Objectives  The near-miss of two ships is a special situation that ships will encounter, and it can be modeled in terms of the time-dimension perspective for evaluating potential maritime traffic risks.

  Methods   In this paper, a temporal near-miss model is first established using the Fujii ship domain to detect near-miss situations, and the near-miss duration and average duration are then defined. An algorithm for calculating near-miss duration is proposed on the basis of the time synchronization processing of automatic identification system (AIS) data. Furthermore, the near-miss duration is used to analyze the time-dimension characteristics of maritime traffic risks. The Bohai Sea area is used as the research example to verify the model.

  Results  The correlations between the average near-miss duration and ship type and length are determined. According to ship type analysis, the longest average near-miss duration is 252.63 s for an oil tanker and at least 180 s for a passenger ship. According to ship size analysis, the longest average near-miss duration is 272.65 s for a large ship and at least 180 s for a mid-size ship.

  Conclusions  The proposed model can be used to accurately calculate near-miss duration, providing a new approach to studying maritime traffic risks from the time-dimension perspective.