A comprehensive understanding of traffic properties is necessary in evaluating and controlling the ship collision risks. AIS (Automatic Identification System) provides a quick and effective method to obtain ship behavior in a wide area, which facilitates gathering of ship traffic data not only in inland seas or restricted congested sea areas, but also in wide coastal sea areas. In this paper, the authors propose a method to utilize AIS data to develop a ship traffic model that represents traffic characteristics, especially focused on lateral direction, ship speed, and course. Nojimazaki area in Japan is taken as a case study. By introducing a definition of virtual gate lines located in the targeted area, and taking intersection points of trajectory with the gate lines, information on ship status at the points are extracted. Statistical analysis shows that speed and course distributions are bilaterally symmetric and distributed in narrow range, while lateral distribution is asymmetric and has a great concentration around the mode. To suit these characteristics, Gauss and gamma distributions are selected. Obtained model reproduced the observation result with reasonable accuracy.