Author(s): Thomas Cochrane; Lea Dasallas; Barry Evans; Markus Pahlow

Linked Author(s): Lea Dasallas, Tom Cochrane

Keywords: Urban flood modelling; Transport risk analysis; Depth–velocity stability; Emergency planning; Surrogate modelling

Abstract: Pluvial urban flooding due to large rainfall events poses severe risks to transport networks and public safety. This study presents an integrated framework using a machine-learning surrogate flood model with a traffic simulation model to simulate urban flood dynamics and assess transport accessibility under hazardous conditions. The surrogate model reduces computation time from approximately 20 hours for high-resolution physics-based simulations to minutes (1–34 minutes depending on grid resolution), while retaining excellent agreement with the original 1D/2D flood model outputs. Furthermore, the surrogate flood model provides both flood depth and velocity outputs, which are combined with stability functions for pedestrians and vehicles to identify high-risk zones. These risk maps are overlaid on the transport network to evaluate accessibility to critical services during floods. A case study in the central business district of Wellington, New Zealand, demonstrates that velocity-inclusive risk mapping significantly increases high-risk areas compared to depthonly assessments, with pedestrian risk zones increasing by over 80% and vehicle high-risk areas by approximately 36%. Transport network analysis reveals substantial reductions in accessibility to hospitals and public transport hubs during peak flooding. The proposed framework supports proactive emergency planning and resilience strategies by integrating rapid flood modelling with transport risk analysis. This approach offers a robust tool for cities facing climate-driven flood challenges. It supports both scenario-based resilience planning and near real-time decision support by producing suggested road avoidance lists and safer alternative routes from forecast rainfall and tide conditions.

DOI: https://doi.org/10.64697/iahr.proc.hic2026.126

Year: 2026