Wind turbine location mapping for microwave link interference assessment

Wind turbines scatter microwave energy unpredictably, threatening licensed backhaul links. SAR amplitude imagery and optical stereo extract hub coordinates and tower heights precisely enough to feed Fresnel-zone and interference models before a link is commissioned.

Sensors

Sentinel-1 (ESA, C-band SAR): 10 m ground-range resolution in Interferometric Wide Swath mode; 6-day repeat at mid-latitudes with both satellites active. Turbine towers produce strong, persistent point-like backscatter returns that are detectable even through cloud and at night. Rotor blades introduce Doppler artefacts that aid discrimination from other tall structures.
TerraSAR-X (DLR/Airbus, X-band SAR): Spotlight mode delivers 1–2 m resolution, resolving individual tower bases and shadow lengths for height estimation. X-band is more sensitive to blade geometry than C-band. Tasked commercially; revisit is typically 2–11 days depending on latitude and acquisition plan.
Pleiades Neo (Airbus, optical): 30 cm panchromatic resolution with stereo and tri-stereo tasking. Photogrammetric processing of stereo pairs yields digital surface models accurate to within 1–2 m vertically, sufficient for hub-height and tip-height extraction needed in Fresnel-zone clearance calculations. Cloud-limited; tasked on demand.
WorldView-3 (Maxar, optical): 31 cm panchromatic, 1.24 m multispectral. Stereo pairs support similar DSM workflows to Pleiades Neo. The 29-band SWIR suite is not relevant here; the value is geometric fidelity and archive depth going back to 2014.

Why a spinning rotor is a microwave engineer's problem

A wind turbine is not a passive obstacle. The rotating blades of a modern 3 MW-class turbine sweep a rotor diameter of 100–130 m and reflect incident microwave energy in directions that shift with blade position, wind speed and pitch angle. For a licensed point-to-point backhaul link operating in the 6–42 GHz bands, a turbine within the first Fresnel zone of the path can introduce amplitude fading, phase disturbance and, in severe cases, persistent interference that degrades link availability below the ITU-R performance objectives the operator is licensed to meet.

Ofcom in the UK, Anatel in Brazil and equivalent regulators in most jurisdictions require interference assessments before new wind farms are consented near existing licensed links, and before operators commission new links near existing farms. Both workflows need the same input: accurate turbine positions, hub heights and rotor dimensions. Planners who rely on developer-supplied coordinates alone risk working from pre-construction design data that may not reflect as-built positions, which can differ by tens of metres.

What a SAR image gives away about a turbine

Steel towers with large horizontal rotor assemblies are among the most distinctive point targets in SAR imagery. In Sentinel-1 IW mode, a turbine typically appears as a bright pixel cluster with a geometric shadow extending in the look direction. The shadow length, combined with the known incidence angle (typically 20–46° in IW mode), yields tower height estimates accurate to roughly 5–10 m at 10 m resolution. That is adequate for initial screening across a wide area but not for Fresnel-zone calculations, which demand better.

TerraSAR-X Spotlight at 1–2 m resolution sharpens that considerably. Shadow-length inversion at X-band has been validated in published literature against surveyed turbine heights with residuals under 3 m for towers above 60 m. Doppler anomalies from rotating blades also provide a discriminator: a static lattice tower and a wind turbine can look similar in amplitude, but the turbine introduces a characteristic spectral smear in focused SAR products that automated classifiers can exploit.

Where optical stereo earns its place

SAR shadow-based height retrieval degrades when turbines cluster together, when shadows overlap, or when the incidence geometry is unfavourable. Optical stereo sidesteps those limitations by treating the nacelle as a photogrammetric target. A Pleiades Neo stereo pair, processed through a standard dense-matching pipeline, produces a digital surface model from which hub height and tip height can be read directly. Airbus documents vertical accuracy of 1–2 m for Pleiades Neo stereo over open terrain, which is the relevant figure for rotor-tip clearance calculations.

The honest constraint is cloud. Optical stereo requires clear sky over the target on the acquisition date. In persistently cloudy climates, such as the Scottish Highlands or coastal Norway where wind resource is high, achieving a usable stereo pair may require multiple tasking attempts across several weeks. Combining a SAR-derived position fix with a single-pass optical height measurement is often the practical compromise: SAR locates the turbine to within a few metres in plan, and a cloud-free optical image, even monoscopic, can confirm hub height by measuring the tower shadow with known sun elevation.

Building the interference input dataset

The output that a microwave planning engineer actually needs is a georeferenced table: easting, northing, hub height above ground level, rotor diameter, and ideally tower base elevation from a digital terrain model. From those five numbers, a propagation tool such as Pathloss, ICS Telecom or a custom ITU-R P.452 implementation can compute the worst-case scattering geometry and estimate the excess interference margin required on the link.

SAR amplitude detection identifies turbine candidates across a 250 km swath in a single Sentinel-1 pass. A change-detection comparison between two Sentinel-1 epochs separated by a year or more flags newly erected turbines that may not appear in planning registers or operator databases. This is where the satellite method adds value that no static dataset can replicate: wind farm construction in many countries outpaces the update cycle of national mapping agencies, and a licensed-link operator may be unaware of a turbine erected 800 m off their path until link performance degrades.

The method cannot predict future turbine positions. Planning-register data from national authorities must be ingested separately and treated as a distinct, uncertain layer. Satellize incorporates that register data as a supplementary input rather than a satellite-derived product.

Resolution floors and what they mean in practice

Sentinel-1 at 10 m resolution will not resolve individual turbines within a tightly spaced array where inter-turbine spacing falls below roughly 200 m. Modern onshore wind farms typically space turbines at 3–5 rotor diameters, so 300–650 m separation is common and individual detection is reliable. Offshore arrays sometimes use tighter rows; there, TerraSAR-X or Pleiades Neo is the appropriate primary sensor.

Rotor diameter cannot be extracted from SAR amplitude alone with useful precision. It must be inferred from turbine model identification (possible from optical imagery if the nacelle markings are legible at 30 cm resolution) or sourced from manufacturer datasheets matched to the detected hub height. A 2 m hub-height error translates to a first Fresnel-zone radius error of under 0.5 m at typical backhaul frequencies and path geometries, which is negligible. The dominant uncertainty in interference models is usually the assumed blade orientation at the moment of worst-case scattering, not the positional accuracy of the turbine itself.

From satellite data to a licensed-link decision

A typical workflow for a national mobile operator assessing a new 18 GHz backhaul route runs as follows. Sentinel-1 archive scenes covering the corridor are processed to detect and geolocate all turbine returns within 2 km of the path centreline. Detections above a confidence threshold trigger a Pleiades Neo or TerraSAR-X tasking order for height confirmation. The resulting turbine dataset feeds an ITU-R P.452 scattering model, and the output is a link margin assessment that the operator submits to the regulator alongside the licence application.

Satellize runs this pipeline for clients who need a defensible, auditable interference dataset rather than a manual desk study. The analytics approach is the same one used in the company's broader remote-sensing work, including the Tonga crop-estimation programme, adapted here for a geometry problem rather than a vegetation-index problem. The deliverable is a GIS layer and an accompanying technical report structured to meet the evidence requirements of the relevant national frequency regulator. If the regulator subsequently asks for updated data after a new turbine is erected nearby, a Sentinel-1 change-detection run can be turned around in days rather than weeks.

Typical figures

Plan position accuracy (Sentinel-1 IW)	~10 m (pixel spacing); sub-pixel interpolation can improve to ~5 m
Hub height accuracy (Pleiades Neo stereo DSM)	1–2 m vertical (open terrain, per Airbus published specification)
Hub height accuracy (SAR shadow inversion, TerraSAR-X Spotlight)	±3 m for towers above 60 m (published validation range)
Sentinel-1 revisit (dual-satellite)	6 days at mid-latitudes; archive from 2014
TerraSAR-X tasked revisit	2–11 days depending on latitude and acquisition plan
Pleiades Neo optical resolution	30 cm panchromatic; tasked on demand, cloud-dependent
Minimum detectable turbine height (Sentinel-1)	~40 m tower (approximate; shorter structures may not produce reliable shadow at 10 m resolution)
Swath coverage per Sentinel-1 IW pass	250 km wide; continental-scale screening in a single acquisition
Delivery format	GeoPackage or Shapefile (turbine point layer with attributes); GeoTIFF DSM; PDF technical report
Sensor frequency bands	Sentinel-1: C-band (5.405 GHz); TerraSAR-X: X-band (9.65 GHz); Pleiades Neo / WorldView-3: visible and near-infrared (optical)

Analytics Satellize can run

Turbine detection and geolocation layer	SAR amplitude thresholding and point-target detection on Sentinel-1 IW scenes; persistent bright-scatterer identification across multi-temporal stack	GIS point layer with easting, northing, detection confidence score and acquisition date
Tower and hub height extraction	SAR shadow-length inversion using known incidence angle (TerraSAR-X Spotlight) and/or photogrammetric DSM differencing from Pleiades Neo or WorldView-3 stereo pairs	Attribute table appended to turbine point layer: hub height AGL, estimated tip height, method flag, uncertainty estimate
Rotor diameter estimation	Optical image measurement of rotor shadow or nacelle geometry at 30 cm resolution; cross-referenced against published turbine model databases where nacelle markings are legible	Rotor diameter field in turbine attribute table, with source flag (measured vs. inferred)
New turbine change detection	Bi-temporal or multi-temporal Sentinel-1 amplitude differencing to identify scatterers present in recent epoch but absent in baseline; flagged candidates reviewed against Pleiades Neo tasking	Change-detection alert report listing newly detected turbines with coordinates and estimated erection date range
Fresnel-zone clearance assessment dataset	Geometric computation of first Fresnel-zone radius at each turbine position along a defined link path, using extracted hub and tip heights combined with a terrain DTM; ITU-R P.526 geometry	Per-turbine clearance table showing encroachment status (clear / marginal / obstructed) for each defined link path
Corridor-wide interference screening report	Batch processing of turbine dataset against a portfolio of licensed link paths; scattering geometry assessed per ITU-R P.452 methodology	Ranked list of link paths by interference risk, with supporting maps and a structured technical report formatted for submission to national frequency regulator

Sources

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.