Using Automatic Identification Systems for Maritime Activities

This information was sourced from "Artificial Intelligence in Action: Selected ADB Initiatives in Asia and the Pacific," Asian Development Bank, 2024, https://www.adb.org/sites/default/files/publication/963831/artificial-intelligence-action-asia-pacific.pdf. Licensed under Creative Commons Attribution 3.0 IGO (CC BY 3.0 IGO).

The Automatic Identification System (AIS) is an automated tracking system used at sea that provides information on ships within a given time. While it was originally designed to help ships navigate and avoid collisions, its data are now increasingly being used for research, port performance analysis, and estimates of trade flows and maritime carbon trade emissions.

ADB explored whether AIS data, which are available through the United Nations Global Platform, could be used as an alternative source of economic statistics, considering that it is available in near real time (data are updated every 4 hours), while official statistics take months or years before they are released.

The key indicators that were identified were not explicitly identified in the raw AIS data. Thus, ADB proposed a framework and methods to extract the information from the data. ADB’s proposed framework covered events of interest (EOIs), which are specific maritime incidents or activities pertinent to a target indicator, and areas of interest (AOIs), which are geographic locations where these events occur identified through manual, distance-based, or cluster-based approaches. The framework was operationalized by creating indicators for ports and passageways that represent major hubs of maritime activity and with cases of maritime disruptions:

1. Port activity. Three indicators were identified under port activity: the count of unique vessels in a port, the number of arrivals in a port, and the median time spent by vessels in a port.

   The EOIs were the port calls (entry, length of stay, and exit of vessels to and from the port) and the AOIs cover the berths, terminals, and anchorages of each port. Squares that were formed by tracing 22 km from the center of each port for every side served as the boundaries to mark the distance-based AOIs. These were supplemented with cluster-based AOIs formed using Hexagonal Hierarchical Geospatial Indexing System (H3), and Density-Based Spatial Clustering of Applications with Noise (DBSCAN). The H3 indices were used to represent the location points while the DBSCAN algorithm identified the group of H3 indices forming the port boundary.

2. Traffic along maritime highways. Three indicators were identified: the number of unique vessels, the count of transits, and the time spent by the vessels in these passageways.

   The EOIs were the vessels’ entry and exit from these passageways. The AOIs, meanwhile, were the mouths of the passageways identified using DBSCAN. Manual AOIs generated by selecting the narrowest areas along the passageways were used as alternative AOIs when it was not feasible to use the mouths.

To study the major hubs of maritime activity, the largest ports identified by the World Shipping Council in 2019 and those with the highest connectivity to different parts of the world were selected. These included the ports of Los Angeles and Long Beach in the United States, the Port of Rotterdam in the Netherlands, and the Port of Shanghai in the People’s Republic of China.

Traffic in the passageways was supplemented to port activities to create a comprehensive view of the maritime industry. Passageways included were the Malacca and Singapore Straits, the Suez Canal, the Strait of Gibraltar, and the Panama Canal. The straits of Hormuz and Bab-el Mandeb were also included considering the role they play in the global oil trade. Ports and passageways that faced significant disruptions in 2022 (the Russian invasion of Ukraine, the Sri Lankan economic crisis, and the Tonga volcanic eruption) were also studied to see whether AIS data significantly changed because of these events.

The ports and passageways that were included in the research because of this consideration were the Port of Odesa, the Port of Colombo, the Port of Nuku’alofa, the Dardanelles Strait, and the Bosporus Strait.

The indicators were able to capture the effects of the lockdowns due to the COVID-19 pandemic, which saw reduced maritime activity. It also reflected the uptick in activity by February 2022. Similarly, it also revealed the impact of the Russian invasion of Ukraine on trade activities. Meanwhile, the blockage of the Suez Canal in March 2021 was also captured by the indicators. On the other hand, the study also revealed some limitations in using AIS data, including signal gaps in cases when transponders are turned off.

Some of the indicators were validated by comparing them with other data sources. While there were some discrepancies, the estimates were nonetheless coherent with official data, with absolute percentage errors ranging from 3%–4%, which lends credence to the value of using AI data to generate timely and accurate statistics.

The results of the study showed that AIS data could be used as an alternative data source to generate timely information on maritime activities. It is important to note that the indicators were developed not to replace official statistics but rather to complement them by providing near-real-time insights to help policymakers make data-driven decisions.