Third-party encroachment and illegal tap detection along onshore pipeline rights-of-way
Unauthorised construction, agricultural encroachment, and illegal hot-tapping along pipeline rights-of-way are detectable from orbit using high-resolution optical change detection and hydrocarbon-induced vegetation stress signals. Detection latency and cloud cover impose real limits that any honest programme must plan around.
Sensors
- Planet PlanetScope: 3 m native resolution, near-daily global revisit across the constellation of roughly 200 Dove satellites. The workhorse for corridor-wide change detection; four multispectral bands (blue, green, red, NIR) allow basic vegetation-index differencing. Revisit reliability degrades over equatorial cloud belts.
- Airbus Pléiades Neo: 30 cm panchromatic, 1.2 m multispectral (6 bands including red-edge). Task-on-demand with same-day or next-day revisit over a given target. The resolution needed to distinguish a freshly welded saddle clamp from a legitimate valve box, or to read construction-vehicle tyre tracks. Narrow swath (14 km) limits cost-effective corridor coverage to flagged hotspots.
- Planet SkySat: 50 cm panchromatic, 1 m multispectral. Taskable, with video-burst capability useful for confirming activity at a flagged site. Sits between PlanetScope and Pléiades Neo in both resolution and cost per km².
- Sentinel-2 MSI: 10 m in visible and NIR bands, 20 m in red-edge and SWIR. Five-day revisit at mid-latitudes (more frequent at higher latitudes due to orbital overlap). Free and openly archived since 2015. The SWIR bands (1610 nm and 2190 nm) respond to soil moisture and hydrocarbon-affected soils, making Sentinel-2 useful for seepage-related vegetation stress mapping over wider areas where 10 m resolution is sufficient.
What a right-of-way buffer is supposed to prevent
Pipeline operators clear and legally protect a strip of land, typically 10 to 30 m wide depending on jurisdiction and pipeline diameter, to allow inspection access and to keep third-party activity away from the buried asset. In practice, these corridors attract exactly the activity they are meant to exclude. Farmers extend fields across them because the cleared ground is flat and fertile. Developers sink foundations for structures because the land appears unused. Criminal groups target them for illegal hot-tapping, drilling into the pipe under a temporary shelter to siphon product, a practice responsible for hundreds of pipeline fires and fatalities globally and for billions of dollars in annual product theft.
Ground patrol of long corridors is expensive, infrequent, and dangerous in insecure regions. Satellite monitoring does not replace patrol; it directs it. The value is in reducing the search space from hundreds of kilometres to a handful of flagged coordinates before a patrol vehicle is dispatched.
What high-resolution optical change detection actually sees
Change detection at 3 m (PlanetScope) compares pixel-level spectral values between a clean baseline image and each new acquisition. Newly disturbed soil has a strongly different reflectance signature from established vegetation or compacted right-of-way surface. A fresh excavation pit of roughly 10 m × 10 m is detectable at 3 m resolution with high confidence; a small hand-dug hole of 1 to 2 m diameter is not. That is an honest floor. For confirmation of small-scale activity, a follow-on Pléiades Neo or SkySat task is required.
Structures are easier. A corrugated-iron shelter erected over a tap site, typically 3 to 8 m on a side, is unambiguous in 30 cm imagery and detectable as a spectral anomaly in 3 m imagery once it exceeds roughly one pixel cluster. Construction vehicles leave track scars that persist for days to weeks. The temporal signature matters: a single changed pixel is noise; the same anomaly present in three consecutive daily acquisitions is an event worth flagging.
Authorised crossings, road intersections, and maintenance pads must be distinguished from encroachments. This requires a georeferenced register of approved features as a mask layer. Without that register, false-positive rates are high and operators quickly lose confidence in the alert feed.
The vegetation kill signal from slow hydrocarbon seepage
Illegal taps are rarely perfectly sealed. Slow seepage of liquid hydrocarbons into surrounding soil suppresses root respiration and kills vegetation in a roughly circular or elongated patch downslope from the tap point. This shows up in the Normalised Difference Vegetation Index (NDVI) as a localised depression, and in Sentinel-2's red-edge band (705 nm, 20 m resolution) as reduced chlorophyll content before visible browning is apparent to the eye.
The spectral signature of hydrocarbon-stressed vegetation overlaps with drought stress, fungal infection, and waterlogging. Context matters enormously. A circular die-off patch of 20 to 50 m diameter centred on the pipeline corridor, with no corresponding drought pattern in surrounding fields, is a meaningful signal. The same patch 500 m from the corridor is probably agricultural. Analysts must apply spatial priors, not just spectral thresholds. Published work in the journal Remote Sensing has documented NDVI-based hydrocarbon-stress mapping in Niger Delta contexts, though detection performance varies sharply with vegetation type, season, and seepage volume.
Slow seepage from very small taps may produce a vegetation anomaly of only a few pixels at 10 m resolution. That is near the practical detection floor for Sentinel-2. Pléiades Neo's red-edge band at 1.2 m can resolve finer stress patterns, but only if the site has already been flagged for tasking.
Cloud cover and revisit: the limits that matter operationally
A near-daily revisit constellation is only useful if the sky cooperates. In tropical regions with persistent cloud cover, such as the Niger Delta, the Congo Basin, or parts of Southeast Asia, usable clear-sky acquisitions from PlanetScope may arrive only once every one to three weeks during the wet season. An illegal tap operation can be established, used, and dismantled in less than a week. That is a real detection gap, and any programme that does not state it plainly is selling false confidence.
SAR-based change detection (covered separately in the pipeline SAR coherence page) partially fills this gap because radar penetrates cloud. Optical and SAR methods are complementary, not competing. For clients operating in persistently cloudy corridors, an optical-only programme will miss events. A hybrid architecture is more reliable, though also more expensive to operate.
In semi-arid and temperate corridors, PlanetScope's near-daily revisit delivers usable imagery four to six days per week on average, making detection latency of two to four days realistic for large-scale encroachments. Smaller events may take longer to confirm across multiple acquisitions.
Building a practical monitoring programme
Effective corridor monitoring rests on three things: a clean baseline, a consistent change-detection pipeline, and a triage workflow that routes alerts to the right people quickly. The baseline should be a cloud-free composite from PlanetScope or Sentinel-2 covering the entire corridor, with all known authorised features masked. New acquisitions are differenced against this baseline using normalised spectral indices and pixel-level statistical thresholds calibrated to the local land cover.
Alert triage works in two tiers. Tier-one alerts, covering large anomalies above a defined area threshold, trigger an automatic notification with coordinates and a thumbnail. Tier-two alerts, covering smaller or ambiguous changes, queue for analyst review before dispatch. This prevents the alert fatigue that kills monitoring programmes within months of launch.
Satellize runs this kind of corridor analytics on open constellations with commercial tasking added for confirmation. The architecture is similar in principle to the crop-monitoring workflow developed for the Kingdom of Tonga, adapted for linear infrastructure rather than agricultural parcels. For a pipeline client, the concrete deliverable is a daily or weekly GIS alert layer with flagged coordinates, a confidence score, and a recommended response tier, not a slide deck of interesting images.
What this method cannot do
Optical change detection cannot confirm product theft is occurring; it can only flag surface activity consistent with an illegal tap. Confirmation requires ground inspection or, in some cases, pressure-drop telemetry from the pipeline's own SCADA system. The satellite flags the location; the operator decides whether to dispatch.
Very small excavations, hand-dug pits less than roughly 2 m wide, and activity conducted under tree canopy or temporary camouflage sheeting will not be reliably detected at 3 m resolution. Organised criminal groups in some regions have learned to work under cover specifically to defeat aerial and satellite observation. This is a real adversarial dynamic, and it means that satellite monitoring should be treated as a probabilistic screening tool, not a guarantee of detection.
Typical figures
| Corridor-wide screening resolution | 3 m (PlanetScope) or 10 m (Sentinel-2 visible/NIR) |
| Confirmation resolution | 30 cm panchromatic / 1.2 m multispectral (Pléiades Neo); 50 cm / 1 m (SkySat) |
| Revisit frequency (screening) | Near-daily (PlanetScope); 5-day (Sentinel-2 at mid-latitudes) |
| Revisit frequency (confirmation tasking) | Same-day to next-day on request (Pléiades Neo, SkySat), subject to cloud and scheduling |
| Minimum detectable surface anomaly (optical) | Approximately 10 m × 10 m excavation at 3 m resolution; structures from roughly 3 m × 3 m at 30 cm resolution |
| Vegetation stress detection floor | NDVI anomaly patches of roughly 20 m diameter at Sentinel-2 10 m resolution; finer at Pléiades Neo 1.2 m |
| Spectral bands used | Visible (RGB), NIR, red-edge (Sentinel-2 at 705 nm / 740 nm), SWIR (1610 nm, 2190 nm) |
| Detection latency (clear-sky conditions) | 2–4 days from event to confirmed alert in temperate/semi-arid corridors |
| Detection latency (persistent cloud) | Up to 2–3 weeks in tropical wet-season conditions; optical-only programmes have a real gap here |
| Archive depth | Sentinel-2 from 2015; PlanetScope from approximately 2016; Pléiades from 2012 |
Analytics Satellize can run
| Daily corridor change-detection alert layer | Normalised spectral differencing (NDVI, normalised bare-soil index) against cloud-free baseline composite; pixel-cluster thresholding with minimum area filter | GIS vector layer (GeoJSON or shapefile) of flagged polygons with coordinates, area, confidence score, and acquisition timestamp; delivered within hours of satellite pass |
| Vegetation stress anomaly map | NDVI and red-edge chlorophyll index time-series analysis on Sentinel-2 and Pléiades Neo; spatial masking against known drought and land-use patterns to isolate corridor-proximate anomalies | Monthly raster and vector product showing persistent stress patches along corridor with severity classification; flagged sites ranked by proximity to pipeline centreline |
| New structure and excavation report | Object-based image analysis on Pléiades Neo or SkySat tasked imagery; comparison against authorised-feature register to classify encroachment versus permitted crossing | PDF or structured JSON report per flagged site with before/after image chips, estimated structure footprint, and classification (probable encroachment / authorised / ambiguous) |
| Baseline right-of-way feature register | One-time manual and semi-automated digitisation of authorised structures, crossings, valve sites, and maintenance pads from high-resolution imagery; used as permanent mask in change-detection pipeline | Georeferenced GIS layer (shapefile or GeoPackage) of authorised features with attribute table; updated on client notification of new permitted works |
| Alert triage and confidence scoring | Two-tier triage: automated tier-one dispatch for anomalies above area threshold; analyst-reviewed tier-two queue for smaller or ambiguous changes; scoring based on anomaly persistence across multiple acquisitions | Structured alert feed with confidence tier (high / medium / requires confirmation), recommended response action, and link to tasking request if confirmation imagery is needed |
| Historical encroachment audit | Retrospective change detection over archived PlanetScope and Sentinel-2 time series to establish when and where encroachments first appeared; useful for legal proceedings or insurance claims | Timestamped image series and written summary per site of interest; earliest detectable date of anomaly with supporting imagery |
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.