Dredging Vessel Activity and Sediment Plume Detection
Active dredging leaves two signatures: a vessel crawling at 1–3 knots and a turbid sediment plume visible in multispectral imagery. Combining SAR vessel detection, AIS kinematics and optical water-quality analysis lets analysts distinguish permitted harbour works from covert land reclamation or illegal seabed extraction.
Sensors
- Sentinel-2 MSI: 10 m resolution in visible and NIR bands (B2–B8A), 20 m in red-edge and SWIR. Five-day revisit at the equator with two satellites. Elevated total suspended matter in the water column raises reflectance measurably in Band 4 (red, 665 nm) and Band 8 (NIR, 842 nm). Cloud cover is the principal limit; coastal haze compounds it.
- Landsat-9 OLI: 30 m multispectral resolution, 16-day repeat at a given path/row but offset from Landsat-8 to give roughly 8-day combined revisit. Band 4 (red) and Band 5 (NIR) respond to suspended sediment concentration. Coarser than Sentinel-2 but the archive runs back to 1972 across the full Landsat series, which matters for long-run reclamation baseline studies.
- Sentinel-1 SAR (C-band): C-band synthetic aperture radar at 5.405 GHz. Interferometric Wide swath mode delivers 10 m resolution across a 250 km swath, all-weather, day and night. Detects vessel hulls as bright radar returns against the sea surface. A dredger's characteristic slow-speed circular or back-and-forth pattern over a work site is distinctive in a time series of detections.
- Spire spaceborne AIS: Spire operates more than 100 low-Earth-orbit satellites carrying AIS receivers. Global AIS message latency is typically under 30 minutes. Hopper dredgers and cutter-suction dredgers transmitting AIS report speed over ground (usually 0.5–3 knots when working) and MMSI-linked vessel type codes. Vessels operating without AIS, or with suppressed transponders, appear in SAR but not in AIS, which is itself a signal.
What the water column gives away
When a cutter-suction or trailing-suction hopper dredger disturbs the seabed, it suspends fine sediment particles that scatter sunlight differently from clear water. Total suspended matter (TSM) concentrations during active dredging can reach tens to hundreds of milligrams per litre, well above background coastal values. That concentration shifts the spectral reflectance of the water surface upward in the red and near-infrared bands, producing a pale or milky signature that Sentinel-2 MSI and Landsat-9 OLI can detect at 10–30 m resolution.
The physics is straightforward: suspended inorganic particles have high backscattering coefficients in the red portion of the spectrum, where phytoplankton absorption is low. Algorithms such as the Nechad et al. TSM retrieval, published in Remote Sensing of Environment, convert band-ratio reflectance values into calibrated sediment concentration estimates. The plume is often larger and longer-lived than the vessel itself, persisting for hours after dredging pauses, which means a single cloud-free overpass can document an event even if the vessel has moved on.
Matching the vessel to the disturbance
Optical imagery shows where the plume is. It does not, on its own, identify which vessel caused it or whether that vessel was operating under a permit. Sentinel-1 SAR fills that gap. A dredger working a defined cut appears in SAR as a bright point target moving at 1–3 knots, often in a systematic lawnmower pattern or stationary over a single position. Repeated Sentinel-1 passes, typically every six days in mid-latitudes with a single satellite and every three days with both, build a kinematic record of the vessel's behaviour.
Spire AIS data adds the identity layer. A vessel transmitting AIS with a dredger type code (IMO type 33) and a speed-over-ground below 3 knots, whose position coincides with a SAR detection and a downstream optical plume, is a high-confidence dredging event. The absence of AIS at a SAR-detected slow-moving target near a known seabed resource area is a different kind of signal entirely.
Permitted works versus covert extraction
Most large dredging operations are permitted, logged and publicly known: harbour deepening, land reclamation for airports or industrial zones, beach nourishment. The monitoring value for those clients is compliance verification and environmental impact tracking, specifically whether the sediment plume stays within the permitted dispersion boundary specified in the environmental consent.
The harder problem is unauthorised extraction. Illegal sand and aggregate dredging is documented across South-East Asia, West Africa and parts of the Mediterranean. Vessels engaged in it often suppress or falsify AIS, work at night and avoid predictable patterns. SAR detects them regardless of AIS status. Night-time optical imagery from VIIRS is too coarse (375 m) to resolve a single dredger, but a persistent SAR anomaly in a zone with no permit record, combined with a daytime plume detection, builds a case. The honest caveat: SAR cannot distinguish a dredger from a similarly sized slow-moving fishing vessel on a single pass. A time series of kinematic behaviour, ideally three or more passes showing the same pattern in the same location, is needed before the classification is reliable.
Limits that matter for procurement decisions
Cloud cover is the dominant constraint on optical plume detection. Sentinel-2 and Landsat-9 are passive sensors; a persistent cloud deck over a tropical port can produce gaps of two to three weeks in usable imagery. SAR is unaffected by cloud, but it detects the vessel, not the plume. A monitoring programme that relies on optical alone will have blind periods. One that relies on SAR alone cannot quantify sediment dispersion.
Spatial resolution sets a floor on what is detectable. Sentinel-2 at 10 m can resolve a plume that is at least a few tens of metres wide, which covers most active dredging scenarios. Very small-scale disturbances from artisanal or small-vessel extraction may fall below the detection threshold. Landsat-9 at 30 m is better suited to large-area baseline mapping than to detecting small or short-duration events. Commercial very-high-resolution optical satellites (sub-metre) can resolve individual vessel details, but their tasking cost and narrow swath mean they are best used for targeted confirmation rather than routine surveillance.
From detection to regulatory evidence
A detection is not a prosecution. Turning satellite observations into actionable regulatory evidence requires a documented chain: the raw imagery, the processing method, the derived TSM or vessel-detection output, and the comparison against the permit record and AIS log. Each step needs to be reproducible and traceable to a published algorithm or calibrated sensor specification.
Satellize structures its dredging-monitoring analytics around exactly that chain, running TSM retrieval on Sentinel-2 and Landsat-9 imagery and correlating outputs with Sentinel-1 vessel detections and Spire AIS records. The Tonga crop-estimation programme demonstrated the same principle in a different domain: that open-constellation data, processed consistently and delivered as calibrated outputs rather than raw files, is what a government client can actually act on. For a port authority or maritime regulator, the deliverable is a dated event log with georeferenced imagery attachments, not a pixel stack.
Archive depth and what it reveals about reclamation history
The Landsat archive extends to 1972. Sentinel-2 coverage begins in 2015. Together they provide a multi-decade record of coastline change that can reconstruct the history of land reclamation projects, including phases that predate any current permit regime. Change detection across that archive, using normalised difference water index (NDWI) to map the land-water boundary over time, can quantify how much area was reclaimed, in what sequence and at what rate.
This is not a trivial capability. Several contested reclamation disputes in international arbitration have turned on questions of when and how quickly particular features were constructed. A calibrated, sensor-documented time series from publicly archived satellites carries evidential weight that anecdotal reporting does not. The archive is free and open; the analytical rigour in processing it is where the value sits.
Typical figures
| Optical spatial resolution (plume detection) | 10 m (Sentinel-2 MSI visible/NIR); 30 m (Landsat-9 OLI) |
| SAR spatial resolution (vessel detection) | 10 m (Sentinel-1 IW mode, single-look complex) |
| Optical revisit (cloud-free) | 5 days at equator (Sentinel-2A+B combined); ~8 days (Landsat-8+9 combined) |
| SAR revisit | 6 days per satellite; 3 days combined (Sentinel-1A+1B where both operational) |
| AIS latency (Spire) | Typically under 30 minutes for global spaceborne AIS |
| Spectral bands used for TSM retrieval | Red (Sentinel-2 B4, 665 nm; Landsat-9 B4, 655 nm) and NIR (Sentinel-2 B8, 842 nm; Landsat-9 B5, 865 nm) |
| SAR frequency | C-band, 5.405 GHz (Sentinel-1) |
| Minimum detectable plume width (optical) | Approximately 20–30 m in favourable conditions (Sentinel-2); larger plumes more reliably detected |
| Archive depth | Sentinel-2 from 2015; Landsat from 1972 (Landsat-1 MSS to present) |
| Delivery formats | GeoTIFF (TSM concentration maps), GeoJSON (vessel detections and AIS tracks), dated event log (PDF or structured JSON) |
Analytics Satellize can run
| Total suspended matter (TSM) concentration map | Semi-analytical or empirical band-ratio retrieval (e.g. Nechad et al. approach) applied to Sentinel-2 B4/B8 or Landsat-9 B4/B5 surface reflectance | Georeferenced GeoTIFF per overpass showing TSM in mg/L, with plume boundary polygon and peak-concentration value |
| Dredging vessel kinematic profile | SAR vessel detection (CFAR or equivalent) cross-referenced with Spire AIS speed-over-ground and MMSI vessel-type code; slow-speed pattern classification | GeoJSON track file per vessel per period, flagged by speed threshold and spatial pattern (linear cut, circular, stationary) |
| Dark-vessel alert at dredging-risk sites | SAR detection without corresponding AIS record within a defined radius of known seabed resource zones or unpermitted areas | Alert report with SAR chip, detection coordinates, timestamp and nearest permitted-work boundary distance |
| Plume dispersion boundary compliance check | TSM map clipped to permitted dispersion zone polygon; pixel-count exceedance calculation against consent threshold concentration | Compliance summary table per overpass, with exceedance area in hectares and annotated imagery for regulatory submission |
| Long-run coastline and reclamation change detection | NDWI time series across Landsat archive (1972 to present) and Sentinel-2 (2015 to present); land-water boundary extraction per epoch | Change map GeoTIFF with reclamation area polygons, dated by first-detection epoch, and area-over-time chart |
| Event log for regulatory or legal use | Fusion of TSM detections, SAR vessel positions, AIS records and permit database; timestamped and source-attributed | Structured PDF or JSON event log with georeferenced imagery attachments, suitable for submission to a port authority or arbitration proceeding |
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.