Hyperspectral optical discrimination of phytoplankton community structure in Funka Bay and its implications for ocean color remote sensing of diatoms

Abstract Identification of phytoplankton functional groups is key to understanding marine biogeochemical cycles. For more accurate understanding of phytoplankton community structure and its implications for ocean color remote sensing applications, we investigated seasonal changes in phytoplankton pigments with high-performance liquid chromatography (HPLC), hyperspectral absorption coefficients of detritus ( a d (λ)), phytoplankton ( a ph (λ)), and colored dissolved organic matter ( a CDOM (λ)), and hyperspectral a ph (λ) derived from remote sensing reflectance ( a ph_ R rs-derived (λ)) in the coastal waters of Funka Bay from 2010 to 2012. Chlorophyll a (Chl a ) concentrations measured by HPLC ranged from 0.29 to 8.6 mg m − 3 . Phytoplankton community compositions, as estimated by chemotaxonomic analysis (CHEMTAX) based on HPLC phytoplankton pigments, showed a seasonal succession of diatoms, chlorophyll b -containing phytoplankton (chlorophytes and prasinophytes), and cyanobacteria. Additionally, to identify the dominant type of phytoplankton with an alternative technique to CHEMTAX analysis, we employed a derivative spectroscopy/similarity index (SI) approach for a ph (λ) as an optical detection technique for discriminating between different types of phytoplankton. In particular for diatom-dominated stations, SI values relative to the second derivative spectra of a ph (λ) of diatom cultures, isolated from our study region, were significantly higher than those for chlorophyll b -containing phytoplankton- and cyanobacteria-dominated stations. Furthermore, we found a strong relationship between the SI values calculated from the second derivative spectra and the composition of diatoms as estimated by CHEMTAX. These results suggest that the two different methods validated each other's performance and precision in estimating relative diatom abundance from bulk samples and that it is possible to optically discriminate the dominance of diatoms using derivative spectra of a ph (λ). We extended this combination approach to hyperspectral a ph_ R rs-derived (λ), using a quasi-analytical algorithm within 400–546 nm range. We found a significant correlation between SI values obtained from the second derivative spectra of a ph_ R rs-derived (λ)/ a ph_ R rs-derived (443) and the composition of diatoms derived by CHEMTAX, but it was not as high as for a ph (λ) measured by filter-pad analysis. These results indicate that using hyperspectral optical data of a ph (λ) and R rs (λ) with derivative spectroscopy is potentially a promising approach to identify seasonal variability in the composition of diatoms in coastal waters. Furthermore, a hyperspectral approach in combination with CHEMTAX analysis as a reference for phytoplankton community structure has proven useful in improving our understanding of phytoplankton community structure in the coastal waters of Funka Bay.