By Alessandro Pititto, Cogea Bip-Group, Coordinator of EMODnet Human Activities
In March 2017, EMODnet was mandated to create vessel density maps of EU waters showing the average number of vessels of certain type (cargo, passenger, fishing etc.) for a given period within a grid cell. This was a direct result of user demand, following a survey carried out in 2016 by EMODnet Human Activities where vessel density maps were, by far, the most requested geospatial data product. The maps went live on EMODnet (https://emodnet.ec.europa.eu) for visualisation and download on 11 March 2019.
What are Vessel Density maps?
EMODnet’s Human Activities’ Vessel Density maps show where ships travel in EU waters and how intensively they use different areas of the sea. Each map visualises vessel density: essentially, the amount of time ships spend in each square-kilometre grid cell over a given month. Unlike simple counts of ship tracks or positions, these maps reflect the true average number of ships present in an area, accounting for vessel speed, irregular AIS transmissions, and the actual time vessels spend within each cell.
How are the maps built?
The method reconstructs a ship’s movement from Automatic Identification System (AIS) data by drawing a line between each pair of consecutive AIS positions. For every segment of these reconstructed routes, the system calculates the duration the vessel spent in each grid cell based on the length of the segment lying within the cell. By summing all such durations for all vessels, the maps show how many ship-hours per square kilometre accumulated in every cell for each month and ship type. This approach produces a realistic measure of density: not merely how many times ships passed through an area, but how likely you would be to find a ship in that area at any moment.
Figure 1. The figure shows an example with a single ship sailing at different speeds in the same grid cell. From these values it is calculated how much time a ship spends in a given cell over a time period, in turn enabling the calculation of the density value for all ships crossing this grid cell per km² (from: EU Vessel Density Map, Detailed method, v1.6 ).
The maps represent the outcome of a multi-step processing approach that includes defining the area of interest, building a 1×1 km geospatial grid, importing and cleaning the collection of AIS records, creating points and lines, and intersecting these lines with the grid. Unlike many existing density products, the EMODnet approach uses a detailed, transparent, and fully documented methodology. This ensures that the final density values are comparable across the entire EU marine area and free for any type of reuse.
Because the maps store raw time spent per cell (rather than pre-normalised density), users can convert them into other density units – such as average instantaneous number of ships per km² or per square degree – according to their analytical needs. The product is therefore both scientifically robust and flexible.
The method originally developed by the EMODnet Human Activities group has become a de facto standard in literature. It’s being used by entities all around the world, including the US NGA.
How much data is needed to build the Vessel Density maps, and how is this approached?
The maps are based on all terrestrial and satellite messages sent by ships sailing in EU waters. Depending on the year, there might be hundreds of billion records to process. On average, the team process approximately 3 billion records per year, after several steps to reduce the volume of data to process.
The first step includes down sampling or reducing the number of AIS messages by enforcing a minimum time interval between consecutive signals from the same vessel. In the dataset used for the Vessel Density maps, AIS messages were down sampled to 3 minutes, meaning that two consecutive messages from the same ship could not be closer than 3 minutes apart, whereas normally messages are a few seconds apart. Down sampling is quite useful because AIS data can be dense and redundant when making a density maps.
The number of records is further reduced by removing points on land (as these points are obvious errors), enforcing allowed AIS message types (only message types relevant to ship positions or static information were kept: 1, 2, 3, 5, 18, 19, 24, and 27, whereby AIS message types that do not contribute to vessel-movement reconstruction are removed), remove duplicate signals, and filter invalid MMSI numbers (i.e. MMSI numbers are the unique identifiers for AIS devices, but many records contain invalid values).
Further, to eliminate satellite-generated “noisy” positions, a Kalman filter, tuned to expected ship movement patterns, was used. The filter checks whether each new observation is consistent with expected track behaviour, vessel speed, likelihood thresholds innovation metrics.
Finally, in the early phase – especially for creating points and lines from AIS data – the team relied heavily on desktop GIS tools. As data volumes grew, the team switched to a workflow that minimised desktop GIS use and leveraged SQL + PostGIS for computationally heavy tasks (database-centric workflow).
Final product
The final output consists of raster maps that reveal spatial patterns of maritime traffic across the EU’s waters. Because the data is broken down by ship category (cargo, tanker, fishing, passenger, pleasure craft, etc.) and by month, the maps highlight both sector-specific traffic patterns and seasonal variations. Busy shipping corridors, port approaches, and fishing hotspots appear as areas of high density, whereas remote or less-travelled areas show lower values. Importantly, the maps are based on both terrestrial and satellite AIS signals, covering nearly all vessel activity that transmits AIS.
These maps therefore function as a long-term, consistent observation tool for marine planners, policymakers, researchers, and commercial users. They provide insight into shipping intensity, maritime safety considerations, environmental pressures (such as noise or emissions associated with traffic), and patterns of human activity at sea.
Screenshot of the Vessel Density maps in the EMODnet Map Viewer