Maritime traffic monitoring systems have recently increased their ability to capture and monitor ship movements everywhere in the world thanks to satellites coverage. These data sources generate a huge flow of position reports which can then be mined in order to derive knowledge about maritime traffic in the Arctic. These information systems also store useful details about ships such as theirs size and category. Combined with other data sources (bathymetric maps, sea ice coverage), an analysis of the Arctic maritime traffic can be done at different space and time scales. As an example, a density analysis of vessels carrying dangerous cargo (classified from A to D) is useful to conduct further research about risks related to maritime traffic in the Arctic. However, for now, the satellite coverage of Arctic maritime traffic is not continuous. It raises numerous problems regarding data fusion algorithms, trajectory modelling and risk analysis.
The application of anomaly detection in the maritime domain is nothing new, but the efficacy of conventional approaches in Arctic waters has yet to be proven. Most approaches to anomaly detection rely on having a basic understanding of what is normal behaviour and what is abnormal. Given that the Artic presents a unique operating environment where first hand experience is limited, are we in a position to say what is normal behaviour? The team at Dalhousie University will investigate anomaly detection and risk using maritime traffic data, with emphasis on surveying common approaches to maritime anomaly detection and qualitatively assessing how their application in the Arctic may be challenged. The next step will be to understand how the anomalous type, quantity, location and behaviour of the vessels relate to risk. The risk assessment component will explore the use of multi-criteria risk analysis to assess the risk posed by vessel behaviour in the Arctic region, and in particular how anomaly detection provides in indicator of actual or impending problems.