Spatial mapping of encounter probability in congested waterways using AIS

Ships in congested waterways are more prone to collision than open sea conditions. Although in most waterways, the overall accident probability is known, their spatial distribution, even for the most dangerous waterways in the world, is not commonly available as a detailed map. Existing solutions distribute the ships along predetermined routes which distorts the actual spatial distribution. In this study, a collision model is developed based on molecular collision theory. The model allows the use of long-term MS data as input to calculate the encounter probability in two-dimensional waterways. The model is tested in the Strait of Istanbul, which is divided into sectors and further subdivided into cells. The encounter probability is calculated using vessel kinematics extracted from one-year AIS data. Results show that the collision probability increases as the narrow passages in the Strait intensify the ship density. Also, sharp turns in the SOI cause additional expansion in collision diameters which further increases the accident probability. The results are compatible with the observed collision accident locations. The sector-based approach provides a better understanding of the spatial distribution of accidents in high-resolution digital maps which are useful for captains, traffic controllers and other stakeholders to take necessary precautions.