Ballast Water Discharge and Turbidity Plume Detection
Ballast discharge plumes alter water-leaving radiance in ways multispectral and hyperspectral sensors can measure. Cross-referencing plume geometry with AIS tracks attributes events to specific vessels, giving port authorities satellite-grade evidence.
Sensors
- Sentinel-2 MSI: 10 m resolution in visible bands (B2, B3, B4) and 20 m in red-edge and SWIR bands; 5-day revisit at the equator with both satellites. Bands 2-4 and B8A resolve suspended sediment concentrations and chlorophyll anomalies in coastal water; the 60 m B1 coastal aerosol band aids atmospheric correction over water.
- PACE OCI (Ocean Color Instrument): Hyperspectral coverage from 340 to 890 nm at roughly 5 nm spectral resolution and approximately 1 km spatial resolution; daily global revisit. Designed explicitly for ocean colour science, OCI can discriminate phytoplankton community composition and suspended particulate matter in ways broadband sensors cannot, making it sensitive to biological load introduced by ballast discharge.
- Landsat 9 OLI-2: 30 m multispectral resolution; 16-day revisit solo, 8-day combined with Landsat 8. Coastal/aerosol band (Band 1, 430-450 nm) and Band 3 (green) are calibrated for water-leaving reflectance and complement Sentinel-2 in time, particularly useful for building historical baselines of ambient turbidity at a given port approach.
- Sentinel-1 SAR (C-band): 6-day revisit in Interferometric Wide Swath mode at 10 m resolution. SAR is insensitive to cloud and works at night. Coherence change detection between repeat passes can flag surface roughness anomalies consistent with discharge events, and backscatter contrasts can reveal biogenic surfactant films associated with biological ballast content, though attribution requires optical confirmation.
What ballast water actually looks like from orbit
A vessel arriving from a distant ocean carries tens of thousands of tonnes of ballast water drawn from its port of origin. When it discharges before entering a new port, it releases a water mass with a different suspended sediment load, salinity, and biological community than the receiving coastal water. That difference is optically measurable.
Water-leaving radiance in the green and red bands (roughly 550-700 nm) rises sharply when suspended particulate matter increases. Sentinel-2 Band 3 (green, 10 m) and Band 4 (red, 10 m) are the primary detection channels. A discharge plume from a large bulk carrier can span several hundred metres within an hour of release, well above Sentinel-2's resolution floor. The plume geometry, an elongated fan trailing from the vessel's stern, is itself a discriminating feature that separates ballast discharge from, say, riverine turbidity or sediment resuspension from dredging.
Spectral physics: why the signal survives atmospheric noise
Ocean colour remote sensing is complicated by the atmosphere, which contributes roughly 90 percent of the top-of-atmosphere radiance over water. Accurate retrieval of water-leaving reflectance depends on atmospheric correction algorithms, and errors in those corrections can swamp a weak biological signal. This is the method's most significant limitation, particularly in hazy or humid coastal environments.
Sentinel-2's B1 band (60 m, 443 nm) supports the ACOLITE and C2RCC atmospheric correction processors, both of which are publicly documented and validated for coastal turbid water. PACE OCI's continuous hyperspectral coverage gives it a structural advantage: it can resolve the spectral shape of the absorption and backscattering coefficients independently, distinguishing mineral sediment (which has a relatively flat backscatter spectrum) from phytoplankton-rich water (which shows the characteristic chlorophyll-a absorption trough near 440 nm and fluorescence peak near 685 nm). That distinction matters because ballast water from biologically productive source regions carries a measurable phytoplankton signal absent from the ambient coastal water at the discharge location.
Attributing the plume to a ship: where AIS enters
Detecting a turbidity anomaly is necessary but not sufficient for enforcement. Attribution requires placing a vessel at the plume's origin at the right time. AIS broadcast positions, available at one-to-few-minute intervals from satellite AIS receivers such as those on Spire's constellation, provide the vessel track. The geometry is simple: the plume's upwelling end, identified by its highest turbidity pixel cluster, should correspond to the vessel's position at the time of discharge onset, corrected for current drift using published tidal or ocean current models.
The method has an honest weakness. A vessel that turns off its AIS transponder before discharging, or that spoofs its position, breaks the automatic attribution chain. SAR can detect a vessel's physical presence independently of AIS, providing a position fix that does not rely on cooperative reporting. Fusing a SAR detection with a Sentinel-2 or Landsat turbidity plume observed within the same tidal window gives a two-sensor attribution that is considerably harder to contest. Night-time discharge, invisible to optical sensors, is the remaining blind spot; SAR coherence change is the only satellite tool with any sensitivity there, and its evidence is circumstantial without a co-located optical confirmation.
Revisit reality and the coverage gap
Sentinel-2's five-day revisit is the operative constraint. A discharge event lasting two to four hours may leave a plume that persists for six to twelve hours depending on tidal flushing, but there is no guarantee a satellite pass occurs within that window. At high latitudes, overlapping orbital swaths improve revisit to two or three days. Tasking commercial optical constellations at sub-metre resolution can close the gap for high-priority port approaches, but at a cost and with a tasking lead time that must be planned in advance.
Cloud cover compounds the problem in tropical port approaches during monsoon seasons. Sentinel-1 SAR is the fallback, but as noted above its evidence for biological ballast discharge specifically is indirect. A realistic operational posture combines persistent Sentinel-2 and Landsat archival monitoring to establish ambient turbidity baselines, triggered commercial tasking when a vessel of interest is expected to arrive, and SAR as a night-and-cloud backstop.
Building a port-approach baseline: the analytic workflow
Effective detection requires knowing what the water normally looks like. A multi-year Sentinel-2 and Landsat time series for a port approach, processed through a consistent atmospheric correction chain, yields a seasonal turbidity climatology. Anomalies are then defined relative to that baseline, not against a fixed threshold, which reduces false positives from natural sediment pulses after storms or river floods.
The detection pipeline has four steps: atmospheric correction and retrieval of remote sensing reflectance (Rrs); calculation of a turbidity proxy such as the Normalised Difference Turbidity Index or a semi-analytical inversion for total suspended matter; spatial clustering to identify plume-shaped anomalies; and temporal matching against AIS vessel tracks within a defined radius and time window. PACE OCI adds a fifth step for biologically loaded plumes: spectral unmixing to estimate phytoplankton absorption coefficients, which can indicate whether the biological community in the plume is consistent with the vessel's reported port of origin. That last step is research-grade at present, not operational, but the published literature from the NASA PACE mission science team documents the underlying capability.
Satellize runs the first four steps as a repeating analytics layer on open constellations for clients with port-approach monitoring requirements, drawing on the same open-data infrastructure used in the Tonga crop-estimation programme. The output is a flagged-event log with plume polygons, estimated turbidity anomaly magnitude, and candidate vessel attribution, delivered as a GIS layer and a structured incident report.
What the method cannot prove
Satellite evidence establishes that an anomalous water mass appeared near a vessel at a given time. It does not directly measure invasive species concentration, pathogen load, or whether the discharge was intentional. Those questions require in-situ water sampling and biological analysis. The satellite layer's role is to direct enforcement resources: it narrows the field from all vessels transiting a port approach to a short list of candidates worth boarding and sampling.
International Maritime Organisation conventions, specifically the Ballast Water Management Convention in force since 2017, require vessels to manage ballast water to defined biological standards. Satellite detection supports flag-state and port-state enforcement by providing independent, timestamped evidence of discharge events outside permitted zones, but the chain of custody from satellite image to legal proceeding requires documented, auditable processing steps. That is a data-governance question as much as a technical one.
Typical figures
| Best spatial resolution (optical) | 10 m (Sentinel-2 MSI visible bands) |
| Best spatial resolution (SAR) | 10 m (Sentinel-1 IW mode) |
| Hyperspectral resolution (PACE OCI) | ~1 km spatial, ~5 nm spectral, 340-890 nm |
| Revisit (Sentinel-2 A+B combined) | 5 days equatorial; 2-3 days mid-to-high latitudes |
| Revisit (Landsat 8+9 combined) | 8 days |
| Revisit (PACE OCI) | Daily global |
| Minimum detectable plume area (Sentinel-2) | Approximately 1-2 ha at 10 m resolution; plumes from large vessels typically exceed this within 30-60 minutes of discharge |
| Turbidity detection limit | Published studies report Sentinel-2 sensitivity to total suspended matter changes of roughly 1-2 g/m³ above background in clear coastal water, degrading in already-turbid ports |
| Archive depth | Sentinel-2 from 2015; Landsat from 1972; PACE OCI from early 2024 |
| Cloud limitation | Optical sensors fully blocked by cloud; SAR unaffected but provides indirect biological signal only |
Analytics Satellize can run
| Port-approach turbidity baseline | Multi-year Sentinel-2 and Landsat time-series, seasonal climatology of remote sensing reflectance and total suspended matter retrieval via ACOLITE or C2RCC atmospheric correction | GIS raster layer (monthly composites) and summary statistics per port polygon, updated quarterly |
| Discharge event detection alert | Anomaly detection against turbidity baseline; spatial clustering of high-Rrs pixels into plume-shaped objects; threshold set at 2 standard deviations above seasonal mean | Flagged-event GeoJSON with plume polygon, anomaly magnitude, image timestamp and sensor ID; delivered within 24 hours of satellite overpass |
| Vessel attribution report | Temporal and spatial matching of plume origin pixel cluster against satellite AIS track data (Spire or equivalent); drift correction using published tidal current model | Structured incident report naming candidate vessel(s), MMSI, flag state, plume-to-vessel distance and time offset, with confidence tier (high/medium/low) |
| SAR night-discharge flag | Sentinel-1 coherence change detection and backscatter anomaly analysis over port approach between consecutive passes; cross-referenced with AIS gap events in the same window | Alert log entry with SAR image chip, coherence difference map, and co-located AIS status (transmitting / dark) for vessels present |
| Biological load indicator (PACE OCI) | Semi-analytical spectral inversion of OCI hyperspectral Rrs to retrieve phytoplankton absorption coefficient (aph) and particulate backscattering; comparison against open-ocean source-water signature from vessel's declared last port | Research-grade supplementary layer flagging plumes with elevated aph inconsistent with local ambient; explicitly labelled as indicative, not legally conclusive |
| Recurring offender pattern analysis | Longitudinal matching of attributed discharge events against vessel identity, route and operator across a rolling 12-month archive | Ranked vessel watchlist with event frequency, port locations and flag-state summary, formatted for port-state control briefing |
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.