Satellites Track Toxic Algal Blooms Well, But Counting Their Cells Is Far Harder

Every late summer, the tributaries of lower Chesapeake Bay turn into a battleground. Toxic dinoflagellates such as Margalefidinium polykrikoides and Alexandrium monilatum bloom in the York and James Rivers, threatening fish, shellfish, and a thriving aquaculture industry. Now, a team of researchers led by R. Wayne Litaker of the University of North Carolina at Chapel Hill and Michelle C. Tomlinson of NOAA’s National Centers for Coastal Ocean Science has put satellite monitoring of these blooms to a rigorous test, with results that are both reassuring and sobering. Writing in Frontiers in Remote Sensing, they show that daily satellite imagery can measure bloom biomass accurately, but converting those measurements into actual counts of harmful cells remains a stubbornly elusive goal.

The stakes are high. During fish kill events, cell densities of these dinoflagellates most commonly reach one million cells per liter or higher, though mortalities have sometimes occurred at concentrations as low as roughly 500,000 cells per liter, depending on the species and conditions. Lethal concentrations for shellfish and other invertebrates range between one and five million cells per liter. Alexandrium monilatum produces goniodomin A, a polyether macrolide toxin that interferes with actin function, limits angiogenesis, and kills isolated liver and nerve cells, along with a suite of unidentified lytic compounds released mainly when cells rupture during grazing or at the end of a bloom. Margalefidinium polykrikoides takes a different route, generating reactive oxygen species that destroy gill membranes and secreting extracellular polysaccharides that can clog gills and asphyxiate fish. Both species form patchy blooms that shift location rapidly with wind, tide, and river flow, making them notoriously difficult to track with boat-based sampling alone.

That difficulty is precisely why the researchers turned their attention to the Ocean and Land Colour Instrument, or OLCI, carried aboard the European Sentinel-3 satellites. OLCI offers a compelling combination of capabilities: it images the same location daily, resolves features at 300 meters, and includes additional spectral bands in the red and near-infrared portions of the spectrum that are less sensitive to the colored dissolved organic matter and sediment that plague estuarine waters. NOAA has already used an OLCI-derived chlorophyll a product operationally to monitor blooms across Chesapeake Bay for over a decade, but the algorithm had only been validated in the Bay’s mainstem. The turbid, CDOM-rich tributaries where these dinoflagellates bloom most intensely had never been tested.

The validation strategy confronted a fundamental mismatch head-on. A field sample represents a few liters of water collected at one point in space and time, whereas a single OLCI pixel covers roughly 90,000 square meters, and the satellite flies over hours away from when a sample bottle was filled. Patchy blooms that migrate vertically and horizontally only sharpen the problem. The team matched field chlorophyll a measurements from Chesapeake Bay Program stations in the York and James Rivers, collected between 2016 and 2025, with median chlorophyll values from 3 by 3 pixel grids of OLCI imagery processed with NOAA’s red-edge algorithm, which switches to the standard global OC4 algorithm at chlorophyll concentrations below 10 micrograms per liter.

The results validated the satellite approach with one caveat. After removing an observed bias of about 5 micrograms per liter in the tributary waters, the satellite-derived chlorophyll estimates showed a median absolute deviation of just 2.58 micrograms per liter with essentially no residual bias. The satellite time series remained stable across the entire nine-year record, capturing realistic seasonal variability and reliably flagging bloom peaks. In practical terms, managers can trust the satellite maps to tell them where biomass is high, how blooms are moving, and where to send sampling boats, extending a monitoring capability previously proven only in the Bay’s mainstem into the challenging tributary environment.

The second objective was more ambitious and produced the study’s most striking finding. Chlorophyll a is a proxy for biomass, but the true measure of risk is cell concentration of the toxic species itself. Converting satellite chlorophyll to cell counts requires knowing how many picograms of chlorophyll each cell contains, and the researchers found that this conversion factor is wildly unstable. Drawing on a decade of matched field cell counts and chlorophyll measurements, plus laboratory culture experiments and published studies from other ecosystems, they showed that chlorophyll per cell at any given chlorophyll concentration can vary by five-fold to more than 100-fold.

The laboratory experiments revealed why. When the team grew both dinoflagellates under combinations of salinity, temperature, light, and nutrient conditions typical of blooms, two factors dominated the variability. Nutrient limitation was the most powerful driver: as batch cultures exhausted their nitrogen supply over two to three weeks, chlorophyll per cell in Alexandrium monilatum declined by 100-fold or more, and by five- to ten-fold in Margalefidinium polykrikoides. Salinity was nearly as influential, with cells grown at a salinity of 20 consistently containing far more chlorophyll than cells grown at 33, often producing differences of 100-fold or more in total chlorophyll at equivalent cell densities. Light and temperature played smaller roles, with lower light generally raising chlorophyll content per cell.

Field data added a second, ecological source of uncertainty. Blooms are almost never pure cultures, and dividing bulk chlorophyll by the cell count of the target species erroneously assigns chlorophyll from co-occurring phytoplankton to the harmful species. The evidence that this matters was written into the statistics: in years when harmful algal bloom cell counts correlated strongly with total chlorophyll, suggesting the bloom was relatively monospecific, the conversion factor was more stable. In years with weak correlations, variability ballooned. Across bloom years for Margalefidinium polykrikoides, for example, the coefficient of determination between cell concentration and chlorophyll ranged from near zero to 0.74, and the chlorophyll per cell conversion factor varied from about 0.1 to 1,147 picograms in field samples.

The implications ripple far beyond Chesapeake Bay. Both dinoflagellate species form blooms in estuaries and nearshore waters worldwide, and the authors argue their findings likely apply to other harmful bloom species as well. Simply borrowing a chlorophyll per cell value from a published study in another region or season will almost certainly produce inaccurate cell estimates. What the satellite product can still deliver is a relative measure: where biomass is highest, how it is changing, and when blooms cross thresholds that warrant field investigation, which has proven genuinely useful to state environmental agencies over the past decade.

There is, however, a path toward better cell estimates. The researchers propose a three-step approach modeled on methods used for estimating microcystin in cyanobacterial blooms: use OLCI chlorophyll to map the bloom, periodically collect field samples during the bloom to measure chlorophyll per cell under current growth conditions, and apply those bloom-specific conversion factors in near real time to translate the next satellite image into cell concentrations. As conditions shift through a bloom, the conversion factors would need updating every few weeks, but the payoff would be satellite-based cell estimates grounded in the actual physiological state of the bloom rather than a static, and potentially wrong, literature value. For now, the study delivers a clear-eyed verdict: satellites have earned their place as the eyes in the sky for tracking toxic blooms, but counting the killer cells still requires boots in the water.

Subject of Research: Validation of satellite-derived chlorophyll a and harmful algal bloom cell concentration estimates using OLCI imagery in Chesapeake Bay tributaries

Article Title: Quantifying the accuracy of estimating harmful algal bloom cell concentrations using OLCI imagery in the tributaries of lower Chesapeake Bay

Article References: Litaker, R. W., Tomlinson, M. C., Holland, W. C., Reece, K. S., Egerton, T. A., Briggs, T. O., Scott, G. P., Small, H. J., Gaillard, S., & Mapes, S. A. (2026). Quantifying the accuracy of estimating harmful algal bloom cell concentrations using OLCI imagery in the tributaries of lower Chesapeake Bay. Frontiers in Remote Sensing, 7, Article 1873211. https://doi.org/10.3389/frsen.2026.1873211

Image Credits: AI Generated

DOI: 10.3389/frsen.2026.1873211

Keywords: harmful algal blooms, OLCI, chlorophyll a, Chesapeake Bay, Alexandrium monilatum, Margalefidinium polykrikoides, remote sensing, dinoflagellates, Sentinel-3, cell counts, NOAA, estuarine monitoring