Port and logistics zone land-use intensity monitoring
Very-high-resolution optical and SAR imagery turns port yards into quantified utilisation metrics, informing site-value assessments and feasibility studies for adjacent industrial land without a single site visit.
Sensors
- Sentinel-1 SAR (C-band, ESA): Interferometric Wide swath mode delivers 10 m ground range resolution with a 6-day repeat at mid-latitudes, cloud-independent. Backscatter contrast between full container stacks and empty concrete is detectable at this resolution, though individual container rows are not separable. Provides the all-weather baseline for yard-state change detection.
- Maxar WorldView-3: Panchromatic resolution of 0.31 m and multispectral at 1.24 m. At this scale, individual 2.4 m wide container rows are clearly resolved and colour classification is feasible. Revisit at a given point is typically 1 to 4.5 days depending on tasking priority and latitude. Cloud cover remains a constraint.
- Planet SkySat: 0.5 m panchromatic, 1 m multispectral, with the ability to task the same site multiple times per day. Revisit frequency is the key advantage for tracking intraday berth occupancy changes. Radiometric depth is shallower than WorldView-3, so fine colour discrimination of container markings is less reliable.
- Umbra SAR (X-band): Spotlight mode delivers 25 cm resolution SAR imagery, sufficient to resolve individual containers and distinguish stack height by shadow length. X-band penetrates cloud and operates at night. Commercial tasking latency is typically sub-24 hours. Archive depth is limited relative to Sentinel-1 given the constellation's recent launch cadence.
What a port yard actually reveals at 0.3 metres
A container terminal is one of the most legible industrial surfaces on Earth when viewed from sufficient resolution. Standard ISO containers are 2.438 m wide and either 6.1 m or 12.2 m long. At WorldView-3's 0.31 m panchromatic resolution, row geometry, stack shadow length and inter-row spacing are all measurable directly. Stack height can be inferred from shadow length given known solar elevation at acquisition time, a method published extensively in the remote sensing literature for urban building stock and directly applicable to stacked freight.
Colour classification adds a further layer. Multispectral imagery at 1.24 m can distinguish broad container colour groups, which correlates loosely with operator markings and dwell patterns, though it cannot read alphanumeric codes. The honest limit: fine-grained consignment identification is not possible from orbit. What is possible is a yard-wide count of occupied bays, an estimate of stack height distribution, and a binary classification of active versus idle yard zones.
SAR fills the gaps cloud leaves behind
Ports in tropical and temperate maritime climates spend a substantial fraction of the year under cloud. A monitoring programme that depends solely on optical tasking will have irregular, weather-biased coverage, which is precisely the wrong property for a time-series used in site-value analysis. Sentinel-1's C-band SAR acquires regardless of cloud or darkness, with a documented 6-day single-satellite repeat and a 12-day exact-repeat cycle for the constellation pair when both satellites are operational.
C-band backscatter from a dense container stack is high and relatively stable. Empty concrete yard produces lower, more specular return. The contrast is detectable at 10 m resolution, meaning Sentinel-1 can track gross yard utilisation state across a full terminal in a single acquisition. It cannot resolve individual container rows at this resolution, and SAR speckle introduces noise that requires multi-look processing or temporal averaging to suppress. Umbra's X-band spotlight mode at 25 cm changes the calculus considerably for sites where budget allows commercial SAR tasking: shadow-based stack-height estimation becomes feasible in SAR as well as optical.
Berth occupancy is a different measurement problem
Vessel presence at a berth is easier to detect than container stack density. A Panamax container ship is roughly 300 m long and 32 m wide, visible in Sentinel-1 at 10 m resolution and trivially resolved in WorldView-3 or SkySat imagery. The analytic task is classification: berth occupied, berth empty, vessel in approach or departure. With Planet SkySat's multiple-pass-per-day tasking, it is feasible to observe the same berth two or three times in a single day and construct an intraday occupancy record.
The complication is partial occupancy. A berth designed for a 350 m vessel may hold two smaller coastal feeders simultaneously, and automated detection models trained on large vessel signatures can misclassify or miss these. Ground-truthing against AIS transponder data, which is publicly aggregated by sources such as Global Fishing Watch's vessel tracking infrastructure, substantially reduces this ambiguity. AIS and imagery together are more reliable than either alone.
Turning utilisation into a land-value input
The connection between port throughput intensity and adjacent industrial land value is not automatic, but it is well-established in property economics. High and rising yard utilisation signals demand pressure that typically propagates outward: logistics operators seek overflow storage, last-mile distribution facilities and light industrial space within practical drayage distance of the terminal gate. Satellite-derived utilisation metrics give a property analyst a time-series of activity that is independent of operator-reported throughput figures, which are often delayed, aggregated or commercially sensitive.
A monthly utilisation index built from Sentinel-1 change detection, calibrated against periodic WorldView-3 spot checks, can cover a three-to-five year archive period for most major ports. Sentinel-1 data is freely available from the Copernicus Data Space, giving genuine historical depth. This time-series can be correlated against published transaction data for nearby industrial land parcels to test whether utilisation leads or lags price movements at a given site. That is a feasibility-study input, not a valuation in itself, and should be presented as such.
Where the method reaches its limits
Several honest constraints apply. First, resolution floors matter for smaller ports. A regional container terminal with 50 m berths and yard bays that are only partially filled may not produce enough backscatter contrast for reliable Sentinel-1 classification. WorldView-3 tasking resolves this but adds cost and cloud-cover risk. Second, temporal resolution is asymmetric: SAR gives frequent but coarse utilisation state; optical gives fine spatial detail but irregular temporal sampling. A combined programme is more expensive than either alone.
Third, port operators sometimes modify yard layouts, add temporary storage structures or change operating patterns in ways that invalidate a model trained on earlier imagery. Periodic retraining or visual validation is not optional. Fourth, this method measures surface activity. It does not capture what is inside containers, the financial terms of vessel calls, or the contractual relationships between terminal operators and shipping lines. It is an intensity proxy, not a throughput audit. Buyers of this analysis should understand that distinction before incorporating it into a valuation model.
Satellize runs port utilisation analytics as part of its broader land-use intelligence offering, applying the same open-constellation approach it uses in the Tonga crop-estimation programme to industrial and logistics precincts where clients need asset-level detail without relying on operator disclosure.
Typical figures
| Finest spatial resolution (optical) | 0.31 m panchromatic (WorldView-3); 0.5 m panchromatic (SkySat) |
| Finest spatial resolution (SAR) | 25 cm (Umbra spotlight); 10 m (Sentinel-1 IW mode) |
| All-weather revisit | 6 days single-satellite (Sentinel-1); sub-24 h tasking latency (Umbra, commercial) |
| Optical revisit at a given point | 1 to 4.5 days (WorldView-3, tasking-dependent); multiple passes per day possible (SkySat) |
| Spectral coverage | Panchromatic plus 8-band VNIR/SWIR (WorldView-3); C-band 5.4 GHz (Sentinel-1); X-band 9.6 GHz (Umbra) |
| Minimum detectable target (optical) | Individual ISO container row (2.4 m wide) at WorldView-3 resolution |
| Minimum detectable target (SAR) | Gross yard utilisation state at Sentinel-1; individual container at Umbra 25 cm |
| Archive depth | Sentinel-1: from 2014 (Copernicus Data Space); WorldView-3: from 2014; SkySat: from 2017 |
| Delivery formats | GeoTIFF orthorectified imagery, GeoJSON utilisation polygons, CSV time-series, PDF site report |
| Latency from acquisition to analysis | Sentinel-1 NRT: 1 to 3 hours post-acquisition; commercial optical: typically 24 to 48 hours after tasking |
Analytics Satellize can run
| Yard utilisation index | Pixel-based change detection on Sentinel-1 backscatter time-series, calibrated against WorldView-3 optical ground truth; supervised classification of occupied versus empty bay areas | Monthly GeoJSON layer with per-zone utilisation percentage; CSV time-series for trend analysis |
| Container stack height map | Shadow-length photogrammetry from WorldView-3 or Umbra SAR imagery using known solar or sensor geometry at acquisition time | Raster height grid at acquisition date; PDF site report with annotated imagery |
| Berth occupancy log | Object detection on optical and SAR imagery for vessel presence classification; cross-referenced against publicly available AIS positional data to reduce false positives | Per-berth occupancy time-series CSV; alert feed for occupancy state change events |
| Empty yard area mapping | Supervised spectral classification on multispectral imagery distinguishing bare concrete, container cover and ancillary structures; validated against SAR backscatter | GeoTIFF classification raster; GeoJSON polygons of idle yard zones with area statistics |
| Multi-year utilisation trend report | Time-series analysis of Sentinel-1 archive from 2014 to present, segmented by terminal zone; linear and seasonal decomposition to separate structural trend from cyclical variation | PDF feasibility-study annex with charts and methodology notes; underlying data in CSV |
| Adjacent land-activity correlation layer | Spatial join of port utilisation index with land-use classification of surrounding parcels; Pearson correlation analysis against available transaction date proxies | GIS layer with correlation coefficients per buffer zone; interpretive memo for property analysts |
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.