POSTER SESSION 2
10:30–12:30, Tuesday, September 15
Poster Session | 1 | 2 | 3 | 4 | Schedule at a Glance
ABSTRACT 1095 | POSTER 70
THE IMPRINT OF PHYTOPLANKTON COMMUNITY STRUCTURE ON HYPERSPECTRAL RRS SIGNALS: COMPARATIVE DIAGNOSTICS FROM IN SITU OBSERVATIONS
Phytoplankton community structure leaves a measurable imprint on hyperspectral remote sensing reflectance. Still, this imprint is distributed across multiple biochemical, taxonomic, and spectral expressions that are not equally visible in raw reflectance alone. In this study, we compared four complementary hyperspectral representations based on a compiled global dataset (N=237): raw Rrs, a smooth-background residual, a semi-analytical spectral mismatch, and a second derivative. Relative HPLC pigment composition was used to define phytoplankton assemblage structure, and a range of classes was identified using self-organizing maps. The spectral representations were evaluated through pigment-ratio class prediction accuracy, wavelength-wise class discrimination, sensitivity, and uncertainty diagnostics. In parallel, pigment ratios and hyperspectral representations were compared with independent phytoplankton-group information derived from psbO metagenomic reads (N=93).
The results show that phytoplankton community structure is expressed through an organized set of relationships. Derived spectral spaces reveal this organization more clearly than raw Rrs alone, systematically achieving higher accuracy in pigment-ratio class discrimination, with each representation emphasizing different components of the ecological signal. Major psbO-derived phytoplankton groups, including diatoms, dinoflagellates, green algae, haptophytes, prokaryotes, cryptophytes, and pelagophytes, are associated with distinct pigment combinations and partially distinct spectral imprints. This organization likely reflects ecological and biogeochemical constraints that cascade from community composition to pigment expression and ultimately to hyperspectral reflectance. By showing that derived spectral representations can expose this organization more clearly than raw reflectance alone, the study supports the development of more ecologically interpretable satellite hyperspectral retrieval approaches for phytoplankton composition.
Roy El Hourany, Laboratoire d’Océanologie et de Geosciences (LOG), [email protected], https://orcid.org/0000-0002-6454-1645
Sasha Kramer, Boston University, [email protected], https://orcid.org/0000-0002-9944-6779
Poster Session | 1 | 2 | 3 | 4 | Schedule at a Glance
