Modeling Phytoplankton Pigments on Global to Local Scales Using Hyperspectral Optics

Sasha Jane Kramer, University of California Santa Barbara, Santa Barbara, United States and David Siegel, Univ of California Santa Barbara, Santa Barbara, United States
Satellite ocean color provides unprecedented spatiotemporal coverage of the global surface ocean; these measurements can be leveraged to characterize phytoplankton community structure by relating variables measured by satellites (i.e., remote sensing reflectance, Rrs) to phytoplankton taxonomy measured in situ. The potential of these ocean color methods is only improved at hyperspectral wavelength resolution (e.g., NASA’s upcoming Plankton, Aerosol, Cloud, and ocean Ecosystem sensor, PACE). Phytoplankton community structure can be defined at low taxonomic resolution from biomarker pigments measured using high-performance liquid chromatography (HPLC). The composition of phytoplankton pigments has an impact on the spectral shapes of Rrs and phytoplankton absorption (aph), allowing relationships to be constructed between biomarker pigments and the derivative spectra of these optical measurements. Here, global and local scale datasets of HPLC pigments and hyperspectral Rrs and aph are combined to target features in the optical data that correspond to groups of phytoplankton quantified by biomarker pigment concentrations. The derivative spectra approach takes advantage of a spectral scale separation between aph and absorption by other oceanic constituents (i.e., CDOM, non-algal particles). The hyperspectral resolution of the Rrs and aph datasets allows for fine-scale absorption features due to phytoplankton pigments to be parsed from larger-scale features. The approach used here also considers the potential and limitations of HPLC pigments for assessing phytoplankton community structure. On global scales, a maximum of 4 taxonomic groups can be identified from HPLC pigments (diatoms and dinoflagellates, haptophytes, green algae, and cyanobacteria); on local scales, often more and different groups are found (i.e., dinoflagellates can be identified locally but not globally). Thus, this model development considers both the spatial scale and the groups present to target phytoplankton biomarker pigments from optics.