2026/04/25 by María Laura Zoffoli, Maria Laura Zoffoli, Victor Pochic +11
Earth and Planetary Sciences · Environmental Science · #Marine and coastal ecosystems #Aquatic Ecosystems and Phytoplankton Dynamics #Oil Spill Detection and Mitigation
paper · doi:10.1016/j.rse.2026.115439
Harmful algal blooms (HABs) in coastal waters are a major concern worldwide and are expected to increase in frequency over the coming decades. Traditional monitoring programmes rely mainly on in situ sampling, while satellite imagery offers broader coverage of chlorophyll- a (Chl- a ) concentration, supporting HAB mapping and monitoring. However, most satellite missions are multispectral and their limited number of spectral bands limits the identification of bloom-dominant phytoplankton classes. Hyperspectral satellite missions, with their detailed reflectance signals, hold promise for detecting diagnostic pigments and improving HAB monitoring. In this study, we developed pigment-specific line height (LH) algorithms using in situ and satellite derived hyperspectral remote-sensing reflectance ( R rs ) data from the PRISMA and EMIT satellite missions. These data were collected from dense phytoplankton bloom events, i.e., water was dominated by one or a few species, with a biomass equivalent to a “massive open-air culture”. Dominant taxa included HAB-forming species of dinoflagellates such as Lepidodinium chlorophorum (containing Chl- b ), Lingulaulax polyedra (Chl- c ), Alexandrium spp. (Chl- c ), raphidophytes such as Heterosigma akashiwo (Chl- c ), and cyanobacteria (phycocyanin, PC). We calculated three types of LH centred at ∼620 nm (LH PC ), ∼628 nm (LH Cc-PC ), and ∼ 646 nm (LH Cb ), to respectively detect PC, Chl- c and Chl- b in highly concentrated blooms. Our analysis revealed that LH Cb ≥ 0 indicates the presence of Chl- b at high concentration, LH Cc-PC ≥ 0 indicates either the presence of Chl- c or PC, and in combination with LH PC allows discrimination between Chl- c and PC, where LH PC > 0.065 denotes the presence of PC. LH algorithms were applied to PRISMA, EMIT and PACE imagery for pigment detection, achieving a global accuracy of 0.76 when validated with independent datasets over phytoplankton blooms of species containing Chl- b , Chl- c or PC. Upon detecting Chl- b or Chl- c , we adapted a semi-analytical algorithm (initially developed to retrieve Chl- a ) to quantify their concentrations. Our results showed that the satellite-retrieved pigment ratios Chl- b :Chl- a or Chl- c :Chl- a agreed with in situ and culture-based pigment measurements. Finally, we demonstrated the usefulness of hyperspectral LH algorithms for monitoring very intense phytoplankton blooms and ecological applications. This study makes two key contributions. Firstly, it enhances the monitoring of highly concentrated bloom events and the identification of accessory pigments, and secondly, it demonstrates the potential of hyperspectral data for this application. It underscores the value of integrating additional spectral bands, particularly in the red region, for more precise detection of key pigments, ultimately advancing species-specific phytoplankton bloom tracking. • Hyperspectral satellite data detect accessory pigments in dense blooms. • Line height algorithms were proposed to identify chlorophyll- b , −c and phycocyanin. • Pigment detection algorithms showed a global accuracy of 0.76 over independent data. • Chl- b and Chl- c were quantified using an adapted semi-analytical algorithm. • Hyperspectral imagery maps HAB pigment types and monitors bloom spatial dynamics.