Risk assessment and regionalization of surface ship navigation in the Northwest Pacific Ocean based on Bayesian networks

Abstract To evaluate the navigational risks of surface vessels and facilitate decision-making for risk mitigation, the paper presents a risk assessment model for ship navigation that utilizes Bayesian network method, based on ship parameters and meteorological ocean reanalysis data. The model incorporates both risk and vulnerability indexes and employs fuzzy comprehensive evaluation method to quantify risk levels. Consequently, the constructed Bayesian network model can rapidly calculate the ultimate risk levels under varying risk and vulnerability conditions. A portion of the Northwest Pacific Ocean (25°–35°N, 130°–140°E) is selected for a case study using the Type 051C destroyer to conduct risk assessment and regionalization, adhering to the principles previously mentioned. The assessment reveals that the risk in this sea area varies significantly with seasons, influenced by factors such as the Kuroshio Current. The risk level is highest during spring (March to May) and autumn (September to November), followed by summer (June to August), with the lowest risk occurring during winter (December to February). Additionally, there are significant geographical differences in risk distribution. The medium to high-risk areas during spring are mainly located in the southern region where the Kuroshio Current flows, while the risk areas during autumn are mainly concentrated in the central and northern regions, which are susceptible to typhoon influence. The research findings indicate that a Bayesian network-based model can effectively assess and delineate risks for naval vessels’ maritime operations, with results having certain implications for planning and decision-making support in maritime military activities.