Sensors
- Maxar WorldView-3: 0.31 m panchromatic resolution; the finest commercially available optical detail for measuring freeboard strips and reading hull markings such as load-line indicators directly from imagery.
- Airbus Pléiades Neo: 0.30 m native resolution, daily revisit over any point globally with the four-satellite constellation; stereo and tri-stereo collection modes support photogrammetric height reconstruction of hull geometry.
- Planet SkySat: 0.50 m resolution, rapid revisit and flexible tasking; useful for monitoring anchorages at scale where fleet-wide load-status screening matters more than single-vessel precision.
- Capella Space SAR (X-band): Spotlight mode delivers roughly 0.35 m resolution; cloud-independent and day/night capable, with backscatter contrast between the waterline and hull side providing freeboard cues even without visible-light colour contrast.
What a floating hull gives away
A Very Large Crude Carrier carries up to around 320,000 deadweight tonnes of oil. When fully laden, its hull sits so deep that the visible freeboard, the painted steel above the waterline, shrinks to roughly two metres or less. In ballast, the same ship rides high enough to expose six to nine metres of hull side, including the distinctive red or dark anti-fouling paint of the underwater section. That colour contrast, combined with the sheer vertical extent of exposed steel, is detectable in sub-metre optical imagery without any specialised instrument.
The physical principle is elementary displacement physics, which is why the method is reliable. A ship cannot misrepresent its draught the way it can misrepresent its AIS position. The hull is where it is, and the waterline is where physics puts it.
The photogrammetric method: rulers, registries and shadows
Estimating freeboard from a satellite image requires three inputs: the image itself, the vessel's known dimensions from a public registry such as Lloyd's or IHS Sea-web, and the satellite's viewing geometry at the moment of collection. With a known overall beam (beam widths for VLCCs typically run 55 to 68 metres), the pixel-to-metre scale of the hull can be calibrated directly from the imagery, since the beam is nearly always visible in a near-nadir pass. Once the scale is fixed, the freeboard strip is measured in pixels and converted to metres.
Shadow-length analysis offers a complementary route. At known solar elevation angles, the shadow cast by the hull onto the water surface has a calculable length proportional to the height of the casting edge. Combining shadow length with solar elevation, derived from the image timestamp and geographic coordinates, yields an independent freeboard estimate. The two methods cross-check each other. Neither requires ground truth beyond the vessel registry entry, which is public.
Load-line markings, the Plimsoll marks painted on the hull, are legible at 0.30 to 0.31 m resolution and provide a third reference. The marks indicate the maximum legal draught for various ocean zones and seasons. Identifying which mark sits at or near the waterline gives a direct categorical read of load status without any calculation.
Where the method breaks down
Oblique viewing angles are the primary source of error. WorldView-3 and Pléiades Neo can collect at off-nadir angles up to around 30 degrees to increase revisit frequency. At high off-nadir angles the hull face foreshortens, the shadow geometry becomes ambiguous, and freeboard measurement uncertainty widens from roughly 0.3 metres at nadir to one metre or more at 30 degrees off-nadir. For a Suezmax vessel where laden-versus-ballast freeboard difference is only three to four metres, that uncertainty is operationally significant.
Haze and thin cloud degrade contrast between hull and water. Coastal industrial haze near major loading terminals, the Persian Gulf in summer, the Strait of Malacca year-round, reduces the reliability of colour-based waterline detection. SAR fills part of this gap: X-band backscatter from the hull-water interface is cloud-independent, though the freeboard signal in SAR is subtler and requires careful speckle filtering.
Vessels moored alongside each other, or obscured by cranes and jetties at a terminal berth, cannot be measured at all until they move to anchorage or open water. This is not a minor edge case: the most commercially interesting moment, loading or discharging at a terminal, is often precisely when the vessel is occluded.
Connecting load status to commodity flows
A single freeboard observation tells you whether a ship is carrying cargo. A time-series of observations across a fleet, correlated with port calls and AIS voyage histories, tells you something more interesting: how much crude is moving, from where, and in which direction.
The canonical application is tracking crude exports from sanctioned or monitored loading terminals. If a VLCC departs a terminal appearing laden in imagery collected within hours of departure, and subsequently appears in ballast at a different anchorage weeks later, the cargo transfer is implied even if AIS was disabled for part of the voyage. The freeboard observation anchors the inference at each end of the voyage. It does not identify the cargo grade, the buyer, or the price, but it establishes that a transfer occurred and roughly when.
Floating storage analysis, covered separately in the sibling page on that topic, uses the same freeboard physics but applies it to vessels that are stationary for weeks at a time as a deliberate inventory strategy. The load-status method described here is most useful for vessels in transit or at anchorage for short periods.
Practical collection parameters and what to expect
For a reliable laden-versus-ballast classification on a VLCC, the working consensus in published remote-sensing literature is that imagery at 0.5 m or better, collected within 20 degrees of nadir, in good atmospheric conditions, yields classification accuracy in the high eighties to low nineties percent range. Smaller vessels, Aframax or Suezmax class, have smaller freeboard differentials and are correspondingly harder to classify; accuracy drops by roughly ten percentage points for the same imaging conditions.
Tasking latency from order to delivery runs four to twelve hours for priority commercial tasking on WorldView-3 or Pléiades Neo, depending on orbital geometry over the target. Archive imagery is available from approximately 2014 for WorldView-3 and from 2021 for Pléiades Neo. Planet SkySat archive depth runs from around 2017. For retrospective analysis of a specific anchorage or chokepoint, archive coverage is usually sufficient to reconstruct fleet load-status histories at weekly or better cadence.
Satellize runs freeboard-based load-status analysis as part of its maritime intelligence analytics layer, applying the photogrammetric pipeline to commercial tasking collected on client licence. Outputs are delivered as structured vessel-level records, not just imagery, so they feed directly into commodity-flow models or compliance workflows.
What a buyer should ask before commissioning this analysis
Three questions determine whether freeboard analysis will actually answer the intelligence question. First, is the vessel class large enough? For vessels below about 80,000 DWT, the laden-ballast freeboard difference may be under two metres, which is at the edge of reliable detection even with the best commercial optical systems. Second, how often does the vessel of interest transit open water or anchorages rather than terminal berths? If it spends most of its time alongside infrastructure, collection opportunities are limited. Third, what is the acceptable false-positive rate for the downstream decision? A commodity-flow model can tolerate occasional misclassifications that average out across a fleet. A sanctions-compliance determination cannot.
Typical figures
| Best available optical resolution | 0.30 m (Pléiades Neo) / 0.31 m (WorldView-3) |
| SAR resolution (spotlight mode) | ~0.35 m (Capella Space X-band) |
| Recommended max off-nadir angle | ≤20° for reliable freeboard measurement; uncertainty grows sharply beyond 25° |
| Freeboard detection floor | ~1.0 m strip width reliably measurable at 0.31 m GSD; practical classification threshold ~2 m difference between states |
| Tasking latency (priority) | 4–12 hours order-to-delivery for WorldView-3 and Pléiades Neo |
| Revisit (commercial tasking) | Daily or better for Pléiades Neo 4-satellite constellation; 1–4.5 days for WorldView-3 depending on latitude and off-nadir tolerance |
| Spectral bands used | Panchromatic (primary); multispectral RGB for waterline colour contrast; X-band SAR for cloud-penetrating fallback |
| Archive depth | WorldView-3 from ~2014; Pléiades Neo from ~2021; SkySat from ~2017 |
| Vessel size applicability | Most reliable for VLCC and Suezmax (>80,000 DWT); accuracy degrades for smaller classes |
| Delivery formats | Structured vessel-level JSON or CSV records; GeoTIFF chips; optional GIS layer |
Analytics Satellize can run
| Binary laden/ballast classification per vessel observation | Pixel-scale freeboard measurement against registry beam dimension; shadow-length cross-check using solar elevation from EXIF/metadata | Vessel-level structured record with classification, confidence score, and source image timestamp |
| Freeboard height estimate in metres | Photogrammetric scaling using known hull beam as ground reference; stereo DEM where Pléiades Neo stereo pairs are available | Per-observation freeboard value with uncertainty bound, delivered as CSV or JSON feed |
| Voyage-level cargo-transfer inference | Time-series of load-status observations correlated with AIS port-call history and voyage waypoints | Voyage summary report: departure state, arrival state, implied transfer event with date range |
| Fleet-level crude export volume index for a monitored terminal | Aggregated laden-departure counts and vessel class DWT from registry, summed over a calendar period | Monthly or weekly export-volume index (in DWT-equivalents, not barrels) delivered as time-series chart and underlying data table |
| Anchorage load-status snapshot for a named waypoint | Systematic tasking of a defined anchorage area; freeboard classification applied to all detected vessels above minimum size threshold | GIS layer of vessel positions with load-status attribute; refreshed on each successful collect |
| Historical load-status reconstruction from archive | Archive search across WorldView-3, Pléiades Neo and SkySat holdings for a named vessel MMSI or IMO number; freeboard analysis applied to all recovered images | Vessel history report with timeline of observed load states, gaps flagged where archive coverage is absent |
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.