Sensors
- Sentinel-1 SAR (ESA/Copernicus): C-band SAR at 5.405 GHz. Interferometric Wide Swath mode gives 250 km swath at 5 x 20 m ground resolution. Repeat cycle 6 days for a single satellite, shorter with both units operational. Free and open archive from 2014. Sufficient to resolve large tankers and detect close-proximity pairs, but marginal for identifying fender deployment on vessels under roughly 100 m.
- ICEYE SAR constellation: X-band SAR with Spot mode resolution down to approximately 1 m and Strip mode at roughly 3 m. Revisit to a specific location can be contracted to sub-daily through tasking. X-band backscatter is sensitive to metallic deck equipment, improving fender and hose detection. Commercial tasking adds cost but is the practical route to catching a transfer in progress.
- Capella Space SAR: X-band SAR with published spotlight resolution of approximately 0.5 m in highest-resolution modes. At that fidelity, deck equipment, mooring lines and freeboard differences between the two vessels become visible. Useful for confirmatory imaging once a candidate rendezvous location has been identified from coarser data.
- Spire Global spaceborne AIS: A constellation of over 100 LEO satellites collecting Automatic Identification System VHF messages globally. Provides near-global AIS coverage with typical latency of minutes to low tens of minutes. Critically, it reports what vessels broadcast, not where they physically are. Discrepancies between Spire-collected AIS positions and SAR-detected positions are the primary indicator of deliberate spoofing.
What a ship-to-ship transfer actually looks like from orbit
Two large tankers moored side by side in open ocean present a distinctive SAR signature. The vessels appear as two elongated high-backscatter targets separated by a gap of roughly 5 to 20 metres, with the gap itself showing a characteristic shadow pattern caused by the fenders and the hull geometry blocking radar return. At Sentinel-1 resolution this appears as a paired bright-line feature. At ICEYE or Capella resolutions, individual deck structures, mooring equipment and the freeboard difference between a loaded and a partially discharged vessel become distinguishable.
The transfer itself typically takes 12 to 48 hours for a very large crude carrier cargo. That duration is both the opportunity and the constraint: the window is long enough that a single SAR pass has a reasonable probability of coinciding with the event, but the ocean is large and tasking must be directed at the right area. Most transfers in documented evasion operations occur in a small number of preferred zones: the Strait of Hormuz approaches, waters off Ceuta, the Laconian Gulf, and the South China Sea. Concentrating revisit capacity on these zones materially improves detection probability.
AIS manipulation and why physics does not cooperate
The standard evasion technique is to broadcast a falsified AIS position placing the vessel hundreds or thousands of kilometres from its actual location while the transfer takes place. From a terrestrial AIS network, this is largely undetectable. From a spaceborne AIS receiver passing overhead, the broadcast position can be compared directly against the SAR-detected position in the same image. A vessel whose AIS signal places it in the Caspian Sea while its radar return is sitting in the Gulf of Oman has a problem.
The comparison is not always clean. AIS messages can be absent entirely (the transponder is simply switched off), the vessel may be transmitting a plausible but incorrect position that falls within the SAR scene, or timing offsets between the AIS collection and the SAR acquisition introduce uncertainty. Spire's constellation collects AIS with latency typically under 30 minutes, and Sentinel-1 acquisition times are known to the second, so temporal alignment is tractable. The harder problem is vessels that go fully dark: no AIS, no spoofed position, just a radar return with no associated identity. Resolving identity from SAR alone requires vessel length and beam measurement, wake geometry and, where available, cross-referencing with optical imagery or prior AIS history. That is a separate analytic chain, covered in the dark-ship detection page in this library.
Revisit frequency determines what you catch
A single Sentinel-1 satellite revisits a given ocean area roughly every 6 days in standard operations. Over a 48-hour transfer window, the probability of a single Sentinel-1 pass coinciding with the event at a specific location is around 30 to 35 percent, assuming the pass falls within the swath. That sounds low. Over a 30-day period, with multiple candidate locations monitored, the cumulative detection probability rises substantially, particularly when combined with commercial SAR tasking triggered by AIS anomaly alerts.
The practical architecture for a monitoring programme is layered. Spire AIS data runs continuously, flagging vessels that enter a watch-list zone, show position discontinuities, or go dark for anomalous durations. Those flags trigger priority tasking of ICEYE or Capella for same-day or next-day imaging. Sentinel-1 provides the background archive that allows analysts to reconstruct a timeline even when commercial tasking was not active. No single sensor closes the gap alone.
Honest limits of the method
SAR works through cloud cover and at night, which is a genuine advantage over optical sensors in this context. But it has its own ambiguities. Two vessels in close proximity for a legitimate reason, such as a rescue, a fuel bunkering operation, or a pilot transfer, produce an identical geometric signature to a sanctions-evasion transfer. Fender deployment is visible at sub-metre resolution but not at Sentinel-1's 5 x 20 m pixel. Duration of proximity, vessel identity, prior voyage history and destination port calls are all required to build an evidential picture. SAR imagery alone is not a legal finding.
Archive depth matters for building that picture. Sentinel-1 data is available from 2014, and Copernicus DataSpace provides programmatic access. This means a vessel's historical behaviour, including previous proximity events, can be reconstructed. Commercial SAR archives are shallower, typically two to three years depending on the operator, and access is licensed rather than free. Resolution improvements in newer commercial systems come at the cost of narrower swaths, so wide-area surveillance still depends on Sentinel-1 or similar broad-coverage assets.
From detection to intelligence product
A detection event, two vessels in close proximity with AIS anomalies, becomes actionable intelligence only when it is placed in context. The analytic workflow typically produces a rendezvous report that includes: the SAR-detected positions and timestamps of both vessels, the AIS broadcast positions at the same time, vessel identity candidates derived from length and beam measurements, prior port calls from AIS history, and the probable cargo state of each vessel inferred from freeboard. Freeboard estimation from SAR is approximate, with uncertainty of roughly one to two metres depending on sea state and incidence angle, but it is sufficient to distinguish a loaded from a ballasted very large crude carrier.
Satellize runs this workflow on a combination of open Sentinel-1 data and contracted commercial SAR tasking, with Spire AIS as the triggering layer. The Tonga crop-estimation programme is a different domain, but the underlying architecture, open data plus commercial tasking plus analytic pipeline, is the same model applied here to maritime intelligence. Clients receive a structured alert feed and, on request, a full evidential package formatted for compliance or regulatory submission.
What the public record establishes
The use of ship-to-ship transfers to evade sanctions on Iranian and Venezuelan crude is documented in United Nations Panel of Experts reports, US Treasury OFAC advisories, and published research using Sentinel-1 and commercial SAR. The UN Panel of Experts on North Korea has specifically cited satellite imagery of STS transfers in its annual reports. These are not hypothetical scenarios.
The physics is also well established. C-band backscatter from steel hulls at maritime incidence angles produces strong, stable returns. Vessel detection in SAR at Sentinel-1 resolution has been validated across multiple published studies, with detection probabilities above 90 percent for vessels over 100 m in low sea-state conditions, falling as sea state rises and vessel size decreases. In high sea states (above Beaufort 5), smaller vessels become difficult to separate from wave clutter. That is a real operational limit and should be factored into any monitoring programme design.
Typical figures
| SAR spatial resolution (Sentinel-1 IW mode) | 5 x 20 m ground range |
| SAR spatial resolution (ICEYE Spot mode) | Approximately 1 m |
| SAR spatial resolution (Capella Spotlight) | Approximately 0.5 m (published) |
| Sentinel-1 revisit (single satellite) | 6 days at a given location |
| Commercial SAR revisit (tasked) | Sub-daily achievable with ICEYE or Capella tasking |
| Spire AIS collection latency | Typically under 30 minutes globally |
| Minimum detectable vessel (Sentinel-1, low sea state) | Approximately 50 m length; detection probability degrades in sea states above Beaufort 5 |
| Sentinel-1 archive depth | 2014 to present, open access via Copernicus DataSpace |
| SAR operating frequency | C-band (5.405 GHz, Sentinel-1); X-band (9.6 GHz approx., ICEYE and Capella) |
| Delivery formats | GeoTIFF (SAR scenes), GeoJSON or Shapefile (detection events), structured JSON alert feed, PDF evidential report |
Analytics Satellize can run
| Proximity event detection | SAR ship detection followed by paired-target geometry analysis; vessels flagged when centre-to-centre distance falls below 50 m and both targets exceed minimum size threshold | Timestamped GeoJSON alert with vessel positions, estimated lengths and confidence score |
| AIS position discrepancy report | Temporal matching of Spire AIS broadcast positions against SAR-detected positions within the same acquisition window; discrepancy calculated in kilometres | Per-event discrepancy table in structured report, flagged by severity band |
| Vessel identity candidate list | SAR-derived length and beam estimation compared against IMO vessel registry dimensional data; ranked candidate list produced | Candidate vessel table with MMSI, IMO number, flag state and dimensional match score |
| Cargo state estimation | Freeboard proxy from SAR incidence angle and backscatter geometry; distinguishes loaded from ballasted condition to ±1 to 2 m under favourable sea state | Load-status indicator (loaded/ballasted/uncertain) appended to proximity event record |
| Voyage history reconstruction | AIS track interpolation from Spire historical archive; prior port calls, dark periods and speed anomalies extracted over a configurable lookback window | KML or GeoJSON track file with annotated events; summary timeline in PDF |
| Watch-zone continuous monitoring feed | Automated AIS anomaly detection (dark events, position jumps, speed discontinuities) across defined ocean polygons; triggers commercial SAR tasking request on threshold breach | Daily alert digest with tasking log; monthly summary of zone activity |
| Evidential package for compliance submission | Structured collation of SAR imagery, AIS discrepancy data, vessel identity candidates and voyage history into a single auditable document | PDF report with georeferenced imagery, metadata chain and analyst commentary, formatted for regulatory or legal use |
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.