Identifying Port Calls of Ships by Uncertain Reasoning with Trajectory Data

Lin Wu, Yongjun Xu, Fei Wang

TL;DR

This paper tackles the challenge of accurately identifying when ships arrive at and depart from ports using Automatic Identification System (AIS) data. They introduce a novel method that represents ports as sets of high-resolution grid cells (berths) and employs uncertain reasoning to detect mooring behavior, leading to more precise port call identification than previous methods.

What Problem Does This Paper Address?

Maritime transport is crucial for global trade, with ports serving as key nodes in this network. Accurately identifying ship arrivals and departures (port calls) is fundamental for analyzing maritime traffic, evaluating transportation networks, and assessing risks. While commercial datasets exist, they often lack transparency and broad coverage. AIS data offers a more comprehensive alternative, but identifying port calls from it is challenging due to inherent uncertainties in ship status reporting and the imprecise definition of port boundaries in existing datasets. Previous approaches often model ports as simple circles with fixed radii, which is insufficient for large or complex ports, leading to missed events or false positives. This paper aims to address this by developing a more robust and spatially precise method for port call detection using AIS trajectory data.

Methods

The authors propose a multi-stage approach:

  1. High-Quality Subset Extraction: They first filter raw AIS data to retain only messages with reliable kinematic information (longitude, latitude, SOG, heading, navigational status), discarding potentially erroneous or incomplete data. This step aims to reduce false positives in subsequent analyses.
  2. Berth Mooring Behavior Detection: This is the core of their novel approach. Instead of relying on simple speed thresholds or fixed port radii, they use fuzzy inference to detect when a ship is likely moored alongside a berth. This involves analyzing factors like the extent of movement, duration of the stop, speed, and heading variations. They define fuzzy rules and use S-membership functions to quantify the degree of berth mooring.
  3. Berth Location Extraction: Once mooring events are identified, their locations are aggregated to define berths. The Earth's surface is discretized into a fine grid (0.6s longitude by 0.6s latitude, approximately 20m x 20m), and each grid cell is assigned a membership value indicating its likelihood of being part of a berth, based on the detected mooring events within it.
  4. Mapping Berths to Ports: Identified berth grid cells are then mapped to actual ports. This involves a combined geospatial and semantic distance calculation, where semantic distance considers country, subdivision, and UN/LOCODE information to ensure accurate port-berth association.
  5. Port Call Detection: Finally, with a high-resolution port map (represented as sets of berth grid cells), port calls are detected by simply checking if a ship's position falls within any of these identified berth grid cells. This step is efficient as it avoids complex reasoning once the port map is established.

Datasets

  • AIS Dynamic Data: Approximately 14 billion AIS messages from July 2015 to February 2016, provided by the China Transport Telecommunications & Information Centre (CTTIC).
  • UN/LOCODE: Used for semantic matching of grid cells and ports. The dataset contained 76,000 locations after preprocessing.
  • World Port Index: Used as a reference for port locations.

No new benchmark datasets were introduced, but the method generates a high-resolution port area database.

Key Insights and Ideas

  • Berths as the True Port Boundary: The paper correctly identifies that the critical areas for ship activity (loading/unloading) are the berths, not the entire port area defined by arbitrary circles or rectangles. Representing ports as collections of berth grid cells offers much higher spatial resolution.
  • Uncertainty Reasoning for Mooring: Using fuzzy logic to detect "mooring behavior" is a smart way to handle the inherent noise and variability in AIS data (e.g., slight movements due to wind/waves, imprecise status updates). It moves beyond simple thresholds to a more nuanced understanding of ship activity.
  • High-Resolution Grid Representation: Discretizing the Earth into fine grid cells for berth identification is a powerful technique. It allows for a much more detailed and accurate representation of port infrastructure compared to previous methods.
  • Combining Geospatial and Semantic Data: The approach of using both geographical proximity and semantic information (like UN/LOCODE) for mapping berths to ports is crucial for disambiguating locations and ensuring accurate associations, especially across international borders.

Key Results

  • The method successfully identified 202,238 grid cells with a non-zero membership to berths.
  • The spatial resolution of the identified port maps is higher than 20 meters, significantly finer than previous methods that operated on kilometer-scale extents.
  • The paper demonstrates high true-positive rates and limited false-positive berth grid cells at sea, as shown in visual examples for Tianjin Port, Rotterdam, and Gibraltar.
  • The method is computationally efficient, with the matching of 202,238 grids and 3685 ports taking only 2.3 seconds after an initial 10-hour processing time for data extraction and berth mining.

Limitations and Future Work

  • Data Volume and Time Span: The study used 8 months of AIS data. The authors suggest mapping global ports from a larger dataset to improve coverage and accuracy.
  • Ship Size Consideration: The current method doesn't explicitly consider ship size, which could lead to fewer misdetections and a better understanding of berth utilization. Future work will incorporate ship static information (type, size).
  • Contextual Knowledge: The authors plan to combine AIS data with mapped ports and berth information to derive contextual knowledge, enabling anomaly detection and more sophisticated maritime network analysis.
  • Real-time vs. Historical: While the framework supports both, the primary implementation and results are based on historical data. Further validation on real-time streams would be beneficial.

Summary updated

Abstract

Considering that ports are key nodes of the maritime transport network, it is of great importance to identify ships’ arrivals and departures. Compared with partial proprietary data from a port authority or shipping company, approaches based on compulsory Automatic Identification System (AIS) data reported by ships can produce transparent datasets covering wider areas, which is necessary for researchers and policy makers. Detecting port calls based on trajectory data is a difficult problem due to the huge uncertainty inherent in information such as ships’ ambiguous statuses and ports’ irregular boundaries. However, we noticed that little attention has been paid to this fundamental problem of shipping network analysis, and considerable noise may have been introduced in previous work on maritime network assessment based on AIS data, which usually modeled each port as a circle with a fixed radius such as 1 or 2 km. In this paper, we propose a method for identifying port calls by uncertain reasoning with trajectory data, which represents each port with an arbitrary shape as a set of geographical grid cells belonging to berths inside this port. Based on this high-spatial-resolution representation, port calls were identified when a ship was in any of these cells. Our method was implemented with around 14 billion AIS messages worldwide over 8 months, and examples of the results are provided.
BibTeX
@article{wu2020identifying,
  author = {Lin Wu and Yongjun Xu and Fei Wang},
  title = {{Identifying Port Calls of Ships by Uncertain Reasoning with Trajectory Data}},
  journal = {ISPRS Int. J. Geo Inf.},
  year = {2020},
  doi = {10.3390/ijgi9120756},
}
Figure