Ro-Ro and Vehicle Carrier Deck Load Monitoring
Satellite imagery can read the loading state of roll-on/roll-off and pure car carriers from freeboard depth and visible deck density, giving automotive supply-chain analysts and sanctions investigators an independent check on declared cargo.
Sensors
- Airbus Pleiades Neo: 30 cm panchromatic resolution at nadir; sufficient to distinguish individual passenger cars on open upper decks and to resolve the waterline against a painted hull with enough precision for freeboard estimation to within roughly 0.3 m under good contrast conditions. Revisit to any point is typically 1 to 2 days with the four-satellite constellation.
- Planet SkySat: 50 cm native resolution in collect mode; resolves vehicle outlines on open decks and gives a useful freeboard reading on high-sided hulls. The constellation supports rapid retasking, making it practical for tracking a vessel across multiple port calls within a single voyage.
- Maxar WorldView-3: 31 cm panchromatic, 1.24 m multispectral. The multispectral bands add colour discrimination useful for vehicle-type sorting on open decks. Archive depth extends to 2014, supporting retrospective voyage reconstruction.
- Sentinel-1 SAR (C-band, ESA): IW mode ground range resolution approximately 5 m x 20 m; insufficient for vehicle counting but confirms vessel presence, heading and approximate length through cloud and at night. Interferometric Wide Swath imagery is freely available with a 6-day repeat at mid-latitudes, falling to 12 days near the equator with a single satellite.
What a high-sided hull gives away
Pure car carriers (PCCs) and roll-on/roll-off vessels (Ro-Ros) are among the most geometrically distinctive ships afloat. Their hulls are unusually tall relative to beam, because internal decks are stacked to maximise cubic volume rather than deadweight. A fully loaded PCC sits noticeably lower in the water than an empty one. The difference in freeboard between a light ship and a full load can exceed three metres on the largest vessels, such as the Höegh Autoliners Aurora class, which carry upwards of 9,000 car-equivalent units.
From a sub-metre optical satellite, the painted freeboard band is clearly visible against the waterline. Measuring its pixel height, correcting for vessel trim and the known geometry of the image acquisition angle, gives a load-state estimate that is independent of any manifest or AIS declaration. The method is the same as for bulk carriers, but the high-sided hull geometry actually makes the freeboard band easier to measure because it presents a large, flat, often light-coloured surface with a sharp waterline boundary.
Counting cars from orbit
Many PCCs carry vehicles on exposed upper decks, typically the weather deck and sometimes one or two open-sided decks immediately below. At 30 to 50 cm resolution, a passenger car occupies roughly 8 to 15 pixels in plan view. Rows are regular. That regularity is the detection signal: a simple grid-fitting algorithm can estimate occupancy on exposed sections with reasonable confidence when image quality is good.
The honest limits are significant. Only the topmost exposed decks are visible. Enclosed lower decks, which hold the majority of cargo on most vessels, are entirely invisible to optical sensors. A vessel with a lightly loaded upper deck may still carry a full complement below. Deck-count data is therefore a floor estimate, not a total. It is most useful as a corroborating signal alongside freeboard, not as a standalone figure.
Shadow geometry also matters. A vessel oriented with its beam to the sun casts shadows from vehicles that can help confirm their presence but also obscure the row behind. Analysts should prefer imagery acquired when the sun angle exceeds 45 degrees and the vessel is roughly broadside to the sensor.
Why automotive supply chains care
The finished-vehicle logistics sector moves roughly 30 million units annually by sea. Production delays, port congestion and allocation disputes between manufacturers and shipping lines create real demand for independent load data. A plant manager waiting on a vessel carrying 2,000 units of a specific model has little recourse if the carrier's position and load status are opaque.
Satellite freeboard monitoring gives a rough but timely signal of whether a vessel departed a loading port full or partially loaded. Combined with AIS port-call history, it can indicate whether a vessel made an unscheduled intermediate stop that might explain a delivery shortfall. The analysis does not require access to proprietary shipping records, which makes it useful in commercial disputes where one party controls the manifest data.
Sanctions screening: the vehicle export problem
Sanctions regimes targeting specific countries frequently restrict vehicle exports. Enforcement depends on knowing which vessels carried vehicles from which loading ports to which destinations. AIS records can be manipulated. Manifests can be falsified. Freeboard imagery cannot easily be faked.
A vessel that loads at a port in a sanctioned jurisdiction and departs sitting measurably lower than its last documented empty state has, with high probability, taken on cargo. If open-deck vehicles are visible and their general type can be inferred from size and shape, the case for a reportable event strengthens considerably. This is not courtroom evidence on its own, but it is the kind of independent corroboration that compliance teams and government investigators need to justify further action.
The practical workflow involves tasking high-resolution optical imagery at the loading port before and after the vessel's call, then again at the suspected destination. Sentinel-1 fills the gaps at sea when cloud or darkness would otherwise leave the voyage unobserved.
Resolution floors, cloud and the revisit problem
Sub-metre tasking is not free and not guaranteed. Commercial satellites are in high demand, and a 24-hour tasking window over a specific anchorage may not always be achievable. Cloud cover over many major vehicle-loading ports, particularly in East Asia during the summer monsoon, can block optical collection for days at a time. Sentinel-1 SAR confirms the vessel is present but cannot read the freeboard with the precision needed for load-state discrimination on a PCC hull.
Freeboard measurement accuracy degrades with image obliquity. Collections at off-nadir angles beyond about 25 degrees introduce perspective distortion that complicates waterline identification. Analysts should specify near-nadir tasking where possible, accepting a slightly longer wait for a geometrically clean collect.
Archive depth helps. WorldView-3 imagery back to 2014 allows retrospective reconstruction of a vessel's loading history across multiple voyages, which is often more persuasive in a sanctions investigation than a single observation.
Putting the analysis together
A practical monitoring programme combines three data streams: routine Sentinel-1 passes for vessel presence and position confirmation, commercial optical tasking at key loading and discharge ports, and AIS history to provide the voyage narrative into which imagery observations are inserted.
Satellize structures this kind of multi-source fusion for government and commercial clients, drawing on the same analytical architecture used in its Tonga crop-estimation programme, where independent physical measurements from satellite data are used to verify conditions that ground records cannot fully capture.
The output is not a precise unit count. It is a load-state classification, a confidence level and a flag where the observed state conflicts materially with declared information. That is enough to direct human investigators to the right vessels, which is the point.
Typical figures
| Best optical resolution (panchromatic) | 30 cm (Pleiades Neo), 31 cm (WorldView-3), 50 cm (SkySat) |
| Freeboard measurement precision (optical, near-nadir) | Approximately ±0.3 m under good contrast; degrades beyond 25° off-nadir |
| SAR resolution (Sentinel-1 IW mode) | ~5 m range × ~20 m azimuth; vessel detection only, not freeboard |
| Optical revisit (commercial tasking) | 1 to 2 days typical (Pleiades Neo 4-sat); 24 to 48 hours (SkySat on demand) |
| SAR revisit (Sentinel-1, single satellite) | 6 days at mid-latitudes; 12 days near equator |
| Minimum detectable vehicle (open deck) | Passenger car (~4.5 m × 1.8 m) detectable at 30–50 cm resolution; type discrimination limited |
| Spectral bands (WorldView-3 multispectral) | 8 bands, 400–1040 nm; supports colour-based vehicle-type discrimination |
| Archive depth | WorldView-3 from 2014; Sentinel-1 from 2014; Pleiades Neo from 2021 |
| Latency from tasking to delivery | Typically 2 to 6 hours after acquisition for commercial optical; Sentinel-1 NRT ~3 hours |
| Delivery formats | GeoTIFF, KMZ, GeoJSON annotations, PDF intelligence report |
Analytics Satellize can run
| Freeboard load-state classification | Waterline pixel extraction and hull-height ratio measurement from sub-metre optical imagery; corrected for trim and acquisition geometry | Per-vessel load-state label (light / partial / full) with confidence score, delivered as structured report or GIS attribute |
| Open-deck vehicle density estimate | Grid-fitting object detection on exposed upper decks using 30–50 cm panchromatic imagery; manual QA on ambiguous rows | Estimated visible vehicle count with uncertainty range and annotated image chip |
| Port-call load-change delta | Before/after freeboard comparison across a single port call, fused with AIS arrival and departure timestamps | Load-change report indicating probable cargo volume taken on or discharged, flagged if inconsistent with declared port function |
| Sanctions-screening vessel alert | Freeboard and deck-count observation cross-referenced against vessel flag, operator, route and designated-port watchlist | Compliance alert with supporting imagery, vessel metadata and narrative summary; formatted for submission to legal or government teams |
| Voyage load-state timeline | Multi-image time series assembled from archive and tasked collections across successive port calls; AIS-anchored narrative | Chronological GIS layer and PDF report showing load state at each observed port call over a defined voyage or date range |
| SAR presence confirmation layer | Sentinel-1 CFAR ship detection on IW GRD scenes; matched to AIS track for dark-ship flagging | GeoJSON vessel detection layer with AIS correlation status, updated per Sentinel-1 pass |
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.