A Dynamic Bayesian Network Probability Framework for Typhoon Risk Analysis in the Land-sea Regions

https://doi.org/10.36777/jag2023.2.2.5

Abstract

The life cycle of a typhoon can generally be divided into three stages: offshore, nearshore or coastal, and landfall. Each stage involves distinct hazards, exposed elements, and potential consequences. During the offshore stage, typhoons primarily threaten maritime activities, particularly navigation and shipping, through strong winds, high waves, heavy precipitation, and rapidly changing sea conditions. As typhoons approach the coast, these hazards increasingly affect coastal infrastructure and human activities, while landfall can cause severe damage to coastal structures, transportation networks, settlements, and other exposed assets. Conventional risk assessment methods often focus on individual hazards or specific stages, limiting their ability to capture the dynamic and interconnected risks throughout the complete typhoon life cycle. To address this limitation, this study develops a Bayesian network-based risk assessment model to evaluate typhoon-related risks across different life-cycle stages. The framework considers both navigational risks to ships and damage risks to coastal structures, integrating marine and terrestrial hazards within a unified assessment framework. A typhoon disaster memory database is also constructed to systematically organise historical disaster information, deconstruct major risk factors, and establish unified descriptive semantics. By linking typhoon characteristics, hazard factors, exposed elements, and potential consequences, the proposed framework provides a comprehensive representation of evolving typhoon risks. The results can support maritime safety, coastal infrastructure protection, disaster prevention, and emergency response planning.