Sea-lane traffic density mapping from multi-source detections
Aggregating spaceborne AIS and SAR detections over weeks or months reveals route-level traffic density, vessel-type mix, and seasonal shifts that no single-pass product can show. AIS-only maps systematically undercount traffic where transponder use is low or deliberately suppressed.
Sensors
- Spire Global spaceborne AIS: A constellation of more than 110 LEO satellites receiving VHF AIS messages (161.975 MHz and 162.025 MHz). Provides global revisit of roughly 20–40 minutes in open ocean, falling to near-continuous coverage in busy straits due to overlapping satellite passes. Position accuracy is that of the vessel's own GNSS, typically better than 10 m, but message collision in high-density areas degrades detection completeness.
- exactEarth spaceborne AIS: Commercial spaceborne AIS service with global coverage and archive depth extending back to approximately 2012, enabling multi-year trend analysis. Useful for constructing long-baseline density maps that reveal route shifts driven by seasonal ice, piracy advisories, or canal closures.
- Sentinel-1 SAR (C-band, ESA): C-band SAR at 5.405 GHz. In Interferometric Wide Swath mode, ground resolution is approximately 5 × 20 m with a 250 km swath. Revisit at the equator is 6–12 days per satellite; the two-satellite constellation halves that interval. Detects vessels regardless of transponder status. Minimum detectable vessel length is roughly 20–30 m in moderate sea states; performance degrades in sea states above Beaufort 5–6.
- RADARSAT Constellation Mission (C-band, CSA): Three-satellite C-band SAR constellation offering daily revisit over Canada's maritime zones and sub-daily revisit at high latitudes. Medium-resolution mode delivers 50 m resolution over 500 km swaths, suited to wide-area density accumulation. High-resolution spotlight modes reach 1–3 m but cover small areas, so they are used for classification rather than density mapping.
What a density map actually is, and what it is not
A vessel traffic density map is an accumulation product, not a tracking product. You bin a region into a grid, typically 0.01-degree to 0.1-degree cells depending on the ocean basin and the required detail, then count how many unique vessel detections fall into each cell over a chosen time window. The result is a heatmap where colour encodes detection frequency. A ship that transits the same lane daily for a month contributes 30 counts; a vessel that crosses once contributes one.
That distinction matters enormously for interpretation. High density in a cell could mean one vessel on a regular schedule, many vessels on irregular schedules, or an anchorage where ships sit for days. Disaggregating those cases requires pairing the density grid with dwell-time statistics and vessel-type filters. A density map that conflates anchored tankers with transiting bulk carriers will mislead an underwriter trying to assess collision exposure on a specific route.
Why AIS alone gives you a biased picture
Class A AIS transponders are mandatory on SOLAS-convention vessels above 300 gross tonnes on international voyages, and on all passenger ships. Smaller fishing vessels, domestic ferries, and many vessels operating in non-signatory flag states carry no transponder at all, or carry one that is switched off. Spaceborne AIS also suffers message collision: in the Strait of Malacca or the English Channel, where hundreds of vessels transmit simultaneously on the same VHF channels, a LEO satellite receiver may decode only a fraction of the messages in a single pass.
The practical consequence is that AIS-only density maps are reliable indicators of commercial deep-sea shipping on major trunk routes, and unreliable indicators of coastal, fishing, and informal traffic. Studies comparing spaceborne AIS with SAR detections in the South China Sea and West African EEZs have consistently found a dark-vessel fraction of 20–50% by count, heavily concentrated in fishing fleets and smaller coastal traders. Any density product used for environmental impact assessment or naval planning that does not account for this fraction is presenting a lower bound, not a true picture.
Fusing AIS and SAR into a calibrated density layer
The standard fusion approach matches SAR detections to AIS positions within a time-position tolerance, assigns matched detections to their AIS-reported vessel type, and flags unmatched SAR returns as dark. Both populations are then gridded separately and summed. The matched layer inherits vessel-type metadata from AIS; the dark layer is typed by SAR backscatter signature and length estimate where resolution permits.
Sentinel-1's 6–12 day revisit means that over a 90-day window you accumulate roughly 7–15 SAR passes per location at mid-latitudes, enough to build a statistically meaningful dark-vessel density layer for most ocean basins. RADARSAT Constellation's daily Arctic revisit makes it the preferred sensor for Northern Sea Route density work, where ice-free windows are short and traffic is growing. The honest limit is that SAR detection probability is a function of sea state, vessel aspect angle, and radar cross-section. A 25-metre wooden fishing vessel in a 3-metre swell may not appear in Sentinel-1 IW imagery at all.
Route-level statistics: what buyers actually use
A raw heatmap is a starting point. The derived statistics are what drive decisions. Route-level analysis extracts the dominant flow axes from the density grid using methods such as principal-curve fitting or kernel density ridge estimation, then computes per-route metrics: mean daily vessel count, vessel-type composition (tanker, bulk carrier, container, fishing, other), and seasonal index (the ratio of peak-month to trough-month density).
Environmental impact assessors use these statistics to estimate underwater radiated noise budgets and strike risk for cetacean populations along proposed cable or pipeline routes. Hull insurers use route-level collision exposure, measured as the mean number of other vessels within a defined radius of a given track per transit, to adjust premium models for specific voyages. Naval planners use seasonal variation to identify when a lane is effectively dormant and when it carries enough commercial traffic to complicate surveillance. Each application needs different temporal aggregation: a 7-day rolling window for near-real-time port approach monitoring, a 12-month baseline for underwriting, a multi-year archive for treaty compliance assessment.
Limits that are worth stating plainly
Density maps do not tell you vessel identity. A high-density cell in the western Indian Ocean tells you that many ships use that corridor; it does not tell you which ships, or whether any of them are of interest. Identity requires AIS correlation or optical/SAR classification, both of which are covered in sibling pages in this library.
Cloud cover is irrelevant for SAR-based density, but SAR archive depth is not unlimited. Sentinel-1 systematic global acquisition only became dense enough for reliable open-ocean density work after 2016, and Sentinel-1B's failure in 2021 reduced revisit until Sentinel-1C's launch in late 2023 restored the two-satellite cadence. Any multi-year trend analysis needs to account for these acquisition gaps explicitly, or the apparent traffic changes may be sensor artefacts rather than real route shifts. Satellize's analytics pipeline flags acquisition-gap periods in all density time-series outputs so that clients do not mistake a data hole for a quiet lane.
Delivering the product: formats and update cadence
Density grids are typically delivered as GeoTIFF rasters with a separate vector file carrying the route-axis statistics, or as a tile service that can be loaded directly into a client's GIS or web platform. Update cadence depends on the sensor mix: an AIS-only product can refresh daily; a fused AIS-plus-SAR product refreshes on the SAR acquisition cycle, typically weekly to fortnightly for Sentinel-1 at mid-latitudes.
For clients who need a near-real-time picture of a specific chokepoint, a hybrid approach works well. The AIS layer updates continuously; the SAR layer provides a periodic dark-vessel correction factor that adjusts the AIS counts upward by the observed dark fraction for that region and season. The correction factor is itself a product of the historical fusion analysis, not a guess.
Typical figures
| AIS position accuracy | Vessel's own GNSS, typically better than 10 m; temporal accuracy depends on message reception latency, usually under 5 minutes for spaceborne AIS |
| SAR spatial resolution (density mode) | Sentinel-1 IW: ~5 × 20 m ground resolution, gridded to 0.01–0.1 degree for density products; RADARSAT Constellation Medium: ~50 m |
| SAR revisit at mid-latitudes | Sentinel-1 (two satellites): 6 days; RADARSAT Constellation (three satellites): 1–4 days depending on latitude |
| AIS revisit (open ocean) | Spire Global: ~20–40 minutes mean; higher in busy straits due to overlapping passes |
| Minimum detectable vessel (SAR) | ~20–30 m length in Beaufort 3–4; detection probability falls sharply above Beaufort 5–6 and for low-RCS wooden hulls |
| Density grid cell size (typical) | 0.01 degree (~1 km) for chokepoint analysis; 0.1 degree (~11 km) for ocean-basin heatmaps |
| Archive depth | Sentinel-1 SAR: systematic global acquisition from ~2016; exactEarth AIS: from ~2012; Spire AIS: from ~2016 |
| Delivery formats | GeoTIFF raster, GeoJSON or Shapefile route-axis vectors, CSV time-series, OGC tile service (WMTS/XYZ) |
| Product latency (fused AIS + SAR) | AIS layer: near-real-time to 6 hours; SAR-corrected density layer: 1–3 days after acquisition depending on processing pipeline |
Analytics Satellize can run
| Ocean-basin density heatmap | Kernel density estimation over gridded AIS + SAR detection counts, with dark-vessel correction applied per region | GeoTIFF raster updated weekly, with per-cell vessel-type breakdown as auxiliary bands |
| Route-axis extraction and statistics | Kernel density ridge estimation or principal-curve fitting to identify dominant flow axes; per-axis vessel count, type mix, and seasonal index computed from 12-month baseline | Vector shapefile of route centrelines with tabular statistics; PDF summary report |
| Dark-vessel density correction layer | SAR-to-AIS matching within time-position tolerance; unmatched SAR returns gridded separately and expressed as a fraction of total detections per cell | GeoTIFF correction-factor layer and accompanying methodology note; updated monthly |
| Seasonal traffic variation index | Monthly density aggregation over multi-year archive; peak-to-trough ratio computed per route segment with confidence interval based on acquisition completeness | CSV time-series per named route segment; interactive chart for web dashboard embedding |
| Chokepoint collision-exposure metric | For each transiting vessel in AIS record, count contemporaneous vessels within defined radius (e.g. 2 nautical miles); aggregate as mean encounters per transit per route | Tabular report per chokepoint, suitable for marine underwriting input |
| Acquisition-gap audit | Cross-reference SAR acquisition log against density time-series to flag periods where reduced revisit may produce artefactual traffic dips | Annotated time-series with gap flags; included as standard in all multi-year trend products |
Who does the work
We can get this done for you. Satellize runs its own analyst desk and a strong science team. You do not buy a data feed and work out what it means; our people source the imagery, run the analysis described on this page, and hand you the answer with its confidence limits stated. Discuss this requirement.