Shipping lane traffic density mapping from fused AIS and SAR

Fusing spaceborne AIS position reports with SAR-detected vessel centroids produces gridded traffic-density maps that expose route shifts, dark fleets and chokepoint stress that neither source can reveal alone.

Sensors

Sentinel-1 SAR (ESA/Copernicus): C-band synthetic aperture radar at 5.405 GHz. Extra Wide Swath mode covers 400 km at 20 m ground-range resolution, revisiting most ocean areas every 6 to 12 days per satellite; the two-satellite constellation halves that interval. Detects vessels as bright point targets regardless of cloud or darkness. Minimum detectable vessel length is roughly 30 m in calm seas; performance degrades in high sea states above Beaufort 5.
Spire Global spaceborne AIS: Over 100 LEO satellites receiving VHF AIS messages (161.975 MHz and 162.025 MHz). Provides global coverage with typical revisit of 20 to 40 minutes in mid-ocean and near-continuous coverage in busy lanes. Carries MMSI, position, speed and heading for reporting vessels. Collision and self-interference in dense lanes reduces message-capture probability.
exactEarth spaceborne AIS: Complementary LEO AIS constellation. Used as a cross-check or gap-filler against Spire, particularly in polar and Southern Ocean corridors where satellite geometry differs. Same VHF band; similar limitations in ultra-dense traffic.
ICEYE SAR constellation: X-band SAR (9.65 GHz) with stripmap resolution down to 1 m and spotlight to sub-metre. Commercially taskable for same-day revisit over specific chokepoints such as the Strait of Hormuz or Bab-el-Mandeb. Better discrimination of vessel class and length than Sentinel-1 in congested scenes, at the cost of narrower swath (typically 30 to 50 km in fine modes).

Why neither source is sufficient on its own

AIS was designed for collision avoidance, not fleet surveillance. Any vessel master who wishes to avoid scrutiny can disable the transponder, and many do. Fishing vessels below 300 gross tonnes are not legally required to carry AIS at all. In contested waters, military and paramilitary craft routinely go dark. The result is that an AIS-only density map systematically under-counts traffic in precisely the areas an analyst most wants to monitor.

SAR solves the detection problem but creates an identity problem. A bright point target in a Sentinel-1 Extra Wide Swath image is a vessel; it is not obviously a crude tanker, a bulk carrier or a fishing boat. Without AIS correlation, you can count ships and estimate their length from the target extent, but you cannot assign flag, owner, cargo type or voyage history. For route-shift analysis, that context is often the whole point.

How the fusion grid is built

The standard approach treats each data source as an independent observation layer over a common spatial grid. Cells are typically 0.1 by 0.1 degrees for ocean-basin work, or 0.01 by 0.01 degrees for chokepoint analysis. AIS position reports are binned into the grid over a chosen time window, weighted by the known revisit probability of the receiving constellation to correct for under-sampling in high-latitude or low-traffic cells. SAR detections are co-registered to the same grid after applying a constant-velocity motion compensation to account for the time offset between AIS snapshot and SAR acquisition, which can be several hours.

The fusion step matches AIS tracks to SAR centroids within a position-uncertainty ellipse derived from vessel speed, heading and the AIS-to-SAR time delta. Matched targets carry AIS identity. Unmatched SAR detections, those with no plausible AIS counterpart within the ellipse, are flagged as non-reporting vessels and counted separately. The ratio of matched to unmatched detections in each grid cell is itself a useful signal: a cell where 40 percent of SAR targets have no AIS match is behaving very differently from one where the figure is 5 percent.

Over time, repeated acquisitions accumulate into a density surface. Monthly or quarterly composites expose structural lane geometry. Differencing two composites, say the quarter before and after a canal disruption, isolates the route shift with reasonable statistical confidence, provided the SAR acquisition cadence is consistent across both periods.

What the density map actually reveals

The most immediate product is a heatmap of vessel counts per cell per unit time. This sounds simple, but the interpretive value is high. Shipping lanes are not fixed physical features; they emerge from the aggregate decisions of thousands of operators responding to fuel prices, piracy risk, insurance zone classifications, seasonal weather and geopolitical events. A density map from before and after the 2021 Ever Given grounding in the Suez Canal, for instance, would show a measurable redistribution of traffic onto the Cape of Good Hope route, with a corresponding increase in voyage durations visible in the AIS track data.

Chokepoints deserve particular attention. The Strait of Malacca, the Bab-el-Mandeb and the Strait of Hormuz each funnel enormous volumes of trade through corridors narrow enough that ICEYE spotlight acquisitions can capture the full width in a single pass. In these locations, daily SAR tasking combined with continuous AIS ingestion produces a near-real-time throughput index: vessel count per day, stratified by vessel class where AIS identity is available.

Non-reporting vessel density is a separate product with its own audience. Sanctions-compliance teams, coast guards and maritime insurance underwriters all want to know where dark vessels concentrate. A persistent cluster of unmatched SAR detections in a cell with low legitimate traffic is worth investigating. The density map does not prove wrongdoing; it produces a ranked list of cells that warrant closer examination.

Honest limits of the method

Sentinel-1's 6 to 12 day revisit per satellite is adequate for monthly composites but too slow to track a single vessel's voyage in real time. ICEYE can task daily, but its narrow swath means you must already know which cell to watch. The combination works well operationally: Sentinel-1 provides the wide-area baseline that identifies anomalous cells, and ICEYE provides the confirmatory close look.

Sea state is a genuine constraint. Sentinel-1 vessel detection performance falls noticeably in significant wave heights above 3 to 4 metres, because the sea clutter power begins to approach the vessel return power for smaller targets. This is not a solvable calibration problem; it is physics. Analyses covering the Southern Ocean in austral winter, or the North Atlantic in winter, carry wider uncertainty bounds and should say so.

AIS message-capture probability in the Strait of Malacca or the English Channel, where hundreds of vessels transmit simultaneously on the same VHF channels, is meaningfully lower than in open ocean. Spire and exactEarth both publish coverage statistics, but those figures are fleet-wide averages. In ultra-dense lanes, the effective under-count can be 10 to 20 percent even for vessels that are nominally transmitting. The density map should be presented with per-cell confidence intervals where the data support it.

From grid cells to decisions

The analytic products that flow from the density grid divide roughly into strategic and operational. Strategic users, port authorities, trade ministries, shipping insurers, want quarterly trend reports showing lane utilisation, emerging routes and chokepoint stress indices. These are best delivered as GIS layers with accompanying statistical summaries, updated on a fixed schedule.

Operational users want alerts. A sanctions-compliance team does not need a beautiful heatmap; it needs a notification when a dark-vessel cluster appears in a cell adjacent to a known ship-to-ship transfer zone. That alert can be generated automatically by thresholding the unmatched-SAR-detection count in a watchlist of cells and triggering when the count exceeds a rolling baseline by a defined margin.

Satellize runs this fusion pipeline on open Sentinel-1 acquisitions combined with commercial AIS ingestion, applying the same gridding and matching logic it uses in its crop-estimation work for the Kingdom of Tonga: a consistent, auditable method that clients can interrogate rather than a black box. If you want to understand the methodology before commissioning a study, the right first step is a scoping call with the analytics team to define the geographic extent, time window and output format.

Typical figures

SAR spatial resolution (Sentinel-1 EW mode)	20 m ground range, 40 m azimuth
SAR spatial resolution (ICEYE spotlight)	Sub-1 m (commercially taskable)
Sentinel-1 revisit (two-satellite)	6 days at equator; shorter at higher latitudes
ICEYE revisit over a named chokepoint	Same-day to daily with commercial tasking
Spire AIS mid-ocean revisit	Typically 20 to 40 minutes
Minimum detectable vessel (Sentinel-1, calm sea)	Approximately 30 m length
Density grid cell size (ocean-basin)	0.1 × 0.1 degrees (~11 km at equator)
Density grid cell size (chokepoint)	0.01 × 0.01 degrees (~1.1 km at equator)
Sentinel-1 archive depth	From April 2014 (Sentinel-1A launch)
Delivery formats	GeoTIFF density rasters, GeoJSON vessel-event points, CSV summary tables, alert feeds

Analytics Satellize can run

Monthly vessel-density composite	AIS position binning plus SAR detection gridding; motion-compensated co-registration	GeoTIFF heatmap layer with per-cell vessel count and confidence interval
Route-shift difference map	Cell-by-cell subtraction of two monthly or quarterly density composites	Signed-difference GeoTIFF and PDF report quantifying lane redistribution
Chokepoint throughput index	Daily ICEYE SAR vessel count correlated with AIS class attribution in a defined corridor polygon	Time-series CSV and dashboard feed of vessel count per day by class
Non-reporting vessel density layer	Unmatched SAR detections after AIS correlation; ratio of dark to AIS-matched targets per cell	GeoJSON layer of dark-vessel cluster cells with count and baseline-deviation score
Sanctions-zone proximity alert	Threshold trigger on unmatched SAR detections in user-defined watchlist cells	Near-real-time alert (email or API push) with cell coordinates, detection count and acquisition timestamp
Historical lane-utilisation trend	Sentinel-1 archive processing from 2014 to present; AIS back-fill where available	Annual density composites as GeoTIFF stack with accompanying statistical summary

Sources

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.