Vessel Trajectory Reconstruction to Establish a Reference Trajectory from AIS Data by HDBSCAN and Interpolations

Abstract

With the widespread availability of automatic identification system (AIS) data, they have been applied in various maritime research. Due to technical issues and environmental interference, AIS data can have missing data points or irregularly reported intervals, which can cause unreliability in research activities and inaccurate trajectory visualization. In this study, AIS data quality issues and the importance of marine traffic in congested areas are addressed first. In this study, the Yangon River and approaches is selected as the area of study due to its significance in the country’s maritime trade. In this study, the reference trajectory reconstruction is done in two steps. The first step is using Hierarchical Density-Based Spatial Clustering of Applications with Noise (HDBSCAN) to cluster AIS data and identify the locations of major vessel traffic activity and extract their dynamic features. The second step is using interpolation methods to reconstruct the reference trajectory from the detected clusters. The reconstructed trajectory is validated against the surveyed benchmark routes for position and their dynamic AIS features by numerical methods. The evaluation result shows that the trajectory reconstructed in this way has a high similarity to the benchmark routes. We hope that our study would provide more future research opportunities for marine traffic in the Yangon Port.