Hierarchical Shipping Network Generation and Crucial Hub Identification Using Two-Tier Clustering on AIS Trajectories

The management of massive Automatic Identification System (AIS) data is challenged by its redundancy and complex structure, obscuring the underlying topology of maritime networks. This paper proposes a multi-stage methodology for hierarchical network generation and crucial hub identification. The process begins with the Ramer-Douglas-Peucker (RDP) algorithm for individual trajectory simplification, efficiently preserving core navigation features. Crucially, a two-level clustering approach is introduced for macroscopic network abstraction. First, the DBSCAN algorithm identifies numerous localized hotspots where vessels converge or anchor. To reveal the functional hierarchy, the centers of these hotspots are then fed into K-Means clustering, which aggregates them into K distinct Macro-Regional Clusters. These macro-clusters define the network’s high-level, simplified structure. Finally, the generated hierarchical network is used to assess the functional flow and simplify complex trajectories. The methodology effectively transforms high-volume trajectory data into an insightful, manageable, and hierarchically structured network model, offering valuable contributions to maritime traffic management and port planning.