Generation of Convergent Ship Traffic Dataset using Sentinel-1 Imagery and AIS

Abstract

Satellite-based ship detection has opened a new era of marine surveillance in recent decades. The ship detection using Sentinel-1, a Synthetic Aperture Radar (SAR) satellite image product, is an excellent measure of monitoring illegal activities at sea. Despite that, it is essential to compare the detection data to other convincible resources to verify whether the detected ships are indeed and to obtain further information. This paper introduces the procedure of matching the information of ships that are detected from Sentinel-1 imagery and Automatic Identification System (AIS) data. Since there are factors including the satellite's trajectory and scanning duration, appropriate data extraction from metadata of Sentinel-1 is required in advance. Azimuth shift, a symptom where a ship position moves to forward or backward of the SAR satellite proceeds in the imagery, also must be compensated. Consequently, a new dataset combining the information from Sentinel-1 and AIS is composed.