Offshore wind foundation scour monitoring from satellite SAR
Seabed scour around monopile foundations threatens structural integrity, yet routine sonar surveys are expensive and infrequent. Satellite SAR detects the surface roughness anomalies that accelerated tidal flow around a scoured foundation produces, providing an early-warning flag between in-situ inspection campaigns.
Sensors
- Sentinel-1 SAR (C-band, 5.405 GHz): Interferometric Wide Swath mode delivers 10 m ground range resolution across a 250 km swath. Repeat pass every 6 days over European waters (12 days globally where only one satellite is operational). C-band backscatter responds to centimetre-scale surface roughness driven by wind-wave interaction with near-surface currents, making it the primary sensor for current-shadow and wake anomaly detection around foundations.
- TerraSAR-X / TanDEM-X (X-band, 9.65 GHz): Stripmap mode achieves 3 m resolution; Spotlight mode reaches approximately 1 m. Higher frequency increases sensitivity to fine-scale surface roughness, improving the signal-to-noise ratio for current anomalies in moderate wind conditions (roughly 3 to 10 m/s). Tasked commercially; not free-to-access.
- Sentinel-2 MSI (optical, 10 m visible bands): In optically shallow water (typically less than 15 m depth), the blue and green bands carry a bathymetry proxy signal. Sediment plumes stirred by scour-driven currents alter water-leaving radiance in ways detectable at 10 m resolution, 5-day revisit. Cloud cover is a hard constraint; this channel is supplementary, not primary.
- Copernicus Marine Service (CMEMS) ocean current model: Provides hindcast and forecast tidal current fields at roughly 1.5 km resolution. Used to condition SAR interpretation: a roughness anomaly aligned with the expected current shadow of a scoured foundation is a meaningful flag; one that appears against the current direction is likely sea-state noise.
What a scoured seabed does to the water column above it
When a monopile foundation is installed in sandy or gravelly seabed, it interrupts tidal flow. The seabed around the pile erodes in a characteristic horseshoe pattern. As the scour hole deepens, the effective cross-section of the water column above it changes. Flow accelerates around the pile base, and the turbulent wake extends further downstream. These are well-documented hydrodynamic effects, studied in physical model tanks and confirmed by in-situ acoustic Doppler measurements at operational wind farms.
The consequence at the surface is subtle but real. Accelerated near-surface currents modulate the short gravity-capillary waves that C-band SAR is sensitive to. Where current speed increases sharply, wave energy is suppressed in the lee of the pile, producing a darker patch in backscatter. Where currents converge or shear, roughness increases, producing a brighter return. The spatial scale of these anomalies, typically tens to a few hundred metres, sits comfortably within Sentinel-1's 10 m resolution, provided the wind speed is in the detectable range.
The wind-speed window that makes detection possible
SAR ocean-surface roughness detection is not uniformly available. Below roughly 2 to 3 m/s wind speed, the sea surface is too smooth for meaningful current-roughness contrast. Above roughly 12 to 14 m/s, wind-driven roughness dominates and swamps the current signal. The usable window is approximately 3 to 12 m/s, which in the southern North Sea and similar offshore wind regions accounts for perhaps 40 to 60 per cent of acquisition opportunities, depending on season.
This is an honest constraint. A single SAR pass cannot confirm scour; a time series of passes, filtered to those acquired within the detectable wind-speed window, is required. With Sentinel-1's 6-day repeat and TerraSAR-X available for targeted tasking, building a 30-pass filtered archive over a foundation cluster takes two to four months. That is long enough to establish a baseline roughness signature for each pile, against which anomalous scenes can be flagged.
What the satellite can and cannot tell you
The satellite method produces a probability flag, not a scour depth. A persistent, current-aligned roughness anomaly around a specific foundation, appearing in multiple filtered SAR passes, is evidence that hydrodynamic conditions around that pile have changed. It is consistent with scour, but it is also consistent with other changes: marine growth on the pile altering drag, a nearby sediment bedform migrating into the wake, or a change in the local bathymetry from a different cause entirely.
Direct scour depth measurement requires a multibeam echosounder or a single-beam sonar survey. That is the only method that gives you the geometry of the scour hole. The satellite's value is in prioritising which foundations warrant an urgent sonar inspection. Operators typically survey tens or hundreds of piles on a fixed schedule; a satellite-derived flag can redirect a survey vessel to a specific pile weeks earlier than the schedule would otherwise allow. That is the realistic use case, and it is a commercially meaningful one.
Optically shallow-water bathymetry as a secondary line of evidence
In water depths below roughly 10 to 15 m, Sentinel-2's blue band (Band 2, centred at 490 nm) penetrates to the seabed. The ratio of blue to green reflectance is a published proxy for relative water depth, used in studies of coral reef and coastal bathymetry mapping. Where scour has deepened the seabed around a pile, a multi-temporal Sentinel-2 comparison can in principle detect a change in the bathymetric proxy signal.
In practice this is difficult. Suspended sediment, phytoplankton, and variable illumination all affect water-leaving radiance. The method works best in clear, calm water, which is rarely the condition around an active scour site. It is most useful as corroborating evidence: if both the SAR roughness anomaly and the Sentinel-2 bathymetric proxy change in the same direction at the same pile, confidence in a real seabed change increases materially.
Building an operational monitoring programme
A credible satellite scour-monitoring programme has three components. First, a baseline archive: every Sentinel-1 scene covering the wind farm from the period before or shortly after commissioning, filtered to the usable wind-speed window using collocated ERA5 or CMEMS reanalysis winds. Second, a per-pile roughness anomaly score, computed as the deviation of each pile's local backscatter from a reference zone in the same scene, normalised by the scene-wide wind-speed proxy. Third, a change-detection threshold, set conservatively to minimise false positives, with alerts issued only when a pile exceeds the threshold in multiple consecutive filtered passes.
Satellize applies this type of SAR time-series analysis on open Sentinel-1 data, with optional TerraSAR-X tasking for higher-resolution confirmation passes on flagged piles. The analytic pipeline is the same class of method used in published oceanographic current-mapping studies. For offshore wind operators managing large foundation portfolios, the output is a monthly ranked list of piles by anomaly score, delivered as a GIS layer and a structured report, ready to brief the marine operations team before the next survey-vessel mobilisation.
Archive depth and the long-term record
Sentinel-1A has been acquiring data since April 2014. For North Sea wind farms commissioned before that date, the archive predates many of the scour events that have since been documented by operators. That decade-long record allows a retrospective analysis: did the SAR roughness anomaly precede the first sonar confirmation of significant scour? If it did, the lead time can be quantified, giving operators a defensible basis for adjusting their survey intervals.
Sentinel-1C, launched in late 2024, maintains the constellation's continuity. The 6-day repeat is preserved. For farms outside European coverage where only one satellite provides coverage, revisit drops to 12 days, which is still sufficient for a monthly monitoring cadence but reduces the number of wind-speed-filtered passes available in any given month.
Typical figures
| Primary sensor spatial resolution | Sentinel-1 IW mode: 10 m range, 10 m azimuth (multi-looked). TerraSAR-X Stripmap: ~3 m; Spotlight: ~1 m |
| Revisit interval | Sentinel-1: 6 days (European waters, two-satellite constellation); 12 days where single satellite. TerraSAR-X: tasked on demand, typically 2 to 4 day revisit achievable |
| Usable wind-speed window for current anomaly detection | Approximately 3 to 12 m/s; outside this range, surface roughness contrast is unreliable |
| Sentinel-1 frequency / polarisation | C-band, 5.405 GHz; VV and VH polarisation in IW mode. VV preferred for ocean surface current signatures |
| Sentinel-2 optical resolution (bathymetry proxy) | 10 m (Bands 2, 3, 4); applicable in optically shallow water typically less than 15 m depth |
| Minimum detectable anomaly scale | Current shadow features of order 50 to 200 m are reliably resolved in Sentinel-1 IW; smaller features may require TerraSAR-X |
| Archive depth | Sentinel-1A from April 2014; Sentinel-2A from June 2015. TerraSAR-X commercial archive from 2008 |
| Ocean current conditioning data | CMEMS tidal current fields at ~1.5 km resolution; ERA5 10 m wind reanalysis at 0.25 degree resolution for scene filtering |
| Delivery format | GeoTIFF anomaly score layers, GeoPackage pile-level time series, PDF monthly report, optional WMS feed |
| Latency from satellite acquisition to alert | Sentinel-1 NRT data available within 1 to 3 hours of acquisition via Copernicus Data Space; processed anomaly score within 24 hours |
Analytics Satellize can run
| Per-pile SAR roughness anomaly score | Local backscatter deviation from scene reference zone, normalised by collocated wind speed proxy (VV NRCS method, published in oceanographic current-mapping literature) | Monthly GeoPackage with anomaly score per foundation, time series plot per pile, wind-speed filter log |
| Filtered SAR pass catalogue | Scene-by-scene wind-speed screening using ERA5 collocated reanalysis; only passes within 3 to 12 m/s window retained for analysis | CSV catalogue of usable passes with acquisition time, mean wind speed, and data quality flag |
| Multi-temporal roughness anomaly map | Stack of filtered Sentinel-1 scenes averaged to reduce speckle; change detection against commissioning-period baseline using per-pixel Z-score | GeoTIFF difference map at 10 m resolution, colour-coded by anomaly magnitude, with pile locations overlaid |
| Sentinel-2 bathymetric proxy change layer | Log-ratio of blue-to-green band reflectance across multi-temporal cloud-free composites; published relative bathymetry method (Stumpf ratio transform) | GeoTIFF showing relative depth change around each foundation; flagged where change exceeds one standard deviation of the baseline period |
| Priority inspection ranking | Composite score combining SAR anomaly persistence (number of flagged passes), anomaly magnitude, and Sentinel-2 corroboration flag; ranked list produced monthly | Structured PDF report with ranked pile list, supporting evidence panels, and recommended sonar survey priority order |
| Retrospective baseline analysis | Full Sentinel-1 archive reprocessing from 2014 to commissioning date; identifies pre-existing seabed current patterns and natural roughness variability at the site | One-off GIS dataset and report establishing site-specific detection thresholds before operational monitoring begins |
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.