Mapping Maritime Risk in the Kattegat Using the Automatic Identification System
Lecq, L. van der
MetadataShow full item record
The Kattegat strait between Denmark and Sweden features high vessel traffic densities and difficult navigational conditions, making it susceptible to ship collisions. This led to the implementation of a new shipping route system as of July 2020. This study has assessed the change in the spatial distribution of maritime risk in the north of the Kattegat after this implementation and its relation to the spatial pattern of vessel movements and recorded maritime incidents. The maritime risk concept is formed by extracting vessel proximity, encounter type, speed, and evasive manoeuvres from non-accident critical events. An automated maritime risk assessment has been developed using vessel tracking data from the Danish Automatic Identification System (AIS). Several millions of data points have been pre-processed in Python and stored in a PostGIS database using a basic line segment data model. A three-dimensional index has been generated to efficiently intersect segments in space and time. A kernel density estimation has been applied in QGIS to create heatmaps of maritime risk from the detected vessel encounter locations and their risk values. A 17% decrease in encounter frequency has been observed across the entire research area, which is in line with projections from before the shipping route changes. Maritime risk values increased slightly. Most ship encounters occurred along the central shipping corridor from and towards the Baltic sea and involved cargo vessels. This may be because these areas see the most vessel movements and cargo vessels are the predominant ship type. The maritime risk pattern along this corridor shifted to the east, which is most likely caused by the shipping route changes and the corresponding shift in traffic density. Encounters involving passenger vessels are relatively rare, but have high risk values. This may be due to their crossing of the busy cargo shipping corridor. It is concluded that maritime risk corresponds to traffic intensity in terms of the spatial pattern and vessel type. Due to spatial and temporal scope constraints, the generalisability of this conclusion is limited. The result do not show a pronounced relationship between the spatial pattern of maritime risk and the locations of recorded accidents. Correcting the estimated maritime risk density for the traffic density bias also did not lead to new insights.