Sensors
- Sentinel-1 SAR (C-band, ESA): Primary detection layer. Interferometric Wide Swath mode covers 250 km at 5 x 20 m ground resolution, day and night, through cloud. Repeat pass over a fixed point is 6 days with one satellite, 3 days with both active. Detects vessels down to roughly 30–50 m length depending on sea state and incidence angle; smaller craft are ambiguous.
- VIIRS Day-Night Band (NOAA/NASA, Suomi-NPP and NOAA-20): Detects fishing and tanker lighting at night at 750 m resolution. The VIIRS Nightfire and boat-detection products from Colorado School of Mines and NOAA STAR document vessel lights independently of AIS. Useful for flagging active transfers after dark, though a vessel running lights-out is not detected.
- Planet SuperDove (PlanetScope): Daytime optical confirmation at 3–4 m native resolution, 8 spectral bands. Near-daily revisit over most ocean areas. Useful for visual confirmation of vessel proximity and hose or fender rigging, but cloud cover and sun glint over open water degrade usability. Cannot replace SAR as a primary layer.
- Maxar WorldView Legion: 30 cm panchromatic, sub-metre multispectral. Taskable for high-confidence vessel identification, flag state, and hull markings once a candidate pair is cued by SAR or VIIRS. Revisit up to 15 times per day over high-priority areas, but coverage is narrow and tasking must be planned ahead of the event.
- AIS terrestrial and satellite feeds (commercial aggregators): Not a satellite sensor in the imaging sense, but the essential negative-space layer. Gaps in AIS transmission, spoofed positions, or identity switches around the time of a SAR-detected rendezvous are the primary analytical signal. Spire Global and others provide historical and near-real-time AIS via satellite receivers.
What the physics gives away when the transponder goes dark
A tanker displacing 100,000 tonnes of crude is not a subtle radar target. At C-band, the metal superstructure, hull, and any cargo-transfer equipment produce a strong, stable backscatter signature that Sentinel-1 records reliably regardless of cloud, fog, or time of day. Two such vessels moored side by side in open water, well away from any port, produce a paired bright return that is geometrically distinct from normal underway traffic.
The evasion logic is straightforward: a sanctioned vessel transfers cargo to a second vessel at sea, both ships disable or spoof their AIS transponders during the rendezvous, and the receiving vessel then proceeds to a legitimate port with a clean paper trail. The origin of the cargo is obscured. What this logic cannot defeat is the physics of microwave backscatter. The vessels are still there, still large, and still detectable by any SAR system with adequate resolution and revisit.
The AIS gap as an analytic signal, not just a data absence
AIS gaps are common and mostly innocent: poor satellite geometry, congested frequencies, legitimate equipment faults. The gap becomes analytically significant when it correlates with other signals. A vessel that goes dark in a known high-risk area, reappears on a different heading, shows a sudden change in draught inferred from waterline imagery, or whose AIS track is inconsistent with the vessel's last known speed, is a candidate for investigation rather than a confirmed violation.
UN Panel of Experts reports on the Democratic People's Republic of Korea, published under Security Council resolution 1874 and its successors, have documented this method in detail. The reports describe specific vessel pairs, coordinates, and dates where SAR imagery corroborated AIS gaps for petroleum product transfers. Iran-related panels have done the same. These are public documents and constitute the clearest open-source validation that the combined method works at operational scale.
The honest caveat is that AIS gap analysis alone produces a large number of false positives. The gap narrows the search space; SAR imagery confirms or rules out physical proximity; optical or high-resolution tasking then provides identification. Skipping any step increases either false positives or false negatives, depending on which layer is dropped.
Sentinel-1 as the workhorse, and where it falls short
Sentinel-1's Interferometric Wide Swath mode is the practical choice for systematic ocean surveillance: free data, global archive back to 2014, and sufficient resolution to detect large vessels reliably. The 3-day revisit with two satellites sounds adequate, but a ship-to-ship transfer lasting 12 to 24 hours can begin and end between passes. A vessel pair present at 06:00 UTC on day one may be gone by the next acquisition on day four.
Resolution is the second constraint. At 5 x 20 m in IW mode, two vessels moored very close together may appear as a single elongated return rather than two distinct targets. Distinguishing a legitimate vessel escort from an active transfer requires either higher-resolution SAR (commercial providers such as ICEYE or Capella Space offer sub-metre SAR, though at higher cost and narrower swath) or corroborating optical imagery. Sentinel-1 is the screening layer. It is not, by itself, the confirmation layer.
Sea state matters too. High wave heights increase clutter and raise the effective detection threshold. Vessels under roughly 50 m in length become unreliable detections in Beaufort 5 or above. Most tankers involved in large-scale petroleum transfers are well above this threshold, but smaller vessels used in arms transfers or fuel bunkering may fall below it.
Night detection and the VIIRS boat-detection product
VIIRS DNB on Suomi-NPP and NOAA-20 passes at roughly 01:30 local time. The Day-Night Band detects emitted light at 750 m resolution, which is coarse for vessel identification but sufficient to flag a lit cluster of ships in open water where no lit cluster should be. The NOAA STAR and Colorado School of Mines Nightfire products process this data operationally and are publicly accessible.
The obvious limitation: a vessel running completely dark, with no deck lights, no navigation lights, and no fishing gear illumination, is invisible to VIIRS DNB. Sophisticated actors conducting high-value transfers at night do sometimes run dark. VIIRS is therefore a useful supplement for detecting less disciplined operators, not a reliable primary layer against a determined, well-resourced evader.
Building a detection workflow that a sanctions analyst can actually use
The operational sequence that emerges from published UN methodology and open remote-sensing literature runs roughly as follows. First, screen AIS feeds for vessels of interest: known sanctioned ships, flag-of-convenience registries under scrutiny, or any vessel showing anomalous dark periods in high-risk ocean zones such as the East China Sea, the Gulf of Oman, or the waters west of the Yucatan Peninsula. Second, task or pull archived Sentinel-1 acquisitions over the gap location and time window. Third, apply vessel detection algorithms to the SAR scene to identify candidate pairs at close proximity. Fourth, cross-reference with VIIRS for any nocturnal light signal. Fifth, task a high-resolution optical or commercial SAR asset for visual confirmation if the candidate pair is flagged.
The output of this workflow is not a legal finding. It is an evidentiary package: timestamped imagery, AIS records, vessel dimensions inferred from SAR, and a documented chain of custody for the data. UN panels and national intelligence agencies then apply their own legal and political judgement. The satellite data provides the factual substrate, not the verdict.
Satellize structures this workflow as a standing monitoring brief for government clients, drawing on open Sentinel-1 and VIIRS data supplemented by commercial tasking on client licence.
What this method cannot do
It cannot identify cargo. SAR sees a metal hull, not what is inside it. Inferring petroleum transfer requires corroborating signals: vessel type, known routes, draught change, and AIS history. Arms transfers in smaller vessels are harder still, because the vessels are smaller, the cargo lighter, and the draught change minimal.
It cannot achieve real-time detection with open data. Sentinel-1's 3-day revisit and the latency of free data distribution (typically 1 to 3 hours after acquisition at best, often longer) mean that by the time an analyst sees a confirmed rendezvous, the vessels have separated and are underway. The value is investigative and evidentiary, not interdiction. For anything approaching real-time cueing, a client needs commercial SAR tasking and a dedicated processing pipeline, which narrows the cost-benefit calculation significantly.
Finally, it cannot distinguish a sanctioned transfer from a legitimate ship-to-ship operation. Bunkering, crew transfers, and cargo lightering are all legal and all look similar in SAR imagery. Context, vessel identity, and location relative to known shipping lanes are what separate the suspicious from the routine. The method generates candidates; human judgement closes the case.
Typical figures
| Primary SAR resolution (Sentinel-1 IW mode) | 5 m range x 20 m azimuth; 250 km swath |
| Sentinel-1 revisit (two satellites active) | ~3 days at mid-latitudes; 6 days with one satellite |
| VIIRS DNB spatial resolution | 750 m; single nightly pass at ~01:30 local solar time |
| Planet SuperDove optical resolution | 3–4 m; near-daily revisit; 8 spectral bands (443–885 nm) |
| Maxar WorldView Legion resolution | 30 cm panchromatic; up to 15 revisits per day over tasked area |
| Minimum detectable vessel (Sentinel-1, calm sea state) | ~30–50 m length; smaller vessels unreliable above Beaufort 4–5 |
| SAR archive depth (Sentinel-1) | Global coverage from April 2014 (Sentinel-1A launch) |
| VIIRS boat-detection archive depth | Suomi-NPP operational from 2012; NOAA-20 from 2018 |
| Data latency (Sentinel-1 open distribution) | Typically 1–3 hours post-acquisition via Copernicus Data Space |
| Delivery formats | GeoTIFF, GeoJSON vessel detections, PDF evidentiary report, AIS-correlated timeline spreadsheet |
Analytics Satellize can run
| AIS dark-event register | Temporal gap analysis on satellite AIS feeds; cross-referenced against vessel class, flag state, and known high-risk corridors | Weekly CSV of candidate dark events with coordinates, duration, and risk score; filterable by region or vessel type |
| SAR vessel-pair detection layer | Constant False Alarm Rate (CFAR) ship detection on Sentinel-1 IW scenes; proximity clustering to flag pairs within 200 m in open water | GeoJSON layer of detected vessel pairs with timestamps, estimated vessel dimensions, and scene metadata; ingestible into standard GIS platforms |
| AIS-SAR correlation report | Temporal and spatial join of AIS gap events to SAR-detected vessel positions; flags cases where a dark vessel's last known position is consistent with a SAR detection | Structured PDF evidentiary package per candidate event, suitable for submission to a sanctions panel or legal team |
| VIIRS nocturnal light anomaly alert | Difference imaging against baseline VIIRS DNB composites; flags anomalous light clusters in open ocean away from known fishing grounds | Near-daily alert feed (email or API) with coordinates and estimated cluster size |
| High-resolution confirmation tasking brief | Cued tasking request for Planet or Maxar assets over SAR-confirmed candidate pairs; analyst-reviewed imagery with vessel identification notes | Annotated image report with hull markings, flag state (where visible), estimated vessel length, and comparison against known vessel registry photographs |
| Historical pattern-of-life analysis | Multi-month retrospective query of Sentinel-1 archive and AIS history for a named vessel or fleet; identifies recurring rendezvous locations and seasonal patterns | Analytical briefing document with mapped rendezvous hotspots and timeline of candidate events; suitable for intelligence or legal preparation |
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.