Offshore wind farm turbine installation progress for project-finance drawdown
SAR and optical satellites track the sequential installation of offshore wind turbines structure by structure, giving project-finance lenders independent milestone verification that does not depend on a developer's own reporting.
Sensors
- Sentinel-1 SAR (ESA): C-band SAR at 10 m resolution in Interferometric Wide Swath mode, 6-day repeat at mid-latitudes with both satellites active. Penetrates cloud and darkness, distinguishing installed steel structures from open water by strong radar backscatter. Free and open archive from 2014.
- ICEYE SAR constellation: X-band SAR with Spot mode down to approximately 1 m and Strip mode at roughly 3 m. Tasked on demand; typical revisit to a specific offshore site within 24 hours. Higher resolution allows separation of monopile, transition piece and tower sections that Sentinel-1 conflates into a single bright point.
- Planet SkySat: 0.5 m optical imagery, tasked on demand with same-day collection possible. On clear days resolves nacelle geometry and rotor installation. Cloud-limited in the North Sea and Baltic; useful for confirmation passes after SAR flags a change.
- Pleiades Neo (Airbus): 30 cm panchromatic, 1.2 m multispectral, stereo collection available. Can distinguish whether a nacelle and rotor assembly are present on a tower, which SAR alone at 10 m cannot confirm. Agile tasking but equally cloud-dependent.
Why lenders need a count they did not get from the developer
Project-finance structures for offshore wind typically release debt tranches against verified physical milestones: foundation installation complete, turbines mechanically complete, first power. Each milestone may represent hundreds of millions in drawdown. The developer's own progress reports are not independent, and site visits by lender representatives to an active offshore construction zone are logistically awkward and expensive.
Satellite observation sidesteps both problems. A SAR pass costs a fraction of a helicopter charter and produces a spatially explicit record: which grid positions are occupied, which are not, and when each transition happened. The archive is timestamped and immutable. For a lender, that is a different class of evidence from a contractor's certificate.
What the physics of steel in open water gives you for free
A steel monopile rising from the sea surface is a near-perfect radar reflector. In C-band SAR imagery such as Sentinel-1, an installed foundation produces a bright point target against the low-backscatter background of open water. At 10 m resolution and with typical turbine spacing of 500 to 1,000 m, individual structures are well separated and individually countable. The method does not require machine learning for the basic count; template matching against a known grid layout is sufficient and has been validated in published literature on wind-farm SAR mapping.
X-band systems like ICEYE improve on this in two ways. First, the shorter wavelength interacts differently with structural geometry, making it easier to distinguish a bare monopile from one with a transition piece and tower attached. Second, the roughly 3 m Strip mode resolution starts to resolve the physical profile of the structure rather than treating it as a point. Neither C-band nor X-band SAR can reliably confirm nacelle installation versus a bare tower top without sub-metre resolution or a supporting optical pass. That ambiguity is real and should be stated plainly in any milestone certificate.
The cloud problem, and why SAR is not a complete answer
The North Sea averages cloud cover exceeding 70 percent of days across most months. The Baltic is only marginally better. Optical imagery from SkySat or Pleiades Neo is genuinely useful but cannot be relied upon for time-sensitive milestone verification in these environments. SAR is the primary workaround, and for foundation and tower counting it works well.
The honest limit is the nacelle. A nacelle and rotor assembly is roughly 15 to 20 m wide on a modern offshore turbine. Sentinel-1 at 10 m cannot resolve it against the tower. ICEYE at 3 m can detect the change in radar cross-section but the interpretation is not unambiguous. Confirming full mechanical completion of a turbine, as opposed to confirming that a steel structure is present, requires either a sub-metre SAR pass, a clear-sky optical pass, or a combination of both. Any milestone definition that includes nacelle installation should specify this and set a protocol for what happens when cloud persists.
Installation vessels add a further complication. A jack-up rig positioned over a foundation during active work can partially mask the foundation's radar signature and makes it impossible to determine the installation state of that specific position until the vessel moves. Typically this affects only one to three positions at any given time across a large array, but it should be accounted for in the verification protocol.
Building the verification timeline
The practical workflow starts before construction. A baseline SAR image of the empty site, combined with the approved layout plan, establishes the expected grid of positions. Each subsequent pass is co-registered to that grid and each position is classified as empty, partially occupied, or fully occupied. The classification history for each position forms a per-turbine installation log with date ranges.
Sentinel-1's six-day repeat provides a free backbone. Commercial tasking from ICEYE fills gaps or provides higher-confidence classification at milestone dates. For a 100-turbine array, a typical project-finance lender might want independent verification at 25 percent, 50 percent, 75 percent and full mechanical completion. Each verification report maps the state of every grid position on the milestone date, flags any positions where vessel interference or cloud prevented confident classification, and states the method and imagery sources used. That last part matters: a report that does not document its own uncertainty is not a useful financial instrument.
Satellize's role and an honest scope statement
Satellize runs this analysis on open Sentinel-1 data supplemented by commercial tasking arranged on client licence, producing milestone verification reports as structured GIS deliverables with supporting imagery mosaics. The analytical approach is the same class of change-detection and object-classification method used in Satellize's crop-estimation work for the Kingdom of Tonga: systematic, grid-referenced, with explicit confidence flags rather than false precision.
What satellite analysis cannot replace is a physical inspection for quality assurance, grouting confirmation, or electrical commissioning status. The satellite record answers one question cleanly: is a structure present at this grid position on this date? For a drawdown milestone defined around physical installation counts, that is exactly the question the lender needs answered independently.
Typical figures
| Primary SAR resolution (Sentinel-1 IW) | 10 m ground range, 20 m azimuth (IW mode) |
| Commercial SAR resolution (ICEYE Strip / Spot) | ~3 m Strip, ~1 m Spot |
| Optical resolution (Pleiades Neo / SkySat) | 0.3 m pan (Pleiades Neo); 0.5 m (SkySat) |
| Sentinel-1 revisit at North Sea latitudes | 6 days with both satellites; 3 days possible with orbit planning |
| Commercial SAR tasking latency | ICEYE: typically within 24 hours of tasking request |
| Minimum detectable structure | Steel monopile (~5 m diameter) reliably detected in Sentinel-1 C-band against open water |
| Turbine spacing (typical offshore arrays) | 500 to 1,000 m; individual structure separation unambiguous at 10 m resolution |
| Sentinel-1 archive depth | From 2014 (Sentinel-1A launch); covers most North Sea and Baltic projects from foundation phase |
| Delivery formats | GeoTIFF annotated mosaics, GeoJSON position-state layer, PDF milestone report |
| Cloud penetration | SAR: unaffected. Optical: cloud-dependent; North Sea cloud cover exceeds 70% of days |
Analytics Satellize can run
| Baseline grid registration | Co-registration of approved layout plan to SAR coordinate frame; point-target extraction per grid position | GeoJSON grid file with position IDs, expected coordinates and initial state; used as reference for all subsequent passes |
| Per-pass structure count | Bright-target detection and template matching against registered grid; each position classified as empty, occupied or uncertain (vessel/cloud interference) | Timestamped GeoJSON layer per acquisition with per-position state and confidence flag |
| Installation timeline log | Change-point detection across the SAR time series; first-occupied date estimated per position with uncertainty range from revisit interval | CSV table of per-turbine first-detection dates and confidence intervals, suitable for lender audit trail |
| Milestone verification report | Snapshot classification at lender-specified milestone date using best available imagery; SAR primary, optical supplementary where cloud-free | PDF report with annotated imagery mosaic, structure count, uncertainty statement and imagery metadata |
| Nacelle-installation confirmation | Sub-metre SAR or optical change detection comparing tower-top radar cross-section or visual profile before and after nacelle lift; flagged as requiring clear-sky optical or ICEYE Spot pass | Per-position nacelle status layer with explicit data-source and confidence notation |
| Vessel interference flag | Automatic detection of large vessel signatures (jack-up rigs typically 80 to 150 m) within one turbine spacing of a grid position; affected positions excluded from count with explanation | Alert layer appended to per-pass deliverable; positions re-classified on next clear pass |
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.