Data-driven urban planning for proactive crowd management: Lessons from the 2022 Seoul Halloween crowd crush

Abstract

Rapid urbanization has heightened the risk of crowd-related disasters during large-scale events in complex urban environments. This study revisits the 2022 Seoul Halloween crowd crush, which resulted in 159 fatalities, to examine how temporal surges and spatial bottlenecks interacted to produce fatal congestion. We aim to generate actionable insights for data-driven urban planning that proactively mitigates crowd risks in mass gathering contexts. Using de facto population data, we identify a sudden surge in crowd density near the incident site, revealing critical temporal vulnerabilities. In parallel, space syntax analysis uncovers spatial conditions characterized by high integration and low connectivity, features that constrained internal circulation and contributed to pedestrian entrapment. These findings indicate that the disaster stemmed not from temporal or spatial factors alone, but from their dangerous convergence within a structurally constrained environment. Furthermore, cluster analysis identifies other areas with similar spatial vulnerabilities, offering a tool for preventive screening. This study highlights the importance of integrating temporal crowd dynamics and spatial constraints into urban planning frameworks to enable proactive crowd safety strategies and the design of safer urban environments.