Tanker loading and discharge event detection from draught proxies
SAR imagery can infer whether a tanker at berth is loading or discharging by tracking the vessel's waterline position between arrival and departure passes. The method is most reliable for VLCCs and Suezmax hulls, where the freeboard change is large enough to exceed sensor resolution floors.
Sensors
- ICEYE SAR constellation: Stripmap mode delivers approximately 3 m ground resolution; Spot mode reaches around 1 m. Sub-daily revisit at many latitudes. The fine resolution allows waterline detection on large tanker hulls with reasonable confidence, though hull curvature and sea state introduce noise.
- Capella Space SAR: Spotlight mode achieves roughly 0.5 m resolution, which is the sharpest commercially available SAR for this application. High-incidence acquisitions are particularly useful for reading freeboard height from hull backscatter gradients.
- Sentinel-1 SAR (ESA): Interferometric Wide Swath mode at 10 m resolution is too coarse for reliable waterline detection on most tankers, but the free archive (back to 2014 in many regions) supports retrospective baseline studies and is adequate for VLCC-class vessels where freeboard change can exceed 10 m between ballast and laden states.
- AIS draught records: Draught is a manually entered, self-reported field in AIS Class A transmissions. Published studies have found systematic under-reporting and stale values that persist across multiple port calls. Used here as a cross-check, not a ground truth.
What the waterline gives away
A laden VLCC sits roughly 20 to 22 metres below the waterline; the same vessel in ballast sits perhaps 10 to 12 metres down. That difference, expressed as freeboard height above the water surface, is the physical signal this method reads. When a SAR satellite images a tanker at berth, the intensity of radar backscatter along the hull changes with the angle between the sensor, the hull plating, and the water surface. A high-freeboard, ballast-condition hull presents a different backscatter signature than a low-freeboard, laden hull. By comparing two acquisitions, one at arrival and one near departure, the direction of draught change can be inferred.
The logic is straightforward. A rising waterline (decreasing freeboard) between arrival and departure means mass was added: the vessel loaded. A falling waterline means mass left: it discharged. No cargo manifest, no port declaration, and no co-operative AIS update is required. The satellite does not care whether the operator has remembered to update the draught field.
Why AIS draught records are not enough
AIS Class A transponders require masters to enter static voyage data including draught, but the field is updated manually and enforcement is uneven. Multiple independent analyses of AIS datasets have found that a significant proportion of tanker draught records are either unchanged from a previous voyage, rounded to implausible integers, or missing entirely. For sanctions-monitoring or cargo-flow intelligence purposes, a self-reported figure from a motivated party is a weak foundation.
SAR-derived waterline position is not a replacement for a direct draught measurement, but it is independent. It cannot be falsified by a crew member who forgets to update a transponder. That independence is precisely its value, particularly when combined with AIS to flag discrepancies between reported and observed draught state.
Resolution floors and the vessel-size constraint
The technique has a hard physical limit: the freeboard change between ballast and laden states must exceed the effective ground resolution of the sensor for the waterline shift to be detectable. For Sentinel-1 at 10 m, only the largest vessel classes, VLCCs and some Suezmax tankers, produce a freeboard delta reliably above that floor. ICEYE Stripmap at 3 m extends the method to Aframax-class vessels in principle, though sea state and mooring geometry add uncertainty. Capella's sub-metre spotlight mode is the most capable for this application, but tasking cost and revisit scheduling mean it is not practical as a fleet-wide continuous monitor.
Sea state matters too. Wave action at an exposed anchorage or single-point mooring introduces apparent waterline motion that is not draught-related. Sheltered berths in enclosed port basins produce cleaner results. Incidence angle also affects how the hull backscatter gradient reads: steeper angles (higher incidence) tend to give better freeboard sensitivity on vertical hull sides. These are not reasons to abandon the method; they are parameters to manage in the acquisition planning.
Fusing SAR with AIS for event classification
A complete detection pipeline combines three data streams. First, AIS provides vessel identity, berth arrival and departure timestamps, and the self-reported draught values for comparison. Second, SAR acquisitions bracketing the port call provide the observed waterline position at two points in time. Third, port call records and terminal-type metadata (crude export terminal versus refinery import jetty) provide prior probability that conditions the interpretation.
When SAR shows a rising waterline at a known crude export terminal and AIS draught is either missing or unchanged, the inference is loading with high confidence. When the waterline falls at a refinery import berth and AIS draught is stale, the inference is discharge. Ambiguous cases, where the SAR delta is below the resolution floor or only one acquisition is available, are flagged as unclassified rather than forced into a category. Honest uncertainty labelling is more useful to a downstream analyst than a spurious confidence score.
What the method cannot tell you
Cargo volume is not directly recoverable from this approach. Freeboard change is proportional to mass change, and converting mass to volume requires knowing the cargo density. Crude oil grades range from roughly 790 to 970 kg per cubic metre; without knowing the specific grade, volume estimates carry a meaningful error band. The method produces a directional event classification (loaded or discharged) and a rough magnitude category (partial versus near-full), not a precise barrel count.
It also cannot distinguish between a vessel that loaded crude and one that loaded refined product, or between a legitimate cargo and a ship-to-ship transfer that occurred before the berth call. For those distinctions, additional data streams are needed. Satellize's analytics stack can incorporate TROPOMI atmospheric data and nighttime thermal signals where relevant, though those are separate analytic products. The crop-estimation programme the company runs for the Kingdom of Tonga uses a similar multi-source inference logic, which is worth noting for clients who want to understand how the methodology generalises.
Practical applications and tasking strategy
The most common clients for this analytic are commodity traders tracking crude flows through specific export terminals, compliance teams monitoring vessels subject to sanctions or cargo restrictions, and port authorities wanting an independent check on declared cargo operations. For a trader, the question is usually directional: is this terminal loading or idle this week? For a compliance team, the question is whether the declared cargo type matches the observed draught behaviour.
A practical tasking strategy pairs persistent Sentinel-1 coverage (free, 6-day repeat at most latitudes, adequate for VLCC detection) with opportunistic Capella or ICEYE tasking triggered by AIS anomalies, such as a vessel with a stale draught field completing an unusually long port call. That combination keeps cost proportionate to the intelligence value of each event. Archive depth for Sentinel-1 runs back to 2014 in most regions, which allows retrospective analysis of historical port call patterns at terminals of interest.
Typical figures
| SAR spatial resolution (commercial) | 0.5 m (Capella Spotlight) to 3 m (ICEYE Stripmap) |
| SAR spatial resolution (open) | 10 m (Sentinel-1 IW mode) |
| Revisit (commercial tasking) | Sub-daily at many latitudes with ICEYE or Capella; depends on constellation scheduling |
| Revisit (Sentinel-1 open) | 6-day repeat at equator; 1 to 3 days at higher latitudes with both satellites |
| Radar frequency | C-band (5.4 GHz, Sentinel-1); X-band (9.6 GHz, ICEYE and Capella). X-band gives finer resolution; C-band penetrates light rain better |
| Minimum vessel size for reliable detection | VLCC and Suezmax (Sentinel-1); Aframax and above (ICEYE Stripmap); Panamax possible with Capella Spotlight under calm sea state |
| Minimum detectable freeboard change | Approximately 3 to 5 m with ICEYE Stripmap; approximately 1 to 2 m with Capella Spotlight; greater than 10 m required for Sentinel-1 |
| Archive depth | Sentinel-1: 2014 to present in most regions. Commercial SAR: varies by operator, typically 2018 onwards |
| Latency from acquisition to analytic output | Typically 2 to 6 hours for commercial SAR with priority processing; 12 to 24 hours for Sentinel-1 standard pipeline |
| Delivery formats | GeoJSON event record, CSV port-call log with loading/discharge flag, GeoTIFF waterline-change layer, API feed |
Analytics Satellize can run
| Loading or discharge event classification | SAR waterline comparison between arrival and departure acquisitions; backscatter intensity gradient analysis along hull | Per-port-call event record (GeoJSON or CSV) with classification, confidence tier, and SAR acquisition timestamps |
| AIS draught discrepancy flag | Cross-comparison of AIS-reported draught fields against SAR-inferred freeboard state; statistical outlier scoring | Alert feed flagging vessels where reported and observed draught state are inconsistent beyond a configurable threshold |
| Terminal activity index | Aggregated loading and discharge event counts per terminal per week, derived from repeated SAR passes and AIS port call records | Weekly time-series report (PDF and CSV) showing terminal throughput direction and approximate utilisation level |
| Retrospective cargo flow reconstruction | Sentinel-1 archive analysis combined with historical AIS draught and port call data; batch processing over user-defined date range and terminal list | Historical GIS layer and tabular dataset of inferred loading and discharge events; suitable for trade flow modelling |
| Anomalous port call alert | AIS-triggered tasking: long dwell time combined with stale draught field triggers commercial SAR acquisition and rapid waterline analysis | Rapid-turnaround alert report within 6 hours of SAR acquisition, naming vessel, berth, inferred event type, and discrepancy summary |
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.