Container Ship Stack Height and Utilisation Estimation
High-resolution optical imagery from WorldView-3 and Pleiades Neo can resolve individual container tiers above the hatch coaming, giving a visible proxy for vessel utilisation. Below-deck stowage is invisible to any optical sensor, so estimates are partial by definition.
Sensors
- Maxar WorldView-3: 31 cm panchromatic resolution at nadir; 1.24 m multispectral. At this resolution individual container tiers (each roughly 2.6 m tall) cast measurable shadows and present distinct roof edges, making manual or algorithmic tier counting feasible on vessels longer than about 200 m. Tasking revisit is 1 day or less at mid-latitudes with off-nadir collection.
- Airbus Pleiades Neo: 30 cm panchromatic, 1.2 m multispectral, with a 4-satellite constellation offering sub-daily revisit over priority areas. Radiometric quality is comparable to WorldView-3 for shadow-depth analysis. Stereo acquisition is available and adds a height-estimation cross-check independent of shadow geometry.
- Planet SkySat: 50 cm panchromatic, 1 m multispectral. Sits at the resolution margin for tier counting; individual tiers on a well-lit, stationary vessel are sometimes distinguishable, but motion blur during transit and lower signal-to-noise ratio make it less reliable than WorldView-3 or Pleiades Neo. Useful for confirmation tasking when cost is a constraint.
- Airbus SPOT-7: 1.5 m panchromatic. Insufficient for individual tier counting on standard 20- or 40-foot containers. Useful for vessel detection and gross load-state classification (clearly empty vs. clearly full) but not for the tier-by-tier counting this method requires.
What the deck actually reveals
A cellular container ship loads containers both below deck (in cell guides) and above deck in stacks rising from the hatch coaming. The above-deck portion is the only part visible to any optical satellite. On a large vessel such as a 14,000-TEU post-Panamax ship, the above-deck capacity can represent roughly 40 to 60 per cent of total TEU, though the split varies substantially by vessel class and operator loading practice. That means a fully stacked deck does not guarantee a full hold, and a lightly stacked deck does not prove the hold is empty.
Within that constraint, the above-deck stack height is a genuine economic signal. Operators load heavy cargo low and light cargo high; a vessel with six or seven tiers showing above the coaming is almost certainly carrying a substantial cargo. Vessels repositioning empty or running with minimal cargo typically show one or two tiers, or bare hatch covers. The pattern across a fleet, or across a port's departing vessels over weeks, carries real intelligence value even when no individual vessel reading is definitive.
How tier counting works in practice
Each standard container is 2.591 m tall (ISO 668). At WorldView-3's 31 cm nadir resolution, a single tier subtends roughly eight to nine pixels in height. That is enough for a human analyst to count tiers directly from the panchromatic image, using the shadow cast by each tier's upper edge as the discriminating feature. Automated methods apply edge detection along the vessel's longitudinal axis, then count intensity transitions that correspond to tier boundaries.
Shadow-based counting requires a sun elevation angle above roughly 20 degrees. Very low sun angles cause tier shadows to merge; very high angles (above about 70 degrees) compress shadows to near zero and reduce contrast between tiers. Collection geometry therefore matters. Off-nadir angles beyond about 25 degrees introduce perspective foreshortening that complicates counting on the inboard stacks. Tasking requests for this application should specify nadir or near-nadir collection and a sun elevation window of 30 to 60 degrees where operationally possible.
Stereo collection from Pleiades Neo offers an independent route: photogrammetric reconstruction of above-deck stack height in metres, divided by 2.591 m per tier, gives a tier count without relying on shadow geometry. This is more reliable in high-latitude winter operations where sun elevation is chronically low.
The below-deck problem, stated plainly
No passive optical sensor can see through steel hatch covers. SAR cannot either, at any commercially available frequency. The below-deck stowage is simply not accessible from orbit with current technology. Any utilisation estimate derived from stack height is therefore a partial estimate, and should be labelled as such in any report or data feed.
The practical implication: this method works best as a relative indicator rather than an absolute one. Comparing the same vessel on departure from port A versus departure from port B, or comparing a fleet of vessels on the same trade lane over time, reveals trends that are meaningful even when the absolute TEU count is unknown. Combining above-deck observations with AIS-derived port call data and published vessel capacity figures narrows the uncertainty range considerably, though it does not eliminate it.
Port throughput and trade-flow applications
The commercial use case is not usually a single vessel reading. It is a time series across a port or a trade lane. If ten vessels depart Shanghai on a given week with an average of five above-deck tiers, and the same vessels departed with an average of three tiers six months earlier, that is a directional signal about export demand that precedes official trade statistics by weeks.
Ports where vessel queues and dwell times are already being monitored (a separate method not covered here) can be cross-referenced with stack-height observations on departing vessels to build a more complete picture of throughput. A port that is congested and sending out lightly loaded ships is in a different situation from one that is congested and sending out heavily loaded ships. Stack height is the variable that distinguishes the two cases from orbit.
Satellize applies this kind of multi-source layering in its analytics work. The approach is similar in logic to the crop-area and yield estimation the team runs for the Kingdom of Tonga, where no single observation is sufficient and the value comes from combining several partial signals into a calibrated estimate.
Resolution floor and why Sentinel-2 does not qualify
Sentinel-2's finest spatial resolution is 10 m in the visible bands. A standard container is 2.4 m wide and 2.6 m tall. At 10 m per pixel, the entire above-deck stack of a large vessel occupies perhaps 20 to 30 pixels in total, with no possibility of resolving individual tiers. Sentinel-2 is excellent for many maritime applications, including turbidity mapping, algal bloom detection, and coarse vessel detection in clear water. Tier counting is not among them.
Planet's SuperDove constellation, at 3 to 5 m resolution, falls in the same category. It can confirm a vessel is present and give a rough sense of whether the deck appears loaded or bare, but individual tier boundaries are below its resolving limit. The resolution floor for this specific method is approximately 50 cm panchromatic, and performance improves materially below 35 cm. That currently means WorldView-3 or Pleiades Neo for reliable results, with SkySat as a lower-confidence option.
Delivery, latency and archive depth
Commercial tasking from WorldView-3 or Pleiades Neo typically delivers imagery within 24 to 48 hours of collection for standard orders, with expedited pipelines available. Archive depth for both sensors extends back to 2014 (WorldView-3 launched September 2014; Pleiades Neo from 2021, with Pleiades 1A/1B archive from 2012). Historical stack-height analysis is therefore possible for vessels and ports of interest, subject to archive availability at the required collection geometry.
Processed outputs from a stack-height analysis can be delivered as a structured data table (vessel identifier, collection date and time, GPS position from AIS correlation, tier count per bay section, estimated above-deck TEU range) or as an annotated image with tier boundaries marked. GIS polygon layers showing vessel position and load-state classification are a common deliverable for port-monitoring dashboards.
Typical figures
| Minimum useful panchromatic resolution | ≤50 cm; ≤35 cm preferred for reliable automated tier counting |
| Primary sensors | Maxar WorldView-3 (31 cm), Airbus Pleiades Neo (30 cm) |
| Tasking revisit (WorldView-3) | ≤1 day at mid-latitudes with off-nadir collection accepted |
| Tasking revisit (Pleiades Neo constellation) | Sub-daily over priority areas with 4-satellite constellation |
| Optimal sun elevation for shadow-based counting | 30° to 60°; below 20° tiers become unreliable |
| Minimum vessel length for reliable analysis | Approximately 200 m (post-Panamax and larger vessels) |
| Tier height (ISO standard container) | 2.591 m per tier; resolves to ~8 pixels at 31 cm resolution |
| Archive depth | WorldView-3 from September 2014; Pleiades 1A/1B from 2012 |
| Imagery delivery latency (standard tasking) | 24–48 hours post-collection; expedited pipelines available |
| Below-deck stowage visibility | None; optical and SAR sensors cannot penetrate steel hatch covers |
Analytics Satellize can run
| Per-vessel above-deck tier count | Manual or semi-automated edge detection on panchromatic imagery; shadow-depth analysis or stereo photogrammetry | Structured data table: vessel ID, date/time, position, tier count per bay section, estimated above-deck TEU range |
| Vessel utilisation classification | Tier count normalised against published vessel above-deck capacity; classified as empty, partial or heavily loaded | Per-vessel load-state flag in CSV or GeoJSON, compatible with AIS feed integration |
| Port departure load-state time series | Repeated tasking of departing vessels correlated with AIS port-call records; aggregated by port and trade lane | Weekly or monthly trend report with confidence intervals; chart-ready data tables |
| Fleet utilisation monitoring | Systematic tasking of a named fleet across multiple port calls; longitudinal comparison of above-deck tier counts | Fleet-level utilisation dashboard layer; anomaly alerts when observed load state deviates from historical baseline |
| Trade-lane demand indicator | Aggregated above-deck TEU estimates on vessels departing a port or region, compared across time periods | Monthly intelligence brief with directional demand signal; uncertainty range stated explicitly |
| Annotated imagery archive | Analyst-reviewed imagery with tier boundaries and bay labels marked; quality-controlled against stereo height where available | Georeferenced TIFF with annotation layer; suitable for audit trail or regulatory reporting |
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.