Naval vessel monitoring in foreign ports and anchorages
High-resolution optical and SAR satellites can identify individual naval vessels by hull class, pennant number, and equipment configuration. This page explains what is detectable, at what resolution, and where the method breaks down.
Sensors
- Maxar WorldView-3: 30 cm panchromatic resolution, 1.24 m multispectral. The sharpest commercially available optical sensor for reading pennant numbers and distinguishing hull-mounted weapons fits. Single-pass swath of 13.1 km; revisit at mid-latitudes roughly 1 to 4.5 days depending on tasking priority and off-nadir angle.
- Airbus Pléiades Neo: 30 cm native optical resolution across a 14 km swath, with a constellation of two satellites giving a revisit of under 24 hours for most latitudes. Useful for detecting configuration changes between passes: cranes deployed, boats in the water, or a vessel that has shifted berth.
- Capella Space SAR (X-band): Spotlight mode delivers down to 50 cm resolution, all-weather, day or night. SAR backscatter distinguishes large metallic superstructures, antenna arrays, and deck fittings even through cloud. Cannot read alphanumeric pennant numbers directly but can confirm vessel presence and approximate class when optical is unavailable.
- Planet SkySat: 50 cm optical resolution, constellation of 21 satellites. Lower per-image sharpness than WorldView-3 or Pléiades Neo, but higher revisit frequency allows detection of short-duration activity such as ammunition loading or boat deployment that a less frequent tasker would miss.
What a berth assignment actually tells you
Naval ports are organised by function. Submarine tenders berth alongside submarine pens or floating dry-docks; amphibious assault ships require deep-water quays with vehicle ramps or well-deck access; surface combatants cluster near fuel and ordnance piers. When a vessel moves from its routine berth to one of these specialised facilities, the shift is analytically significant before a single weapon is loaded.
High-resolution imagery makes these distinctions tractable. At 30 cm resolution, a WorldView-3 image of a destroyer shows not just the hull outline but the arrangement of vertical launch system hatches, the radar mast configuration, and whether the helicopter hangar doors are open. These details, cross-referenced against published naval recognition guides and open-source vessel registries, allow class identification with reasonable confidence for most major surface combatants.
Reading hull markings from orbit
Pennant numbers on naval vessels are typically painted in characters 1 to 2 metres tall on the bow and, for larger ships, the stern. At 30 cm optical resolution, a two-digit pennant number subtends roughly 4 to 6 pixels per character, which is enough for a trained analyst to read directly or for an automated template-matching algorithm to score against a known fleet inventory. Three-digit numbers are somewhat harder but still achievable at nadir or low off-nadir angles.
Lighting matters considerably. A vessel moored with its bow in shadow, or imaged at a steep off-nadir angle that foreshortens the hull, can defeat even the best optical sensor. Tasking strategy therefore includes requesting multiple passes at different times of day when a port is under persistent watch, and combining panchromatic sharpness with multispectral colour information to separate hull grey from water and quay concrete.
Equipment configuration as a readiness indicator
The presence or absence of specific deck equipment is often more informative than hull identity alone. Cranes extended over a vessel's side indicate loading activity. Boats deployed in the water suggest crew transfer or security patrols. Radar arrays in a stowed versus operational position, canvas covers removed from gun mounts, and gangways rigged or withdrawn all carry meaning when tracked across a sequence of images.
Change detection between two registered images of the same berth is a well-established method. Pixel-level differencing, after co-registration to sub-pixel accuracy, highlights any physical change larger than roughly one to two metres at 30 cm resolution. This is sufficient to detect a new crane position, a vehicle on the flight deck, or a boat that was not there yesterday. The method is computationally inexpensive and can be automated to generate alerts when a threshold area of change is exceeded.
Where cloud and geometry defeat optical sensors
Optical sensors are useless through cloud. Ports in tropical or high-latitude regions can be obscured for days or weeks at a time, and a vessel can arrive, load, and depart within a cloud-covered window that leaves no optical record. This is not a minor caveat; it is a structural limitation that any honest monitoring programme must plan around.
SAR fills part of this gap. Capella's X-band spotlight mode at 50 cm resolution cannot read a pennant number, but it can confirm whether a large metallic vessel is present at a berth, estimate its length to within a few metres from the radar shadow, and detect significant superstructure changes between passes. Combined optical-SAR tasking, with SAR providing continuity and optical providing identity when skies clear, is the standard approach for persistent port monitoring. Even so, a vessel that moves during a prolonged overcast period may be difficult to track without corroborating signals intelligence or AIS data, which are covered in separate pages in this library.
Archive depth and pattern-of-life analysis
A single image of a warship in a foreign port is a data point. A year of images of the same port is a pattern-of-life baseline. Commercial archives now extend back to 2008 for Planet imagery and to 2014 for WorldView-3. Sentinel-1 SAR data from 2014 onwards is freely available through the Copernicus Data Space. These archives allow an analyst to establish what is normal for a given facility: which berths are routinely occupied, how often vessels rotate, and what seasonal maintenance cycles look like.
Departures from that baseline are the analytic product. An unusual vessel class appearing at a port it has never previously visited, a berth occupied for longer than historical norms, or a sudden increase in support vessel activity around a submarine tender are all detectable against a well-constructed baseline. Satellize structures this kind of persistent-watch tasking for government clients who need a documented, auditable record of port activity over time, similar in analytical logic to the crop-state time series it runs for the Kingdom of Tonga, adapted here for a very different observational target.
Honest limits of the method
Resolution floors matter. At 50 cm, class identification is feasible for vessels above roughly 100 metres in length; smaller patrol craft become ambiguous. Pennant number reading at 50 cm is unreliable. SAR cannot substitute for optical on identification tasks. Even at 30 cm, a vessel with fresh grey paint over its pennant number, or one moored stern-to with its bow obscured by a pier structure, may resist positive identification.
Revisit is the other constraint. Even with aggressive multi-constellation tasking, a port can realistically be imaged once or twice per day in good conditions. A vessel that arrives and departs within a six-hour window may leave no imaged record. Night-time optical is possible with very long exposure times but is operationally impractical for moving targets; SAR has no such limitation and is the preferred sensor for overnight coverage. Analysts should treat any port-monitoring product as a probabilistic record rather than a complete one.
Typical figures
| Best optical resolution (commercial) | 30 cm panchromatic (WorldView-3, Pléiades Neo) |
| Best SAR resolution (commercial) | 50 cm spotlight mode (Capella Space X-band) |
| Optical revisit (tasked) | Less than 24 hours with multi-constellation tasking; 1 to 4.5 days for single-sensor WorldView-3 |
| SAR revisit | Capella constellation: multiple passes per day at most latitudes in tasked mode |
| Minimum vessel length for class identification | Approximately 80 to 100 m at 50 cm resolution; smaller vessels require 30 cm optical |
| Pennant number readability threshold | Characters of 1 m height or larger at 30 cm resolution, nadir or near-nadir geometry |
| Archive depth (optical) | WorldView-3 from 2014; Planet from 2009 (lower resolution) to 2017 (SkySat) |
| Archive depth (SAR, open access) | Sentinel-1 from 2014, freely available via Copernicus Data Space |
| Tasking latency (image to delivery) | Typically 2 to 6 hours for priority commercial tasking; same-day feasible |
| Cloud penetration | SAR: full; optical: none |
Analytics Satellize can run
| Vessel presence and class report | Object detection and template matching against known hull silhouettes; co-registered optical and SAR imagery | Structured report per tasking event: vessel count, probable class, berth location, confidence score |
| Pennant number extraction | Optical character recognition on panchromatic imagery at 30 cm, validated by analyst against published fleet registries | Annotated image chip with extracted identifier and confidence rating; flagged for analyst review when score is below threshold |
| Configuration change alert | Sub-pixel co-registration followed by pixel-difference change detection between sequential images of the same berth | Automated alert with change-mask overlay and before/after image pair, delivered within hours of new image ingestion |
| Pattern-of-life baseline and anomaly scoring | Time-series analysis of vessel occupancy, berth assignment, and support activity across archive imagery; statistical anomaly scoring against rolling baseline | Monthly port activity summary with anomaly flags; GIS layer of berth occupancy history |
| Readiness indicator assessment | Structured analyst review of deck equipment state, crane positions, boat deployment, and gangway status across sequential images | Readiness state rating (routine, elevated, uncertain) per vessel per observation, with supporting image evidence |
| Cloud-gap SAR continuity record | Automated SAR vessel detection using constant false alarm rate (CFAR) processing on Capella or Sentinel-1 imagery during optical blackout periods | Continuous presence/absence log with SAR-derived length estimate; flagged gaps where neither sensor achieved coverage |
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.