Vessel Collision and Allision Incident Reconstruction from Satellite Evidence
Post-incident reconstruction of maritime collisions and allisions using SAR positioning, AIS track replay, wake geometry, and sub-metre optical damage assessment. Covers evidentiary chain-of-custody and the temporal gaps that bound reconstruction precision.
Sensors
- ICEYE SAR constellation: Stripmap mode delivers approximately 3 m ground resolution; Spot mode reaches sub-1 m. Revisit to any point on Earth within hours under tasking, with same-day archive delivery possible. Provides vessel position, heading and Doppler-derived speed independent of visibility or cloud cover.
- Sentinel-1 SAR (ESA): Interferometric Wide Swath mode covers 250 km at 5 x 20 m resolution with a 6-day repeat for a single satellite; the two-satellite constellation halves that in many latitudes. Free archive from 2014 onward. Useful for pre-incident baseline and for detecting vessel wakes and oil sheens.
- Maxar WorldView Legion: Panchromatic resolution of approximately 0.3 m. Up to 15 revisits per day over priority areas under tasking. Resolves hull deformation, paint-transfer smears, anchor-chain geometry, and fixed-structure contact points at allision sites.
- Planet SkySat: 0.5 m native resolution, tasked on demand. Rapid collect scheduling (often same-day) makes it practical for acquiring post-incident optical imagery before salvage operations alter the scene. Useful for documenting vessel orientation, freeboard change, and debris distribution.
What the record actually contains before lawyers arrive
Maritime incidents rarely have witnesses who are also instruments. AIS transponders record position, speed over ground, course over ground, and heading at intervals that vary by vessel class: Class A transponders transmit every 2 to 10 seconds when under way. That time-stamped track is the first layer of the reconstruction. The second layer is satellite imagery, which may capture the vessels at or near the moment of contact, or within minutes to hours afterward depending on orbital geometry and tasking response.
The critical distinction is what each source measures. AIS is self-reported and can be delayed, interpolated, or in some cases manipulated. Satellite SAR measures physical radar backscatter from the hull and wake, independently of what the transponder says. Where the two disagree, the physics tends to be more credible in arbitration, provided the satellite data carries a proper chain-of-custody record from tasking order through to processed product.
SAR Doppler and wake geometry: reading speed and heading from radar
A SAR sensor illuminates a target with coherent microwave pulses and measures the phase shift in the return signal. A moving vessel introduces a Doppler frequency offset proportional to its velocity component along the radar line of sight. For Sentinel-1 in IW mode, published studies place the velocity estimation uncertainty at roughly 0.5 to 1.5 knots under good signal-to-noise conditions, which is sufficient to distinguish a vessel making 4 knots from one making 12 knots, but not to resolve small speed differences within that range. ICEYE Spot mode, with its finer resolution, narrows that ambiguity somewhat, though published figures for operational accuracy remain sensor- and geometry-dependent.
Wake geometry adds a second, independent heading estimate. A Kelvin wake subtends a half-angle of approximately 19.5 degrees regardless of speed in deep water, a consequence of the dispersion relation for surface gravity waves. The wake axis bisects that angle and points directly along the vessel's track. In shallow water or at very high Froude numbers the geometry deviates, so analysts must account for bathymetry. The intersection of the Doppler-derived velocity vector and the wake-axis heading gives a position-and-trajectory fix that is geometrically overdetermined and therefore more defensible than either measurement alone.
Sub-metre optical imagery and what hull damage reveals
At 0.3 to 0.5 m resolution, WorldView Legion and SkySat imagery can resolve features that are directly relevant to contact mechanics. Paint transfer, which appears as a colour anomaly against the hull's baseline livery, indicates the height of the contact zone and therefore the relative freeboard of the two vessels at the moment of impact. Buckled plating, missing railings, and displaced anchor hawsepipes each carry geometric information about the angle of approach. An allision with a fixed structure such as a berth, bridge pier, or offshore platform typically leaves a linear crush pattern whose orientation, read against the structure's known geometry, constrains the vessel's heading at contact to within a few degrees.
The honest limit here is time. Optical imagery is cloud-dependent, and most post-incident response windows are short before salvage vessels and repair crews alter the scene. If the first clear optical collect arrives 48 hours after the incident, some damage evidence will already have been obscured. SAR is cloud-independent and can be tasked within hours, but SAR cannot resolve paint-transfer colour anomalies. The two sensor types are complementary, not interchangeable.
Evidentiary chain of custody: what arbitration panels actually require
Satellite data is increasingly accepted in maritime arbitration, but acceptance is not automatic. The International Maritime Organization's casualty investigation code and most flag-state procedures require that evidence be traceable to its source without unexplained gaps. For satellite data, that means the tasking metadata (operator, time of acquisition, satellite ID, orbit parameters), the processing log (which algorithms were applied, by whom, at what version), and the delivery record must all be preserved and producible.
Commercial SAR operators including ICEYE provide acquisition certificates that include UTC timestamp, satellite position, and imaging mode. These are analogous to a camera's EXIF metadata but with cryptographic or operator-signed authentication. The processed analytic outputs, velocity estimates, wake-axis bearings, damage annotations, must be tied back to those certificates in the final report. Gaps in that chain, for instance a velocity figure derived from a processing step that cannot be reproduced, will be challenged. Analysts should retain intermediate products, not just the final map.
Where reconstruction breaks down: honest limits
The biggest temporal gap is almost always the interval between the last pre-incident satellite pass and the incident itself. If no satellite was tasked and no archived pass exists within a useful window, the reconstruction must rely entirely on AIS interpolation for the pre-contact period. AIS interpolation assumes constant velocity between reported positions, which is physically reasonable in open water but unreliable during manoeuvring. A vessel executing an emergency turn may change heading by 90 degrees in less than two minutes, a period during which AIS may record only one or two position reports.
Cloud cover compounds the problem for optical sensors. In tropical or high-latitude winter conditions, the first cloud-free optical collect may arrive days after the incident. SAR fills part of that gap but cannot establish the visual state of the hull. Doppler velocity estimation also degrades when a vessel is nearly stationary or moving perpendicular to the radar line of sight, precisely the geometry that arises when a vessel has already stopped after a collision. Analysts should state the uncertainty bounds on every derived parameter explicitly. A heading estimate of 247 degrees plus or minus 8 degrees is useful. A heading stated as 247 degrees without qualification is not credible.
Satellize structures incident-reconstruction packages around these documented limits, presenting each derived parameter with its uncertainty range and the sensor evidence that constrains it, following the same analytical discipline applied in the Tonga crop-estimation programme where quantified uncertainty was a contractual deliverable.
Typical figures
| SAR spatial resolution (ICEYE Spot) | Sub-1 m (approximately 0.5 m) |
| SAR spatial resolution (Sentinel-1 IW) | 5 x 20 m ground range |
| Optical spatial resolution (WorldView Legion) | Approximately 0.3 m panchromatic |
| SAR Doppler velocity uncertainty | 0.5 to 1.5 knots (published range, Sentinel-1 IW; geometry-dependent) |
| Wake-axis heading uncertainty | Typically 2 to 5 degrees in deep water; larger in shallow or high-Froude conditions |
| AIS position reporting interval (Class A under way) | 2 to 10 seconds |
| Tasked SAR revisit latency (ICEYE) | Hours to same-day under priority tasking |
| Sentinel-1 archive depth | From 2014 onward, freely accessible |
| Optical cloud dependency | 100% cloud-blocked; SAR unaffected by cloud or darkness |
| Delivery formats | GeoTIFF, KMZ, PDF annotated report, GIS shapefile, acquisition certificate (metadata) |
Analytics Satellize can run
| AIS track reconstruction with interpolated trajectory | Kinematic interpolation of Class A/B AIS position reports; gap-filling flagged with confidence intervals | Time-stamped GIS track layer with per-segment speed and heading, uncertainty annotations at manoeuvre gaps |
| SAR-derived vessel position and heading fix | Bright-target detection in SAR amplitude image; wake-axis geometry measurement; Doppler offset extraction | Point feature with UTC timestamp, position (lat/lon), heading bearing, estimated speed, and uncertainty range |
| Doppler velocity profile at time of imaging | Along-track Doppler shift analysis on SAR complex data (published method class: along-track interferometry and sub-aperture processing) | Velocity estimate with confidence interval, tied to acquisition certificate for evidentiary use |
| Hull damage annotation from sub-metre optical imagery | Manual and semi-automated change detection against pre-incident optical baseline; colour anomaly mapping for paint transfer | Annotated orthorectified image with damage-zone polygons, contact-height estimate, and approach-angle inference |
| Oil sheen and debris field mapping | SAR backscatter dampening for oil-on-water detection; optical spectral anomaly for debris extent | Polygon layer of sheen and debris extent with timestamps, suitable for MARPOL reporting and salvage planning |
| Incident timeline synthesis report | Multi-source fusion of AIS, SAR fixes, optical damage state, and vessel particulars (LOA, beam, draught from public registers) | Structured PDF report with chain-of-custody appendix, per-parameter uncertainty table, and timeline graphic; formatted for submission to flag-state or P&I club investigation |
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.