Assessing vessel pollution risk in Asian areas: A comparative analysis based on data-driven Bayesian Network approach

Abstract

Vessel emission is gradually becoming one of the major sources of environmental pollution in Asian areas such as the Greater Bay Area (GBA) and Southeast Asia (SEA). Accurate identification of vessels with high pollution risks can effectively control their emissions. This research develops data-driven Bayesian network models to assess vessel pollution risk in GBA and SEA regions through a novel machine-learning methodology. A comprehensive analysis based on the newly proposed ‘pollution risk index’ reveals the key variables affecting vessel pollution risk, as well as similarities and differences between two regions. Furthermore, managerial implications are provided to help different coastal authorities better control the vessel pollution, i.e., the pre-assessment of vessel risk before onboard inspections, the formulation of specific regulations targeting on vessels with high pollution risks. This research provides a good reference for assessing vessel pollution risks, controlling vessel emissions and ensuring environmentally-friendly navigational waters in GBA and SEA areas.