Ship-to-shore crane activity as port productivity indicator
Ship-to-shore gantry cranes are the physical rate-limiters of container terminal throughput. Satellite imagery and SAR time-series can classify crane posture, estimate move-rate proxies, and flag productivity shifts without a single ground sensor.
Sensors
- Airbus Pléiades Neo: 30 cm panchromatic resolution; boom angle, trolley position and shadow geometry are directly interpretable at this scale. Tasking latency of one to two days for standard priority; daily revisit over most ports with the four-satellite constellation.
- Maxar WorldView-Legion: 30 cm native resolution across a six-satellite constellation targeting up to 15 revisits per day over priority sites. High temporal density makes it the strongest option for intra-shift move-sequence reconstruction, though cloud remains a hard constraint.
- Planet SkySat: 50 cm resolution with on-demand tasking and up to 12 collects per day over a given site. Slightly coarser than Pléiades Neo but sufficient to distinguish boom-raised from boom-lowered posture and to detect trolley travel along the boom girder.
- Sentinel-1 SAR (C-band, IW mode): 10 m ground-range resolution in Interferometric Wide Swath mode; six-day repeat at the equator, shorter at higher latitudes. Individual cranes are not resolved geometrically, but a terminal's aggregate radar cross-section changes measurably when multiple cranes shift posture. Coherence change between passes flags structural rearrangement. Cloud-transparent.
What a crane's posture actually reveals
A ship-to-shore gantry crane has three observable states from orbit: boom lowered over a vessel (active working position), boom raised to the travel or parked angle (typically 75 to 85 degrees from horizontal), and trolley position along the boom girder. At 30 cm resolution, the boom is a resolved linear feature roughly 60 to 70 metres long on the largest post-Panamax cranes. Shadow length on the apron gives boom angle to within a few degrees when solar elevation is known. Trolley position is detectable as a small high-reflectance object on the boom, though this requires near-nadir viewing and good contrast.
Boom angle alone separates working cranes from idle cranes with high confidence. A crane serving a vessel has its boom horizontal or slightly inclined; a crane not in use has its boom raised clear of the ship's superstructure. This binary classification is achievable even at 50 cm resolution, which means SkySat collects contribute usefully to a multi-sensor time-series even when the finer-resolution assets are unavailable.
Estimating moves per hour from multi-pass sequences
A single image gives posture; a sequence of images gives a proxy for throughput. If a crane is observed with its trolley at the landside end of the boom at time T1 and at the waterside end at time T2, one cycle has occurred. The standard industry metric, moves per hour (MPH), requires counting complete lift cycles. With optical satellites, the practical ceiling on intra-shift temporal resolution is set by revisit frequency and cloud probability, not by sensor capability.
WorldView-Legion's up to 15 daily revisits over priority sites is the current practical maximum for commercial optical tasking. At that cadence, the inter-image gap averages roughly 90 minutes, which is far too coarse to count individual moves on a crane running 25 to 30 MPH. What multi-pass sequences can do is establish a lower bound on activity: if the trolley position changes between consecutive images, at least one move occurred in the interval. Aggregated across a full port call, this yields an activity fraction rather than a precise MPH figure. For benchmarking terminals against each other, or tracking the same terminal across months, that proxy is genuinely informative.
The honest ceiling: satellite imagery cannot replicate the per-move counting that terminal operating systems or quayside laser sensors provide. The value is in independent, externally verifiable activity indices, not in replacing operational instrumentation.
SAR coherence as a cloud-tolerant complement
Sentinel-1 C-band SAR does not resolve individual cranes at 10 m, but it detects changes in a terminal's backscatter signature between repeat passes. A gantry crane is a large, geometrically complex metallic structure; its radar cross-section varies substantially depending on boom angle and trolley position. When multiple cranes shift from parked to working posture between two passes, the terminal's aggregate backscatter and coherence change in a measurable way.
Coherence change detection, which compares the phase correlation between two SAR acquisitions over the same area, is well-established in the literature for detecting surface change. A terminal with active crane operations between two Sentinel-1 passes will show lower coherence over the crane apron than the same terminal during a port closure or low-activity period. The six-day repeat cycle limits temporal resolution, but the cloud-transparency of SAR makes it the only reliable option for tropical ports where optical collection is frequently blocked. The two sensor types are complementary: optical for posture classification and move-sequence proxying when skies are clear, SAR for confirming activity or inactivity under overcast conditions.
The shadow geometry method and its limits
Shadow length on the apron is a direct function of boom angle and solar elevation at the time of acquisition. Given the solar ephemeris (which is precisely known for any location and time), measuring shadow length in the image allows back-calculation of boom angle. Post-Panamax cranes typically stand 75 to 100 metres above the quay at the apex; their shadows at moderate solar elevation can extend 50 to 150 metres across the apron, making them easy to measure even at 50 cm resolution.
The method fails when shadows overlap with vessel superstructure, adjacent cranes, or container stacks. It also requires near-nadir collection; off-nadir angles above roughly 25 degrees introduce parallax that complicates shadow-to-height conversion without a known crane height reference. Archive imagery, where collection geometry was not controlled for this application, requires per-image geometric correction before shadow measurements are reliable.
Turning crane counts into a port productivity index
The analytic product that matters to a port authority, shipping line, or trade-finance desk is not a raw crane posture classification but a normalised productivity index: what fraction of deployed cranes were in active working posture during a given vessel call, and how does that compare across berths, terminals, or time periods. This requires combining crane posture data with vessel presence data from AIS, which provides the start and end times of a port call with minute-level precision.
The combined output is a crane utilisation ratio per vessel call: cranes observed working divided by cranes deployed at the berth, averaged across available image passes during the call. This metric is independent of terminal operator reporting, which makes it useful for third-party due diligence, trade-lane benchmarking, and sovereign port development planning. Satellize applies this kind of multi-source index construction in its analytics work, including the Tonga crop-estimation programme, where the same principle of combining sparse satellite passes into a continuous activity proxy applies in a different domain entirely.
Index values are most meaningful as relative measures. A terminal consistently showing 60 percent crane utilisation during vessel calls, against a peer group averaging 80 percent, is a signal worth investigating. The absolute figure depends on how many image passes fell during the call and the cloud fraction over those passes, both of which must be reported alongside the index to make it interpretable.
Archive depth and what it enables
Pléiades and WorldView commercial archives extend back roughly a decade over major ports, with tasking frequency having increased substantially since 2018 as constellation capacity grew. Sentinel-1, freely available through the Copernicus Data Space, provides a continuous SAR archive from late 2014. Together, these archives support multi-year productivity trend analysis without any new tasking cost for the historical component.
The practical constraint on archive analysis is cloud fraction. For a tropical port such as Singapore or Colombo, a large fraction of historical optical collects will be partially or fully obscured. Usable-image counts per month vary from near-zero during monsoon periods to several per week in the dry season. Any historical analysis must report usable-image density alongside the productivity index, or the gaps will be misread as low activity rather than missing data. SAR archive fills some of those gaps but at coarser analytical resolution.
Typical figures
| Best optical spatial resolution | 30 cm (Pléiades Neo, WorldView-Legion) |
| SAR spatial resolution | 10 m ground range (Sentinel-1 IW mode) |
| Optical revisit (priority tasking) | Up to 15 passes per day over a given site (WorldView-Legion); 1 pass per day typical (Pléiades Neo four-satellite constellation) |
| SAR revisit | 6 days at equator, shorter at higher latitudes (Sentinel-1 ascending and descending combined) |
| Minimum detectable feature | Boom posture (raised vs. lowered) detectable at 50 cm; trolley position requires 30 cm and near-nadir geometry |
| Cloud tolerance | Optical: zero cloud tolerance. SAR: fully cloud-transparent |
| Archive depth | Commercial optical: approximately 10 years over major ports. Sentinel-1 SAR: from October 2014 (Copernicus Data Space) |
| Tasking latency | 1 to 2 days for standard commercial optical priority; near-real-time SAR acquisition within current 6-day repeat |
| Spectral / frequency | Panchromatic and multispectral (optical); C-band 5.4 GHz VV/VH polarisation (Sentinel-1) |
| Delivery formats | GeoTIFF orthoimage, GeoJSON crane-state classification layer, CSV activity index time-series |
Analytics Satellize can run
| Crane posture classification | Object detection on very-high-resolution optical imagery; boom angle from shadow geometry using solar ephemeris | GeoJSON layer per image pass: each crane labelled working, parked, or boom-raised, with confidence score |
| Crane utilisation ratio per vessel call | Posture classifications fused with AIS vessel-presence timestamps; active-crane fraction averaged across passes during the call | CSV report per port call: vessel IMO, call duration, passes used, cloud fraction, utilisation ratio |
| Terminal productivity index (monthly) | Aggregated utilisation ratios normalised by usable-image count; compared across berths or terminals | Monthly time-series chart and GIS layer; peer-terminal benchmarking table on request |
| SAR activity flag | Sentinel-1 coherence change detection between repeat passes over the crane apron; backscatter amplitude difference | Binary activity/inactivity flag per terminal per SAR pass; delivered as GeoJSON alert feed |
| Multi-year productivity trend | Archive optical and SAR time-series analysis; seasonal decomposition to separate weather-driven data gaps from genuine activity change | Annual trend report with usable-image density disclosed alongside index values |
| Boom angle time-series | Shadow-length measurement in calibrated orthoimage; trigonometric inversion using known solar elevation and crane height reference | Per-crane boom-angle estimate per image pass, with geometric uncertainty range |
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.