The many shades of red tides: Sentinel-2 optical types of highly-concentrated harmful algal blooms

Harmful algal blooms (HABs) have severe environmental and economic impacts worldwide. Improving HAB detection is crucial because massive blooms are likely to increase in both frequency and amplitude in the next decades due to global warming and escalating coastal eutrophication. While satellite remote sensing has proved useful to detect red tides and support HAB monitoring, the discrimination of the dominant bloom-forming species is still a challenge, all the more as the observation of highly concentrated phytoplankton patches can be hampered by a too coarse spatial resolution. Moreover, the majority of HAB studies generally had a regional focus, and a limited number of species were separately documented so far. Here, we provide a broader perspective for red tides remote sensing to better resolve HAB optical and taxonomical diversity. The main objective of the present study was to identify how many optical bloom types could be recognized with the high spatial resolution Sentinel-2 (S2) satellite mission. For that purpose, an extensive database of massive, nearly monospecific blooms, both documented in situ and using synchronous S2 observation was compiled. More than 100 S2 images of various red tides were selected worldwide. Altogether, the S2 database covered the typical reflectance spectra of 27 red tide forming species. The remote-sensing reflectance of each red tide was analysed to evaluate S2 ability to distinguish the dominant species of the bloom. A hierarchical clustering analysis suggested that six optical bloom types could be identified: (1) surface accumulation of cyanobacteria or of green Noctiluca scintillans, (2) surface accumulation of red N. scintillans (a purely heterotrophic plankton devoid of chlorophyll a), (3) red tides of Mesodinium rubrum (a phycoerythrin-bearing ciliate), (4) green seawater discolorations of Lepidodinium chlorophorum (a dinoflagellate with unusual carotenoids), (5) blooms dominated by a dinoflagellate such as Prorocentrum, Gymnodinium, Lingulodinium polyedra, Gonyaulax or Alexandrium, and (6) brown tides dominated by a dinoflagellate (such as Karenia, Karlodinium veneficum, Protoceratium reticulatum, Margalefidinium polykrikoides, or Tripos fusus), a prymnesiophyte (Phaeocystis), or a pelagophyte (Aureococcus anophagefferens). While the results presented here are inherently limited by the concomitant availability of in situ and S2 observations, as well as by S2 spectral resolution, it is a step forward to an improved understanding of HAB bio-optical diversity.