCP8.5 climate change scenario, and examine their implications for evacuation decision-making and urban risk governance.

Probabilistic flood maps are developed by overlaying inundation extents simulated for multiple rainfall return periods, allowing flood risk to be expressed in terms of likelihood rather than as a single deterministic outcome. Changes in flood hazard patterns under climate change are further assessed using composite hazard indicators, including flood depth, flow velocity, and water level rise rate. To evaluate decision-level impacts, probabilistic flood information is incorporated into urban road network analysis to compare evacuation strategies based on a conventional depth threshold (50 cm inundation) and a probability-based decision threshold (70% flood likelihood).

Results indicate that under climate change conditions, areas characterized by high flood probability and moderate hazard levels expand significantly within the urban fabric, affecting neighborhoods not readily identified by depth-based criteria alone. The comparison of evacuation strategies reveals substantial differences in priority evacuation zones, routing options, and shelter allocation for vulnerable populations. Probability-based planning reduces the risk of over-evacuation while enabling earlier and more targeted evacuation actions in high-likelihood risk areas.

The findings demonstrate that integrating probabilistic flood risk into urban evacuation planning can enhance anticipatory decision-making and support more adaptive and equitable urban risk governance. By reframing flood risk from static depth thresholds to probabilistic decision logic, this approach contributes to strengthening urban resilience and improving disaster preparedness for vulnerable populations under climate change.