Analysis of marine oil spill accidents using a human-centered approach

Global shipping, responsible for 90% of trade, is a primary carbon emitter. Oil spills from accidents deteriorate environmental damage and hinder shipping decarbonization. This study analyzes maritime oil spill risk factors and their interdependencies, aiming to develop effective preventive strategies. First, common risk factors in oil spills were identified via comprehensive literature reviews and global accident reports. A data-driven Bayesian Network is then constructed, incorporating human factors into maritime pollution accident analysis using Tree Augmented Network (TAN) and LASSO algorithms. Next, sensitivity analysis and model validation are conducted to explore critical risk influencing factors affecting pollution levels in ship accidents. The results indicate that the number of seafarers onboard significantly influences low and moderate pollution accidents, while ship damage is the primary cause leading to severe pollution. The findings explain ship oil spill risks from a human-centered perspective, enabling maritime stakeholders to develop management strategies to reduce oil spill pollution.