Subsea pipeline seabed exposure and free-span detection from SAR backscatter
In shallow water, a pipeline lifted clear of the seabed by scour returns a distinctive double-bounce SAR signal. Sentinel-1 and high-resolution commercial X-band data can map these free spans at scale, prioritising where inspection vessels go first.
Sensors
- Sentinel-1 (C-band SAR, ESA): IW mode delivers 10 m ground range resolution with a 250 km swath. Repeat pass every 6 days over Europe and many offshore regions, 12 days elsewhere. Free and open archive back to 2014. C-band (5.405 GHz) penetrates cloud but is more susceptible to sea-clutter at moderate sea states than X-band; practical use limited to Beaufort 3 or below for shallow-pipeline detection.
- ICEYE X-band SAR constellation: Spotlight mode achieves approximately 0.5 m azimuth resolution; Strip mode around 3 m. Revisit can be tasked to sub-daily over a specific offshore asset. X-band (9.65 GHz) produces a stronger double-bounce return from a cylindrical steel target than C-band, improving signal-to-clutter ratio in marginal sea states. Commercial tasking adds cost but the geometry is better suited to narrow pipeline detection.
- Capella Space X-band SAR: Spotlight mode resolution down to approximately 0.5 m by 0.5 m. Useful for confirming ambiguous Sentinel-1 detections and for producing higher-confidence free-span length estimates on known pipeline routes. Same X-band physics advantages as ICEYE.
- COSMO-SkyMed (X-band SAR, ASI): Second-generation system (CSG) offers Spotlight-2 mode at approximately 0.35 m resolution. Established archive from 2007 onward. Useful for retrospective analysis of scour events and for comparing current exposure against historical baselines on well-documented pipeline corridors.
What a lifted pipe looks like to a radar
A steel pipeline resting on the seabed is acoustically opaque but electromagnetically interesting. When a section is scoured free and sits proud of the sediment in water shallower than roughly 20 metres, it creates a geometry that SAR illumination exploits: the radar pulse bounces from the water surface down to the pipe wall and back up to the sensor. This double-bounce mechanism produces a localised bright return, typically 5 to 15 dB above the surrounding sea-surface clutter in calm conditions, depending on pipe diameter, incidence angle and wavelength.
The effect is well documented in the radar-physics literature and is the same mechanism that makes bridge piers and ship hulls bright in SAR imagery. Pipeline diameters in the 0.3 m to 1.2 m range common in offshore oil and gas infrastructure fall within the resonance region for X-band, making the return particularly strong. C-band (Sentinel-1) still detects the signature but with a lower signal-to-clutter margin, which is why sea state matters so much: a 1-metre swell raises the background noise floor enough to swamp the pipe return entirely.
The 20-metre depth ceiling and why it is hard
SAR is a surface-scattering sensor. It does not image through water in any operationally useful sense. The double-bounce signature only exists because the pipe is close enough to the surface for the two-way path to produce a coherent return before the signal is attenuated by the water column. Published electromagnetic modelling and empirical observations from North Sea and Gulf of Mexico surveys suggest the practical detection depth is 15 to 20 metres in calm conditions, degrading rapidly below that.
This is not a limitation that better satellites will fix. It is physics. The consequence is that this method is genuinely useful for nearshore export lines, river-crossing sections, and shallow-shelf segments, but has nothing to say about deepwater flowlines. Operators of deepwater infrastructure should not expect this technique to substitute for remotely operated vehicle inspection or multibeam sonar survey. What it does do is tell you which shallow sections warrant urgent acoustic attention and which can wait, which is a meaningful operational input when survey vessel day-rates run into six figures.
From bright pixel to free-span length estimate
Detection starts with a known pipeline route, usually from operator GIS data or from published maritime charts. The analyst masks the expected corridor, typically a 200 to 500 metre buffer around the centreline, and applies a constant false-alarm rate (CFAR) detector tuned to the expected target brightness and spatial extent. Detections that align with the pipeline centreline and show the elongated bright-dark-bright pattern characteristic of a cylindrical target are flagged as candidate free spans.
Free-span length is estimated from the along-track extent of the bright return, corrected for the point-spread function of the sensor. At Sentinel-1 IW resolution (10 m), spans shorter than 20 to 30 metres are below reliable detection. ICEYE or Capella Spotlight data push that floor down to roughly 3 to 5 metres. The uncertainty on span length is typically plus or minus one to two resolution cells, so a reported 25-metre span from Sentinel-1 data carries a real uncertainty of plus or minus 10 to 20 metres. That is honest and it still narrows the search area for a survey vessel considerably.
Multi-temporal stacking helps. Averaging several calm-state acquisitions over weeks suppresses transient clutter returns from wave action, vessel wakes and fish shoals, all of which can produce false positives in single-pass analysis. A detection that persists across four or more independent passes is unlikely to be artefact.
Sentinel-1 archive as a scour chronology
One underused capability is retrospective analysis. Sentinel-1 has been acquiring consistently since 2014, giving a decade of C-band imagery over most offshore regions. A pipeline that was not exposed in 2018 but shows a bright return in 2024 has experienced scour in that interval. Correlating the appearance of the signature with storm records, dredging activity or nearby construction narrows the likely cause and helps operators understand whether the scour is ongoing or has stabilised.
This kind of temporal analysis is cheap relative to a mobilised survey and can inform the inspection programme before any vessel leaves port. The limit is that Sentinel-1 archive coverage is not uniform globally and some tropical offshore regions have significant gaps due to acquisition scheduling priorities.
What this method cannot do, stated plainly
It cannot measure the gap between pipe and seabed. SAR gives no depth information. A pipeline that appears exposed in SAR imagery might be 0.05 metres clear of the sediment or 0.5 metres; those two situations have very different fatigue implications and only sonar or ROV inspection can distinguish them.
It cannot detect spans in turbid or rough water. A sea state above Beaufort 3 to 4 typically renders the method unreliable for C-band and marginal for X-band. Operators in the North Sea or offshore Norway should expect seasonal windows, broadly spring and summer, when calm-state acquisitions are achievable. Tropical offshore regions with persistently low swell offer better year-round utility.
It cannot detect spans on buried pipelines. If the pipe is trenched and backfilled, there is no surface return regardless of what happens to the sediment above. The method is specific to pipelines that were designed to rest on or near the seabed surface.
Satellize runs this detection workflow on Sentinel-1 open data combined with commercial X-band tasking for confirmation passes, delivering flagged span candidates as GIS polygon layers with confidence scores and a per-span summary report. The same platform that runs crop-area estimation for the Kingdom of Tonga handles the temporal stacking and CFAR detection here; the underlying change-detection architecture is the same.
Fitting SAR screening into an inspection programme
A rational inspection programme for a shallow-water pipeline network might run Sentinel-1 screening monthly during calm-season windows, flag any new or extended bright returns against the known route, and task a commercial X-band pass within days to confirm. Confirmed detections above a threshold span length trigger a survey vessel deployment. Sections that show no persistent SAR signature can remain on a longer acoustic inspection interval.
The economic case is straightforward. A Sentinel-1 pass costs nothing beyond processing. A commercial X-band confirmation costs a fraction of a survey vessel day-rate. If SAR screening prevents even one unnecessary mobilisation per year, or catches a developing free span before it reaches fatigue-critical length, the analytics cost is trivial relative to the avoided risk. The method does not replace the acoustic survey. It makes the acoustic survey more targeted.
Typical figures
| Spatial resolution (Sentinel-1 IW) | 10 m ground range × 10 m azimuth |
| Spatial resolution (ICEYE / Capella Spotlight) | 0.5 m × 0.5 m (approx.) |
| Revisit (Sentinel-1, open) | 6 days (Europe/major offshore); 12 days elsewhere |
| Revisit (commercial X-band, tasked) | Sub-daily feasible; subject to constellation scheduling |
| Radar frequency | C-band 5.405 GHz (Sentinel-1); X-band ~9.65 GHz (ICEYE, Capella, COSMO-SkyMed) |
| Maximum detection depth | ~15–20 m water depth; degrades rapidly below |
| Minimum detectable span length | ~20–30 m (Sentinel-1); ~3–5 m (X-band Spotlight) |
| Sea-state limit | Beaufort 3 or below for reliable detection; Beaufort 4 marginal for X-band |
| Archive depth (Sentinel-1) | 2014 to present |
| Delivery formats | GeoJSON / Shapefile polygon layers, per-span CSV summary, PDF report |
Analytics Satellize can run
| Free-span candidate map | CFAR detection on SAR intensity along known pipeline corridor mask | GeoJSON polygon layer with confidence score per detection, updated per acquisition |
| Span length estimate | Along-track extent of bright return corrected for sensor point-spread function | Per-span attribute table with estimated length and uncertainty range (±resolution cells) |
| Scour chronology | Multi-temporal Sentinel-1 archive stack; first-detection date and change rate per span | Time-series chart and GIS layer showing span appearance date and growth over archive period |
| False-positive suppression layer | Multi-pass persistence filtering; detections present in fewer than N passes flagged as probable clutter | Filtered candidate list with pass-count confidence metric appended |
| X-band confirmation report | Commercial Spotlight tasking on Sentinel-1 flagged locations; higher-resolution CFAR re-run | Per-location confirmation PDF with annotated SAR chip and updated span geometry |
| Survey vessel prioritisation ranking | Span length, persistence score and proximity to known high-consequence zones combined into risk rank | Ranked inspection schedule in CSV and PDF, ready for vessel mobilisation planning |
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.