Harmful algal bloom detection and extent mapping

Toxin-producing phytoplankton blooms threaten fisheries, drinking water and public health, yet cloud cover and coarse revisit rates make reliable single-pass detection difficult. Multi-sensor compositing and Sentinel-3 OLCI's 300 m ocean-colour products now give coastal managers a practical detection workflow.

Sensors

Sentinel-3 OLCI: 21 spectral bands from 400 to 1020 nm at 300 m resolution; ~2-day revisit at mid-latitudes with twin satellites (Sentinel-3A and 3B). The primary workhorse for coastal HAB mapping: its fluorescence line height (FLH) and chlorophyll-a products are operationally processed by EUMETSAT and freely available within hours of overpass.
MODIS-Aqua: 250 m to 1 km bands; daily global coverage. Chlorophyll-a and FLH products have a heritage archive back to 2002, making it the standard for long-term bloom climatology. The 1 km ocean-colour bands limit utility in narrow estuaries and fjords.
NASA PACE OCI: Launched February 2024. Hyperspectral ocean-colour imager covering 340–890 nm at ~5 nm spectral resolution and approximately 1 km spatial resolution, daily global revisit. Its fine spectral sampling improves discrimination between HAB species and benign phytoplankton communities, a capability MODIS and OLCI can only approximate.
Landsat-8/9 OLI: 30 m multispectral resolution in coastal/aerosol, blue, green and red bands. Useful for resolving bloom boundaries in small bays and reservoirs where 300 m pixels average over too much water. Revisit is 8 days per satellite (16-day for a single satellite), which frequently misses fast-evolving bloom events.

What the water is actually saying

Phytoplankton scatter and absorb light in ways that vary by pigment composition, cell density and depth. Chlorophyll-a absorbs strongly at around 443 nm and 675 nm, producing a reflectance peak near 550 nm that gives productive water its green cast. Harmful species add their own spectral fingerprints on top of this baseline.

Karenia brevis, the Florida red-tide dinoflagellate, produces a characteristic brown-red discolouration driven by its high carotenoid content. Cyanobacteria such as Microcystis and Nodularia carry phycocyanin, a pigment with an absorption feature near 620 nm that chlorophyll-a does not share. Alexandrium species are harder: they do not produce a distinctive accessory pigment, so remote sensing can flag the bloom but cannot confirm the genus without in-water sampling. This is an honest limit of the method, and any workflow that claims otherwise is overstating the science.

Fluorescence line height: the signal that cuts through ambiguity

All photosynthetically active phytoplankton re-emit a small fraction of absorbed red light as fluorescence, centred near 685 nm. Fluorescence line height (FLH) measures the magnitude of this emission above a baseline drawn between adjacent non-fluorescent bands. MODIS-Aqua and Sentinel-3 OLCI both carry dedicated band sets for FLH retrieval.

FLH is valuable because it responds to active chlorophyll rather than to coloured dissolved organic matter or suspended sediment, both of which confound simple band-ratio chlorophyll-a estimates in turbid coastal water. In practice, FLH still saturates at very high cell concentrations (the signal quenches when cells shade each other), so surface scum conditions can produce lower FLH than a moderately dense sub-surface layer. Interpreting FLH alongside remote-sensing reflectance ratios and, where available, in-situ fluorometry gives a more reliable picture than any single product alone.

Cloud cover is the dominant operational problem

Coastal regions prone to HABs, the Gulf of Mexico, the Baltic, the Yellow Sea, the Norwegian fjords, are also frequently cloudy. A single overpass under cloud produces nothing usable. During a fast-developing K. brevis event in the Gulf, a bloom can travel tens of kilometres in 48 hours on wind-driven surface currents. By the time a clear overpass arrives, the geometry has changed.

The standard mitigation is multi-day compositing: selecting the clearest pixel from each location across a rolling 3-to-8-day window. This extends spatial coverage at the cost of temporal precision. For slow-moving cyanobacterial blooms in enclosed basins, a 5-day composite is often acceptable. For fast-moving marine dinoflagellate events, it can be misleading. Combining optical data with wind-field reanalysis and ocean-circulation model output helps distinguish where a bloom was from where it is likely to be now.

Sentinel-3's twin-satellite configuration (3A and 3B) reduces the revisit gap at mid-latitudes to roughly two days, a meaningful improvement over the four-day single-satellite cycle. At polar latitudes, the orbits converge and revisit improves further, which is relevant for monitoring cyanobacterial blooms in Arctic coastal waters as sea-ice retreat extends the open-water season.

Separating harmful from merely productive

High chlorophyll-a is not, by itself, evidence of a HAB. Upwelling zones, river plumes and post-storm nutrient pulses all produce elevated chlorophyll without toxin risk. The spectral features that help distinguish harmful from benign blooms are pigment ratios (phycocyanin-to-chlorophyll for cyanobacteria), anomalous FLH-to-chlorophyll ratios, and surface texture visible in high-resolution imagery.

NASA's PACE OCI changes the calculus here. Its hyperspectral coverage allows phytoplankton community composition to be inferred using spectral unmixing methods that MODIS and OLCI's discrete bands cannot support. Early validation studies suggest PACE can distinguish broad functional types (diatoms, dinoflagellates, cyanobacteria) with greater confidence than heritage sensors. The caveat is that PACE data products are still maturing; operational HAB agencies are validating them against in-situ pigment measurements collected by research cruises and Argo floats.

Landsat-8/9 OLI contributes a different capability: spatial detail. At 30 m, it can resolve the streaks and Langmuir circulation patterns that concentrate surface cells, giving bloom-boundary maps that are genuinely useful for beach closure decisions and aquaculture cage positioning. Its spectral range does not include a dedicated fluorescence band, so it is most useful as a complement to OLCI or MODIS rather than a replacement.

Building an operational monitoring workflow

A practical HAB monitoring system for a coastal government agency typically combines three layers. First, near-real-time OLCI or MODIS chlorophyll-a and FLH alerts, flagging pixels that exceed a locally calibrated anomaly threshold. Second, a rolling composite product that fills cloud gaps and tracks bloom extent over days to weeks. Third, a contextual layer: sea-surface temperature (available from the same Sentinel-3 SLSTR instrument), wind fields, and river discharge data that help explain why a bloom is where it is.

Satellize runs this kind of multi-source analytics stack on open constellations, adding commercial tasking on client licence where sub-daily revisit or very high resolution is needed. The architecture is similar in principle to the crop-estimation work done for the Kingdom of Tonga, where compositing and anomaly detection on open satellite data formed the analytical core.

Alert latency matters. EUMETSAT distributes Sentinel-3 OLCI Level-2 water products within three hours of overpass under its Near-Real-Time processing stream. MODIS-Aqua products are available from NASA Earthdata within similar timescales. A monitoring system that ingests these feeds automatically and applies pre-defined thresholds can issue a coastal alert the same morning as the satellite pass, provided the sky was clear.

Honest limits and what they mean for decisions

Remote sensing detects surface optical signals. A bloom that has sunk below the first optical depth (roughly 1–2 m in turbid coastal water, up to 20 m in clear open ocean) is invisible to any passive sensor. Alexandrium, which can migrate vertically through the water column, can be present at toxic concentrations while the surface appears clean. No satellite product replaces in-water sampling for shellfish toxin monitoring; it redirects the sampling effort to where it is most likely to be needed.

Atmospheric correction over coastal water remains an active research problem. Aerosols, sun glint and adjacency effects from land can all introduce errors in retrieved water-leaving reflectance, particularly in the blue bands most sensitive to phytoplankton pigments. OLCI's dedicated 400 nm band and its improved aerosol correction scheme perform better than MODIS in many coastal geometries, but errors of 20–30% in chlorophyll-a retrievals near turbid river plumes are not unusual. Treat satellite-derived concentrations as indices for anomaly detection, not as absolute values for regulatory thresholds.

Typical figures

Best coastal spatial resolution	300 m (Sentinel-3 OLCI); 30 m with Landsat-8/9 OLI (no fluorescence band)
Typical revisit (twin Sentinel-3)	~2 days at mid-latitudes; daily MODIS-Aqua global coverage
NRT product latency	~3 hours post-overpass for EUMETSAT OLCI Level-2; similar for NASA MODIS
Key spectral bands	Chlorophyll absorption: ~443 nm, ~675 nm; FLH emission: ~685 nm; phycocyanin: ~620 nm
PACE OCI spectral resolution	~5 nm from 340–890 nm; enables phytoplankton community discrimination
Minimum detectable chlorophyll-a	Approximately 0.05–0.1 mg/m³ in clear open water; higher detection floor in turbid coastal water
Archive depth	MODIS-Aqua: 2002–present; Sentinel-3A: 2016–present; Landsat-8: 2013–present
Cloud limitation	Single-pass optical data lost under cloud; 3–8 day compositing standard mitigation
Delivery formats	GeoTIFF, NetCDF-4, OGC WMS/WFS; alert feeds via email or API

Analytics Satellize can run

Chlorophyll-a anomaly alert	Threshold detection on OLCI or MODIS Level-2 chlorophyll-a relative to a locally derived seasonal climatology	Same-day email or API alert with bloom centroid coordinates and estimated area in km²
Fluorescence line height map	FLH retrieval from Sentinel-3 OLCI bands 7, 8 and 9 (665, 674, 681 nm) using the standard NASA/EUMETSAT baseline algorithm	GeoTIFF layer at 300 m resolution, updated each clear overpass
Multi-day bloom composite	Maximum valid-pixel compositing over a rolling 5-day window, cloud-masked using OLCI quality flags	Weekly GeoTIFF and PDF summary showing bloom extent evolution
Phycocyanin index map (cyanobacteria)	Band-ratio index using reflectance near 620 nm and 665 nm to isolate phycocyanin signal; validated against published OLCI cyanobacteria indices	GIS polygon layer of probable cyanobacterial surface accumulations with confidence classification
Bloom trajectory forecast overlay	Satellite-detected bloom centroid combined with ERA5 or Copernicus Marine Service surface-current fields to project 24–48 hour drift	Map overlay in client GIS showing probable bloom position range with uncertainty envelope
Seasonal HAB risk climatology	Statistical analysis of MODIS-Aqua archive (2002–present) to derive bloom frequency, typical onset timing and peak extent by location	Annual risk report with monthly probability maps; suitable for aquaculture site planning or coastal management review

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.