Urban noise barrier and acoustic infrastructure mapping
Very-high-resolution optical imagery and airborne LiDAR can locate, classify, and audit roadside noise barriers at city scale, revealing material type and continuity gaps that ground surveys miss or price out.
Sensors
- Pleiades Neo: 30 cm panchromatic, 50 cm multispectral (8 bands including red-edge and deep blue); daily revisit over most urban areas. At 30 cm, a 1-metre-wide concrete wall casts a measurable shadow and occupies multiple pixels, making it detectable and measurable in plan.
- WorldView-3: 31 cm panchromatic, 1.24 m multispectral (8 bands), 29 cm SWIR (8 bands). SWIR bands help discriminate bare earth bunds from concrete or metal panels by exploiting differences in shortwave reflectance, though this remains a proxy rather than a definitive material identification.
- Airborne LiDAR: Point densities of 4 to 50 points per square metre are routinely achieved in urban surveys. At 8 pts/m² and above, a Digital Surface Model derived from the point cloud resolves barrier height to roughly ±15 cm vertically, enabling height profiling along the full barrier length and detection of gaps or step-downs.
- SPOT 7: 1.5 m panchromatic, 6 m multispectral. Insufficient resolution to detect individual barrier panels, but useful for corridor delineation and change detection over multi-year periods when a more precise baseline already exists.
Why planning authorities keep getting this wrong
Noise barriers along motorways and ring roads are legally mandated infrastructure in most jurisdictions with environmental impact assessment requirements. Yet inventories of what actually exists, where gaps occur, and whether barriers meet specified heights are almost universally out of date. Ground surveys are slow and expensive; they also cannot easily capture the plan-view geometry needed to assess continuity. A barrier that looks solid from the road may have a 10-metre gap at an access ramp that undermines its acoustic performance entirely.
Satellite and airborne data change the economics of this problem. A city-scale corridor audit that would take a ground team months can be completed analytically in days, with repeatable geometry that ground observers cannot match. The catch is that the physics of detection sets real limits, and understanding those limits is what separates useful analysis from confident nonsense.
What a floating roof gives away
At 30 cm resolution, a noise barrier appears in imagery as a linear feature with a characteristic shadow. The shadow length, combined with the solar elevation angle at acquisition time, gives a direct estimate of barrier height. This is not a proxy: it is basic trigonometry applied to a measurable pixel extent. Pleiades Neo and WorldView-3 both support this approach reliably for barriers taller than roughly 1.5 metres, provided the acquisition is not nadir (straight down), so that shadows fall to one side. Off-nadir angles of 15 to 25 degrees are generally optimal.
Material discrimination is harder and more honest about its limits. Earth bunds have a characteristic spectral signature in the near-infrared and SWIR: lower reflectance, often with vegetation establishing on the slope, and a rounded profile in LiDAR cross-sections. Hard walls, whether concrete, timber, or metal, tend to show higher visible-band reflectance, sharper edges in the DSM, and distinct specular returns in some LiDAR configurations. WorldView-3's SWIR bands add useful separation, but the classification is probabilistic. Expect accuracies in the 80 to 90 percent range for a two-class bund-versus-wall distinction, not certainty.
Height profiling along the barrier length
LiDAR is where the analysis becomes genuinely rigorous. A normalised Digital Surface Model, produced by subtracting a bare-earth Digital Terrain Model from the full DSM, gives barrier height above ground at every point along its length. At 8 points per square metre, the vertical precision of a well-calibrated airborne survey is typically ±10 to 20 cm. That is sufficient to flag sections where a nominally 3-metre barrier has settled or been modified to 2.4 metres, which may matter for compliance with the design specification.
The limitation is coverage and currency. Airborne LiDAR surveys are expensive to commission and are rarely repeated more than once every three to five years in most cities. Where a recent survey exists, it is the most powerful input available. Where it does not, the analysis falls back on shadow-based height estimation from VHR optical imagery, which is less precise but still useful for flagging anomalies worth ground-checking.
Detecting gaps: the continuity problem
A gap in a noise barrier is, acoustically, a near-total failure of that section. Sound diffracts around gaps in ways that make even a short interruption disproportionately damaging to the protected zone. From a planning compliance perspective, gaps at access ramps, utility crossings, and property boundaries are the most common defects and the hardest to catch without a systematic spatial analysis.
Object-based image analysis (OBIA) is the standard published method for this task. The barrier is first extracted as a linear object using spectral and geometric rules. The extracted centreline is then checked for continuity: any break longer than a threshold, typically 2 to 5 metres depending on the specification, is flagged as a gap candidate. False positives arise where vegetation overhangs the barrier and interrupts the spectral signal, or where a road bridge crosses the corridor. These require manual review, and any honest workflow acknowledges that automated gap detection at sub-5-metre precision still needs a human check before it becomes a compliance finding.
From raw data to a GIS layer a planning officer can use
The deliverable that actually gets used is a vector layer: a polyline for each barrier segment, attributed with estimated height, material class, continuity status, and a confidence flag. That layer drops directly into any standard GIS environment and can be overlaid on cadastral data, road centrelines, and the original planning consent geometries to identify where the as-built condition diverges from the approved design.
Satellize runs object-based barrier extraction and height profiling as part of its urban analytics stack. The workflow is the same one used for linear infrastructure in the Tonga crop-estimation programme, adapted from agricultural field boundaries to built linear features. Outputs are delivered as GeoPackage or shapefile, with an accompanying report summarising gap locations and height anomalies by road corridor. If a client already holds an airborne LiDAR dataset, that is the preferred input. If not, Pleiades Neo tasking is arranged through commercial licence.
Honest limits, stated plainly
Cloud cover blocks optical acquisition entirely. In temperate climates, getting a usable cloud-free image over a specific urban corridor may require multiple tasking attempts across several weeks. LiDAR is unaffected by cloud but is collected by aircraft, not satellite, so it is a commissioned survey rather than an on-demand product.
Very low barriers, below about 1 metre, are at the edge of reliable detection from space at current commercial resolutions. Barriers obscured by tree canopy are a persistent problem: the canopy registers in the DSM, not the barrier beneath it. And material classification from spectral data alone will always carry uncertainty; it is a guide to where ground inspection is warranted, not a substitute for it. Any programme using this analysis for formal compliance decisions should treat the satellite output as a screening tool that prioritises field resources, not as a final audit record.
Typical figures
| Spatial resolution (optical) | 30 cm (Pleiades Neo panchromatic); 31 cm (WorldView-3 panchromatic) |
| Spatial resolution (LiDAR DSM) | 12.5 to 25 cm grid, from point densities of 8 pts/m² and above |
| Vertical height precision (LiDAR) | ±10 to 20 cm for well-calibrated airborne surveys |
| Minimum detectable barrier height (optical shadow method) | Approximately 1.5 m at 30 cm resolution with off-nadir acquisition |
| Minimum detectable gap length | Approximately 2 to 5 m (OBIA method; sub-5 m gaps require manual validation) |
| Revisit (Pleiades Neo constellation) | Daily over most urban areas; cloud-free acquisition may require multiple attempts in temperate climates |
| Spectral bands used | Visible, NIR, red-edge, deep blue (Pleiades Neo); SWIR bands 1–8 (WorldView-3) for material proxy |
| Archive depth (WorldView-3) | Operational since 2014; significant archive over major cities |
| Delivery format | GeoPackage or shapefile (vector); GeoTIFF (raster DSM or orthoimage); PDF summary report |
| Coverage per tasking | Pleiades Neo single scene: 100 km² at nadir; strip acquisitions cover full urban corridors in one pass |
Analytics Satellize can run
| Barrier presence and centreline mapping | Object-based image analysis (OBIA) on VHR panchromatic and multispectral imagery; spectral and geometric rule sets for linear feature extraction | Vector polyline layer of all detected barrier segments, attributed with location and length, delivered as GeoPackage |
| Barrier height profile | Shadow-length trigonometry from off-nadir VHR optical imagery; or normalised DSM differencing from airborne LiDAR point cloud | Height attribute on each barrier segment polyline; longitudinal height profile chart per road corridor in PDF report |
| Material proxy classification (earth bund vs. hard wall) | Supervised spectral classification using NIR and SWIR bands (WorldView-3); LiDAR cross-section profile shape as secondary discriminant | Material class attribute on barrier vector layer, with per-segment confidence score |
| Continuity gap detection | Centreline break analysis on extracted barrier polylines; gaps flagged where interruption exceeds configurable threshold (default 3 m) | Point layer of gap locations with estimated gap length; ranked priority list for field verification in PDF report |
| Compliance divergence report | Spatial overlay of extracted barrier geometry and height against planning consent polygons and design-specification attributes | Table of non-conforming sections with coordinates, measured vs. specified height, and gap status; suitable for submission to planning authority |
| Multi-date change detection | Bitemporal OBIA comparison of barrier extent and height between two acquisition dates; change flagged where barrier length decreases or height drops beyond threshold | Change vector layer highlighting modified or removed sections; summary statistics in PDF |
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.