Road and rail corridor construction monitoring
Optical and SAR satellites track earthwork, vegetation clearance and pavement progress along road and rail corridors. Where tropical cloud blocks optical sensors, Sentinel-1 backscatter change reveals surface transitions from bare earth to compacted sub-base.
Sensors
- Planet SkySat: 50 cm native resolution (resampled to 50 cm product), tasked on demand. Captures fine detail such as machinery positions, culvert headwalls and pavement edge definition. Revisit depends on tasking priority but can achieve same-day collection over a fixed point. Archive from 2014.
- Sentinel-2 MSI: 10 m resolution in visible and near-infrared bands, 5-day revisit at the equator with both satellites. Useful for corridor-scale NDVI change (vegetation clearance) and bare-soil mapping. Free, open archive from 2015. Cloud cover is the principal limitation in humid tropical corridors.
- Sentinel-1 SAR (C-band): Interferometric Wide Swath mode delivers 10 m ground range resolution, 6-day revisit (12-day for single satellite). C-band backscatter is unaffected by cloud or darkness. Backscatter intensity change from rough bare earth to smooth compacted sub-base provides a measurable proxy for earthwork and pavement progress.
- Airbus Pléiades Neo: 30 cm resolution, four-satellite constellation enabling revisit of less than 24 hours over priority sites. Tri-stereo tasking supports 1 m contour digital surface models useful for cut-and-fill volume estimation along embankments and cuttings.
Why a linear corridor is harder to monitor than a building site
A point construction site fits inside a single high-resolution image. A 400 km road or rail corridor does not. It crosses multiple satellite swaths, several climate zones and, in tropical regions, near-permanent cloud decks that can render optical imagery useless for weeks at a time. Progress at kilometre 220 may be running ahead of schedule while kilometre 310 sits flooded. No single image date captures the full picture.
The standard response, ordering commercial tasking on a fixed monthly cadence, is expensive and still cloud-prone. The practical answer is sensor fusion: use Sentinel-1 SAR as the always-available backbone, Sentinel-2 for vegetation and soil-type context when skies permit, and commercial optical tasking (SkySat or Pléiades Neo) selectively, at the specific chainages where contract milestones require photographic evidence.
What a SAR backscatter shift actually means on a road corridor
C-band radar (5.4 GHz, roughly 5.6 cm wavelength) interacts with surface roughness at the centimetre scale. Freshly graded bare earth, with its clods and ruts, scatters the radar signal diffusely and returns a relatively high backscatter value. A compacted granular sub-base, rolled smooth, returns less energy to the sensor. Asphalt or concrete returns even less, and at a characteristic specular angle. This progression, rough to smooth, is detectable as a systematic decrease in mean backscatter along a corridor segment between two acquisition dates.
The detection is not clean. Moisture content confounds it: wet soil can return higher backscatter than dry compacted material, temporarily masking the roughness signal. The practical approach is to compare acquisitions from the same orbital pass (same incidence angle, same polarisation) separated by at least one dry-season interval, or to use the ratio of VV to VH polarisation as a partial moisture correction. Published studies using Sentinel-1 over road construction in sub-Saharan Africa and Southeast Asia have demonstrated that a change in mean backscatter of roughly 2 to 3 dB across a 100 m segment is distinguishable from seasonal noise, though the threshold varies by soil type and local rainfall regime. Treat that figure as indicative, not universal.
Vegetation clearance: NDVI tells you where the machines have been
Before earthworks begin, vegetation clearance is the first detectable signal. Sentinel-2's red-edge and near-infrared bands support NDVI calculation at 10 m resolution. A corridor polygon drawn from the design alignment can be intersected with a time-series NDVI stack to produce a chainage-versus-date matrix showing where canopy has been removed. The 5-day revisit means that, outside prolonged cloud cover, clearance events are typically captured within one to two weeks of occurrence.
This matters for two reasons. First, it gives project financiers and environmental monitors an independent check on whether clearance is staying within the approved right-of-way. Encroachment into adjacent forest or agricultural land shows up as anomalous NDVI loss outside the design corridor buffer. Second, it provides an early-progress signal before any SAR-detectable earthwork has begun. The sequence, clearance then grading then sub-base then pavement, each has a spectral or backscatter signature, and tracking all four gives a more complete construction timeline than any single sensor.
High-resolution optical for milestone verification
Some contractual milestones require photographic evidence rather than a change-detection metric. Culvert installation, bridge deck completion, drainage channel lining and pavement marking all need to be individually visible. At 10 m, Sentinel-2 cannot resolve a 1.2 m culvert headwall. At 50 cm, SkySat can, and at 30 cm, Pléiades Neo can resolve the headwall, the wingwalls and the apron.
The practical workflow is to use the SAR and Sentinel-2 time series to identify which corridor segments have progressed and then task high-resolution optical only over those segments for milestone photography. This concentrates commercial tasking budget where it is needed rather than imaging the entire corridor at high resolution every month. Over a 400 km corridor, selective tasking of 20 to 30 km of active front at any one time is typically far more cost-effective than blanket coverage.
Pléiades Neo tri-stereo tasking adds a further capability: photogrammetric surface models with vertical accuracy in the 0.5 to 1 m range (depending on ground control), sufficient to estimate cut-and-fill volumes at embankments and cuttings. This is not a substitute for ground survey at final payment, but it is a credible independent check on contractor-reported earthwork quantities at interim valuations.
Honest limits: what satellite monitoring cannot tell you
Satellite monitoring of a road or rail corridor is a progress-tracking and anomaly-detection tool. It is not a quality-assurance instrument. Compaction density, bearing capacity, sub-base gradation and drainage performance are not remotely sensed properties. A corridor segment can look geometrically complete in imagery and still fail a plate-bearing test.
Cloud cover in equatorial regions remains the most stubborn operational constraint for optical sensors. During the wet season over parts of Central Africa or insular Southeast Asia, Sentinel-2 may return fewer than two usable acquisitions per month over a given corridor segment. SAR fills much of that gap, but the backscatter-to-progress interpretation requires careful calibration against known ground conditions. Archive depth helps: Sentinel-1 data is available from 2014, which means pre-construction baseline backscatter is almost always available for comparison, even on corridors that began construction recently.
Revisit frequency also sets a floor on how quickly anomalies are caught. A 6-day Sentinel-1 revisit means that a sudden event, an embankment slip, a flood washout, a right-of-way encroachment, may not appear in analysis until several days after it occurs. For time-critical monitoring, commercial SAR constellations (ICEYE, Capella) can reduce that latency to hours, though at higher cost and outside the scope of this page.
Building a monitoring programme for a corridor project
A workable monitoring programme for a 200 to 500 km corridor typically combines three layers. The first is a continuous SAR change-detection layer, updated every 6 to 12 days from Sentinel-1, covering the full corridor and flagging segments where backscatter change exceeds the site-specific threshold. The second is a monthly Sentinel-2 NDVI and bare-soil composite, cloud-masked and gap-filled where possible using temporal interpolation, providing the vegetation-clearance and soil-exposure context. The third is on-demand high-resolution optical tasking, triggered by the first two layers or by contractual milestone dates.
Satellize structures corridor-monitoring analytics around this layered logic, drawing on open Sentinel data and adding commercial tasking on client licence. The Tonga crop-estimation programme demonstrated that time-series change detection over dispersed, hard-to-visit sites is operationally tractable at national scale. Road and rail corridors present the same fundamental challenge: many kilometres, infrequent ground access, and a client who needs to know what is happening before the next site visit.
The output is not a map for its own sake. It is a chainage-referenced progress table, updated on a defined schedule, with flagged anomalies and supporting imagery clips that a project manager or lender's technical adviser can read without specialist training. If that is the deliverable you need, the sensible next step is to share the corridor alignment and the contract milestone schedule so that a monitoring design can be scoped against actual ground conditions.
Typical figures
| Best optical resolution (tasked) | 30 cm (Pléiades Neo); 50 cm (Planet SkySat) |
| Medium-resolution optical | 10 m, Sentinel-2 MSI (bands 2–8A); free, open archive from 2015 |
| SAR resolution | 10 m ground range, Sentinel-1 IW mode; C-band, VV+VH dual polarisation |
| SAR revisit | 6 days (two-satellite Sentinel-1 constellation); 12 days single satellite |
| Optical revisit (open data) | 5 days at equator, Sentinel-2 (cloud permitting) |
| Minimum detectable clearance unit | ~100 m corridor segment for SAR backscatter change; ~30 m for Sentinel-2 NDVI loss |
| Volume estimation vertical accuracy | 0.5–1 m (Pléiades Neo tri-stereo with ground control) |
| Archive depth | Sentinel-1 from April 2014; Sentinel-2 from June 2015; SkySat from 2014 |
| Typical analysis latency | 24–72 hours after satellite acquisition for automated change layers |
| Delivery formats | GeoTIFF change layers, chainage-referenced CSV progress tables, PDF milestone reports |
Analytics Satellize can run
| SAR backscatter change map | Bi-temporal or multi-temporal C-band backscatter differencing (VV, VH, VV/VH ratio); site-specific threshold calibrated against known ground conditions | GeoTIFF layer flagging corridor segments with statistically significant surface-roughness change; updated every 6–12 days |
| Vegetation clearance timeline | NDVI time-series differencing on Sentinel-2 10 m bands; cloud-masked composites; corridor polygon intersection | Chainage-versus-date matrix showing clearance progress and any exceedance of approved right-of-way buffer; monthly GIS layer and PDF summary |
| Pavement-front progress report | Supervised classification of Sentinel-2 and high-resolution optical imagery distinguishing bare earth, compacted sub-base and sealed pavement by spectral signature | Chainage-referenced progress table with percentage of corridor in each construction phase; suitable for lender's technical adviser review |
| Earthwork volume estimate | Photogrammetric DSM differencing from Pléiades Neo tri-stereo acquisitions; cut-and-fill calculation against design DTM | Cut-and-fill volume report per corridor segment, with uncertainty bounds; PDF and CSV; timed to interim valuation dates |
| Right-of-way encroachment alert | NDVI loss or backscatter change detected outside approved corridor buffer polygon; automated spatial intersection with design boundary | Alert feed with coordinates, date and supporting image chip; delivered within 72 hours of acquisition |
| Milestone verification image pack | On-demand SkySat or Pléiades Neo tasking triggered by SAR/Sentinel-2 progress signal or contractual date; orthorectified to <1 m positional accuracy | Annotated image pack with chainage references and feature identification; formatted for inclusion in contract milestone certificates |
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.