Understanding housing flood resilience through Bayesian networks: A data-driven framework

Flooding poses a significant threat to housing facilities, particularly in flood-prone areas. To mitigate ongoing flood damage, it is crucial to enhance the understanding of underlying factors and the dynamic interactions among elements that contribute to such damage. This study introduces a data-driven Bayesian network model to enhance housing flood resilience, with a particular focus on the interactions between flood-related components. The resilience was focused on the key parameters of rapidity, redundancy, robustness, and resourcefulness, and these parameters were assessed to evaluate their impact on overall housing flood resilience. By considering the intricate interrelationships between floodwaters, terrain and housing, data variables relevant to flood resilience were first collected from three UK regions: Manchester, Cumbria, and York, across 18 zones categorised by high, medium, and low flood risk levels. Principal Component Analysis was employed to identify the most influential variables at each risk level, followed by Pearson correlation to determine their significant interrelationships and dependencies. In total, nine separate Bayesian network models were developed for each risk level using GeNIe software. The models quantified resilience levels, offering a flexible and robust tool for real-time flood resilience assessment. The results revealed distinct patterns of resilience across different risk levels, highlighting the importance of geographic and risk-specific strategies for enhancing flood resilience in housing. This study offers a transferable methodology and adaptable models, which is a significant contribution to the knowledge base for improving flood resilience globally, and it sets a new standard for data integration in resilience assessment.