A method for detecting possible near miss ship collisions from AIS data

• We propose a method detecting possible near miss ship–ship collisions from AIS data.
• The model we propose can be used to gain further insight in safety of maritime transportation.
• The method considers factors: distance, relative speed, and phase defined by course.
• The model behaves as expected as maritime traffic over the Northern Baltic Sea.
• The risk model can be used to rank ship–ship encounters of various risk levels.

Various methods have been proposed to analyse safety and risk of maritime transportation, through modelling techniques and empirical studies, focusing on ship–ship collision. However, an efficient method for screening traffic data for the near miss collisions is lacking. Therefore the contribution of this paper is two-fold. First, it proposes a novel method detecting possible near miss ship–ship collisions from AIS data. Second, it discusses how such near miss data can be used to gain further insight in safety of maritime transportation. The developed Vessel Conflict Ranking Operator (VCRO) considers the following factors affecting the complexity of an encounter between two ships: the distance between the two ships, the relative speed of the ships, and the difference between the headings of the ships. The model content and structure are based on expert judgment, whereas the parameterisation is based on AIS data points of a reference encounter situation dataset. The developed model has been applied to the traffic data from the Northern Baltic Sea. The obtained results and the model validity tests suggest the model being adequate for ranking encounters, and for prioritising the encounters to subject to further expert scrutiny.