Challenges to identify phytoplankton species in coastal waters by remote sensing

摘要

During spring and summer 2004, intensive field campaigns were conducted in the Eastern English Channel. This region is characterized by relatively intense phytoplankton blooms, low bathymetry, strong tide ranges and great river inputs. The sampling period accounts for episodic blooms of prymnesiophyceae Phaeocystis globosa and diatoms. Hyperspectral radiometric measurements (TRIOS; 350-950 nm, with a 3 nm spectral resolution) were concurrently performed with water sampling for biogeochemical and optical characterization. The remote sensing reflectance, R_(rs), is analyzed in conjunction with variation of the water composition. We particularly focus on the capability to identify some phytoplankton species from R_(rs) in this very variable environment. Different methods, based on multispectral and hyperspectral data are tested and compared for that purpose. We show that no R_(rs) ratio allows to discriminate between diatoms and Phaeocystis. In contrast, the derivative analysis applied to hyperspectral data stresses large differences in some part of the R_(rs) spectra collected in diatoms or Phaeocystis dominated waters.