Seismic spatiotemporal assessment of indoor occupant casualties in regional buildings: A Bayesian network approach incorporating population density dynamics

Abstract

In the context of rapid population growth and accelerated urbanization, the impact of earthquake disasters has intensified, making human casualties a critical concern. Accurate prediction of earthquake-induced casualties is essential for developing emergency response strategies, optimizing rescue resource allocation, and facilitating post-earthquake decision-making. This study employs a Bayesian Networks (BNs) model to systematically integrate information on both structural and non-structural components, as well as social factors, to construct a multifactorial assessment framework. This framework accurately estimates the probability of indoor occupants' life safety status during an earthquake. It considers the effect of population density, escape capability and rescue efficiency on individual safety and develops life safety status fragility surface with ground motion intensity measures and population density. Monte Carlo simulation is utilized to enable real-time prediction of indoor occupant life safety status within a single building structure. The study further extends to regional scale by incorporating the spatial correlation of ground motion intensity measures. This innovation facilitates a dynamic, spatiotemporal assessment of life safety status within a regional building portfolio. The findings highlight the critical role of ground motion intensity measures and population density in assessing indoor occupant casualties. By dynamically monitoring population distribution and real-time updating of ground motion intensity measures, the assessment framework proposed in this study can effectively enhance the accuracy and efficiency of earthquake disaster emergency response, providing a reliable scientific basis for government departments to facilitate rapid rescue operations and optimize resource allocation in the aftermath of an earthquake.