Dark-vessel detection on known smuggling routes
SAR radar detects vessel hulls regardless of AIS status. Cross-referencing SAR positions with transponder gaps on documented smuggling routes in the Mediterranean, Gulf of Guinea and Strait of Malacca identifies vessels of operational interest.
Sensors
- Sentinel-1 SAR (ESA): C-band (5.405 GHz) SAR with 5 x 20 m resolution in Interferometric Wide Swath mode and a 250 km swath. Revisit is 6 days at the equator for a single satellite, 3 days with both Sentinel-1A and 1B operational. Covers all three focus routes systematically and free of charge. Sufficient to detect vessels above roughly 30 m in length reliably; smaller craft become ambiguous.
- ICEYE SAR (commercial): X-band SAR constellation offering Spot mode at approximately 0.5 m resolution and Strip mode at around 3 m. Revisit on a tasked basis can be sub-daily over a specific area. Resolves vessel type and deck configuration that Sentinel-1 cannot, though tasking cost limits routine wide-area use.
- Capella Space SAR (commercial): X-band SAR with Spotlight imagery at approximately 0.5 m resolution. Coherent change detection between passes can reveal whether a vessel at anchor has moved or been loaded. Useful for confirming identity on vessels already flagged by Sentinel-1 screening.
- Spire AIS (commercial): Spire operates a constellation of over 100 small satellites collecting AIS messages globally. Maritime AIS coverage in open ocean reaches near-complete; coverage degrades in congested nearshore areas where message collision is common. The gap record from Spire feeds the AIS-dark interval calculation used to score vessels of interest.
Why radar and not optical imagery
Optical sensors are useless at night and unreliable under cloud. The Mediterranean in winter, the Gulf of Guinea year-round, and the Strait of Malacca under monsoon conditions are all routinely obscured. Smuggling does not pause for clear skies. SAR illuminates the scene with its own microwave energy and is indifferent to cloud, haze or darkness.
The physical mechanism is straightforward. A steel or fibreglass hull presents a large radar cross-section relative to the surrounding sea surface, which returns very little energy at low sea states. The vessel appears as a bright point or elongated blob in the image. At Sentinel-1's 5 x 20 m resolution in Interferometric Wide Swath mode, a vessel of 30 m or more is detectable as a distinct target. Below that length, detection probability drops sharply and false positives from wave clutter increase, particularly at sea states above Beaufort 4.
The AIS gap as the primary indicator
AIS is mandatory under SOLAS for vessels above 300 gross tons on international voyages, and for all passenger vessels. Switching it off is not, by itself, proof of wrongdoing. Vessels legitimately disable AIS to avoid piracy in certain corridors, and technical failures occur. The signal becomes meaningful when it is absent on a route where AIS-dark behaviour correlates with documented illicit activity.
The analytical workflow is correlation, not accusation. A SAR image detects a vessel at a known position and time. The AIS record for that area and time window is queried. If no AIS message matches the detected vessel's position within a plausible drift radius, the vessel is flagged as AIS-dark. The flag is then scored against route risk, vessel size class, time of day, proximity to known transshipment anchorages, and historical patterns on that corridor. The output is a probability-weighted list, not a verdict.
Published casework from the UN Office on Drugs and Crime and the IMO's reports on the Gulf of Guinea document that AIS manipulation is a consistent feature of vessels involved in narcotics transshipment, illegal bunkering and arms movement. The Mediterranean route from North Africa to southern Europe has been extensively documented in EU border agency operational reports.
What the three focus routes actually look like in data
The Mediterranean corridor between Libya, Tunisia and the Italian coast is narrow enough that Sentinel-1's 250 km swath covers it in a single pass. Revisit at these latitudes is closer to 2 to 3 days. The challenge is volume: hundreds of legitimate vessels transit daily, so the signal-to-noise problem is one of filtering, not detection.
The Gulf of Guinea presents a different geometry. Distances are larger, vessel traffic is sparser in some sub-corridors, and the threat mix includes oil bunkering vessels operating without AIS near offshore platforms. Sentinel-1 archive depth back to 2014 allows baseline behaviour modelling: a vessel appearing repeatedly in the same offshore anchorage area without AIS, at intervals consistent with a bunkering cycle, is a different risk profile from a one-off gap.
The Strait of Malacca is among the world's most congested waterways, with AIS message collision a genuine data-quality problem in the nearshore zone. Spire's space-based AIS avoids the terrestrial receiver saturation issue, but even satellite AIS has message decode limits in very dense traffic. SAR detection here is valuable precisely because it is independent of the AIS channel entirely.
Resolution, revisit and the honest limits of the method
Sentinel-1 reliably detects vessels above roughly 30 m. A semi-submersible or a go-fast boat of 10 to 15 m presents a radar cross-section that can fall below the detection threshold, particularly in moderate sea states. Commercial X-band SAR from ICEYE or Capella resolves vessel type at sub-metre resolution but must be tasked: you need to know approximately where to look before committing a tasking order.
Revisit is the other constraint. A 3-day repeat means a vessel can transit the entire Mediterranean in the gap between passes. The practical answer is multi-source fusion: Sentinel-1 for systematic area screening, commercial SAR for targeted confirmation, and Spire AIS to reconstruct the transponder history. No single sensor closes the gap alone.
Atmospheric conditions affect SAR less than optical sensors, but very high sea states (Beaufort 6 and above) increase clutter and reduce detection confidence. Vessel detection algorithms tuned for calm water will generate false positives in rough conditions if thresholds are not adjusted dynamically.
From detection to an actionable intelligence product
Raw SAR vessel detection is a list of pixel coordinates and timestamps. The analytic value is in what comes next. Identity resolution draws on vessel databases (IMO numbers, MMSI records, flag registry data) to match detected vessels against known histories. A vessel with a history of flag changes, name changes, or previous AIS-dark incidents on the same route carries a materially higher risk score than an unknown first appearance.
Satellize runs this correlation layer on open SAR constellations and adds commercial tasking on client licence, similar in structure to the data pipeline used for the Kingdom of Tonga crop-estimation programme, adapted here for maritime rather than agricultural targets. The deliverable for an enforcement client is typically a scored vessel list with timestamps, SAR image chips, AIS gap intervals and a route-segment risk classification, formatted for integration into existing maritime operations centre workflows.
The method does not produce evidence admissible in court without chain-of-custody documentation and expert interpretation. It produces targeting intelligence: a prioritised list of vessels worth watching, intercepting or requesting flag-state information on. That distinction matters and should be stated plainly to any client.
Typical figures
| Sentinel-1 spatial resolution (IW mode) | 5 x 20 m ground range |
| Commercial SAR resolution (ICEYE / Capella Spotlight) | Approximately 0.5 m |
| Sentinel-1 revisit (dual satellite) | 3 to 6 days depending on latitude |
| Commercial SAR revisit (tasked) | Sub-daily over a defined area of interest |
| SAR frequency | C-band (5.405 GHz, Sentinel-1); X-band (approx. 9.6 GHz, ICEYE and Capella) |
| Minimum detectable vessel length (Sentinel-1, calm sea state) | Approximately 30 m; smaller vessels unreliable |
| Sentinel-1 swath width (IW mode) | 250 km |
| Sentinel-1 archive depth | 2014 to present |
| AIS data source | Spire space-based AIS constellation (100+ satellites) |
| Typical detection-to-alert latency | 2 to 6 hours post-acquisition for near-real-time SAR products |
Analytics Satellize can run
| AIS-dark vessel list | SAR vessel detection (CFAR algorithm class) cross-referenced against Spire AIS position records within a defined time-position window | Timestamped GIS point layer of AIS-dark detections with vessel size estimate, route segment and gap duration |
| Route-segment risk score | Historical frequency analysis of AIS-dark detections per corridor segment using Sentinel-1 archive back to 2014 | Heat-map layer and summary report showing high-frequency dark-vessel zones by route and season |
| Vessel identity resolution report | Matching SAR-detected vessel dimensions and position history against IMO registry and flag-state databases | Per-vessel PDF report with AIS history, flag changes, size class, and prior dark-interval record |
| Confirmed vessel type classification | High-resolution commercial SAR Spotlight imagery (ICEYE or Capella) with manual and algorithmic interpretation of deck structure | Image chip with vessel type annotation and confidence rating, delivered within 6 hours of tasking |
| Recurring anchorage anomaly alert | Coherent change detection on repeat-pass SAR over known offshore anchorage areas to identify vessels appearing at regular intervals without AIS | Automated alert with image comparison and interval statistics, integrated into client maritime operations centre feed |
| Multi-source fusion intelligence brief | Combination of SAR detections, Spire AIS gaps, vessel registry data and published route risk classifications into a single scored priority list | Weekly or on-demand PDF brief with top-ranked vessels of interest, supporting imagery and recommended follow-up actions |
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.