Pipeline construction progress monitoring with SAR coherence
SAR coherence change detection tracks right-of-way clearing, trenching, pipe stringing and backfill along overland pipeline corridors without waiting for cloud-free optical windows. Sentinel-1 C-band provides free, systematic coverage; ICEYE and Capella Space X-band adds the resolution needed for segment-level progress reporting.
Sensors
- Sentinel-1 A/B C-band SAR: 5.405 GHz, IW stripmap mode at 5 × 20 m ground range resolution, 6-day repeat at mid-latitudes with two satellites. Free, systematic archive from 2014. C-band coherence is sensitive to surface roughness change from soil disturbance and bare metal, but vegetated canopy can partially mask shallow trenches.
- ICEYE X-band SAR: 9.65 GHz stripmap at approximately 3 m resolution, spotlight at approximately 1 m. Tasked on demand; revisit to a specific corridor achievable within 24–48 hours commercially. X-band coherence decorrelates faster over vegetation, making fresh disturbance stand out more sharply than C-band.
- Capella Space X-band SAR: Spotlight mode at approximately 0.5 m resolution, sliding-spotlight at approximately 1 m. Provides the finest spatial detail available commercially for identifying individual pipe joints, weld stations and valve-pit excavations. Revisit is on-demand; latency from tasking to delivery is typically under 12 hours.
- ALOS-2 PALSAR-2 L-band SAR: 1.27 GHz, stripmap at 3–10 m resolution. L-band penetrates forest canopy and dry soil to a greater depth than C- or X-band, making it useful for detecting sub-canopy right-of-way clearing and buried-pipe settlement in forested or arid corridors. JAXA's 14-day repeat limits rapid progress tracking.
What disturbed ground tells a radar
Coherence in SAR interferometry measures how similar the radar signal is between two passes over the same ground. Stable, undisturbed surfaces, bare rock, dry pavement, standing structures, return high coherence values close to 1. Freshly turned soil breaks that stability immediately. The random orientation of clods, stones and exposed subsoil scatters the radar signal differently on every pass, driving coherence toward zero. A pipeline right-of-way under active construction is, from a coherence perspective, a long moving wound in the landscape.
Backscatter also changes. Bare soil is rougher at radar wavelengths than the grass or scrub it replaces, so the raw signal return increases. Strung pipe, particularly large-diameter steel, produces very strong specular and double-bounce returns at certain incidence angles. Both effects are detectable independently of coherence and serve as cross-checks. The combination of coherence loss and backscatter increase is specific enough to distinguish active construction from seasonal vegetation dieback or agricultural tillage, provided the analyst accounts for local land cover and acquisition geometry.
Matching the sensor to the construction phase
Right-of-way clearing produces the largest areal coherence loss. Felled timber and churned topsoil across a corridor 30–50 m wide is detectable in a single Sentinel-1 six-day pair at 5 × 20 m resolution. The signal is coarse but sufficient to confirm that clearing has started and to measure the advancing front kilometre by kilometre.
Trench excavation is narrower, typically 1–3 m wide, and may fall below Sentinel-1's ground-range resolution in a single pixel. Coherence loss is still detectable because the spoil mounds on either side of the trench widen the disturbed zone. For precise trench-front localisation, ICEYE or Capella stripmap acquisitions at 1–3 m resolve the spoil heap geometry directly.
Pipe stringing lays individual joints along the right-of-way before welding. Steel pipe 0.5 m or more in diameter produces a bright linear return in X-band imagery. Capella's 0.5 m spotlight mode can distinguish strung pipe from construction equipment parked alongside it, which matters when a client needs to verify that material has actually arrived on a segment rather than merely that the ground has been disturbed.
Backfill and reinstatement restore coherence over weeks to months, depending on soil moisture and rainfall. Monitoring coherence recovery is therefore a proxy for reinstatement completion. A segment that was backfilled two months ago but still shows low coherence may indicate poor compaction, ongoing settlement or incomplete restoration, each of which carries schedule and environmental compliance implications.
Frequency choice is not academic
C-band (Sentinel-1) and X-band (ICEYE, Capella) differ in how quickly coherence decays over vegetation. C-band coherence over dense forest decorrelates substantially within a single 6-day repeat even without any disturbance, because wind moves canopy elements between passes. X-band decorrelates even faster over vegetation, which sounds like a disadvantage but is actually useful: the contrast between a disturbed corridor and its forested surroundings is sharper in X-band because the baseline coherence of the forest is already low, making the construction scar stand out against a uniformly decorrelated background rather than a partially coherent one.
L-band (ALOS-2 PALSAR-2) sits at the other extreme. Canopy coherence is higher, and the signal penetrates several centimetres into dry soil. For corridors crossing dense tropical forest where C-band cannot see the ground at all, L-band may be the only frequency that detects sub-canopy clearing. The 14-day revisit is a real constraint for fast-moving construction fronts, but for monthly progress reporting on remote corridors it remains a credible option.
Arid and semi-arid corridors are the easiest environment. Dry soil has high coherence in all bands, so any disturbance produces a clean, persistent coherence drop. Humid tropical corridors are the hardest: high soil moisture, dense vegetation and frequent rainfall all accelerate decorrelation, reducing the contrast between disturbed and undisturbed ground. Honest programme design acknowledges this and increases acquisition frequency accordingly.
Building a progress curve from radar time series
A practical monitoring workflow begins by establishing a coherence baseline from archive imagery acquired before construction starts. Sentinel-1's archive runs from 2014, giving years of pre-disturbance data for almost any corridor on Earth. The baseline characterises normal seasonal coherence variation, which is essential: agricultural land adjacent to a pipeline corridor loses coherence every planting season regardless of construction activity, and conflating tillage with trenching would produce false progress reports.
Once construction begins, each new SAR pair is compared against the baseline. Pixels where coherence drops below a threshold, typically 0.3–0.4 for C-band over stable ground, are flagged as disturbed. Spatially, these flags form a front that advances along the corridor as work progresses. Measuring the distance of that front from the start point on each acquisition date produces a progress curve: kilometres completed against time, comparable to contractor milestone reports.
Segment-level reporting adds a second layer. The corridor is divided into named segments matching the construction contract structure. For each segment, the analysis reports the fraction of the segment showing coherence loss consistent with active disturbance, the fraction showing backfill-stage coherence recovery, and the fraction still undisturbed. Delivered as a GIS layer updated on each acquisition cycle, this gives project managers a spatial view of the schedule that is independent of what the contractor reports.
Honest limits of the method
SAR coherence cannot distinguish between types of disturbance without supporting context. A wildfire, a flood, or a new access road crossing the corridor all produce coherence loss. Interpretation requires knowing the construction schedule and local land-cover history. Automated alerts should be treated as flags for human review, not as definitive progress confirmations.
Very narrow trenches in fine-resolution imagery may still be sub-pixel. Capella's 0.5 m spotlight covers small areas at high cost; using it across hundreds of kilometres of corridor for every acquisition cycle is not practical. The sensible architecture is to use Sentinel-1 for corridor-wide weekly tracking and task high-resolution commercial SAR only on segments where the schedule is contested or where a specific verification is needed.
Coherence recovery after backfill is soil-moisture-dependent and cannot be converted to a precise compaction figure without ground truth. The method tells you that the surface has stabilised, not that the backfill meets engineering specification. Similarly, the presence of pipe-like bright returns in X-band imagery confirms that large metal objects are present; it does not confirm that the pipe is correctly aligned, welded or coated.
Satellize runs this workflow on Sentinel-1 open data for corridor-wide tracking and adds commercial ICEYE or Capella tasking for segment-level verification, following the same analytic architecture used in its Tonga crop-estimation programme: open data for breadth, commercial tasking for depth.
Typical figures
| Spatial resolution (Sentinel-1 IW) | 5 × 20 m ground range |
| Spatial resolution (ICEYE stripmap) | ~3 m; spotlight ~1 m |
| Spatial resolution (Capella spotlight) | ~0.5 m |
| Revisit (Sentinel-1, mid-latitudes) | 6 days with two satellites; 12 days with one |
| Revisit (ICEYE / Capella, commercial tasking) | On-demand; 24–48 hours typical |
| Radar frequency | C-band 5.4 GHz (Sentinel-1); X-band ~9.6 GHz (ICEYE, Capella); L-band 1.27 GHz (ALOS-2) |
| Archive depth (Sentinel-1) | From 2014 for most land areas |
| Coherence detection threshold (typical) | Disturbance flagged at coherence < 0.3–0.4 over stable-ground baseline |
| Minimum detectable disturbed width | ~10–15 m corridor disturbance detectable in Sentinel-1; ~3 m in ICEYE stripmap |
| Delivery formats | GeoTIFF coherence difference layers, GeoPackage / Shapefile segment-status polygons, CSV progress curve, PDF milestone report |
Analytics Satellize can run
| Construction-front advance curve | Coherence change detection on sequential SAR pairs against pre-construction baseline; front localisation by leading-edge pixel clustering | Time-series CSV and chart: kilometres of corridor disturbed per acquisition date, updated each Sentinel-1 pass |
| Segment-status layer | Per-segment aggregation of coherence-loss and backscatter-increase pixels, classified into undisturbed / active / backfilled / reinstated states | GeoPackage polygon layer with status attribute per named contract segment, refreshed weekly |
| Pipe-stringing detection | X-band backscatter anomaly detection for high-return linear features consistent with large-diameter steel pipe, validated against expected corridor geometry | Point or line GIS layer flagging confirmed pipe-presence zones, with confidence score per segment |
| Reinstatement progress report | Coherence recovery monitoring: segments where coherence returns to within one standard deviation of the pre-construction baseline are classified as reinstated | Monthly PDF report with map and table of reinstatement completion percentage per segment |
| Schedule-deviation alert | Comparison of observed construction-front position against client-supplied milestone schedule; statistical test for significant lag or unexplained halt | Email or API alert when a segment shows no detectable progress over two consecutive acquisition cycles despite being on the active schedule |
| Sub-canopy clearing detection (forested corridors) | ALOS-2 PALSAR-2 L-band coherence change detection for corridors where C-band is masked by dense canopy | GeoTIFF coherence difference layer and cleared-area polygon, delivered on each ALOS-2 acquisition (14-day cycle) |
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.