POSTER SESSION 4
10:30–12:30, Thursday, September 17
Poster Session | 1 | 2 | 3 | 4 | Schedule at a Glance
ABSTRACT 1334 | POSTER 232
EOF‐BASED HYPERSPECTRAL MODEL TO RETRIEVE PHYTOPLANKTON FUNCTIONAL TYPES ON COASTAL WATERS
Due to its complexity, phytoplankton are often represented as Phytoplankton Functional Types (PFTs), to facilitate understanding of their role in climate regulation, biogeochemical cycles, and trophic webs. Several methods currently estimate phytoplankton composition using ocean color data. This study aimed to employ in situ hyperspectral reflectance to develop and evaluate models for discriminating PFTs in a coastal region with optically complex waters, with a focus on their application to hyperspectral satellite data. To this end, scientific cruises were conducted with concurrent collection of samples for HPLC pigment analysis and hyperspectral surface reflectance data. PFT models based on chemotaxonomic groups were developed using Empirical Orthogonal Functions (EOF) and Generalized Linear Models (GLM). Model performance was assessed through cross-validation, indicating successful estimates (R > 0.75) for total chlorophyll a (TChla) and five functional groups (diatoms, cryptophytes, dinoflagellates, haptophytes, and prasinophytes). For cyanobacteria (Prochlorococcus and Synechococcus), performance was moderate (R = 0.66) and unsatisfactory (R = 0.37), respectively. In an initial application to an image from the OCI/PACE sensor, the models estimated spatial patterns of PFTs consistent with the expected coast ocean gradient in phytoplankton composition observed in the area, with diatom dominance nearshore and cyanobacteria in oligotrophic waters. Although these results indicated the potential for applying the models to remote sensing products, extensive validation with independent data is still needed. In summary, this study advances in the remote estimation of phytoplankton functional types in optically complex waters, while highlighting challenges for more reliable operational applications.
Nikolas Heinz*, Federal University of Rio Grande – FURG, [email protected], https://orcid.org/0009-0002-2258-5559
Catharina Cardoso, Federal University of Rio Grande – FURG, [email protected]
Ana Paula Forgiarini, Federal University of Rio Grande – FURG, [email protected]
Carlos Rafael Mendes, Federal University of Rio Grande – FURG, [email protected]
Áurea Ciotti, Universidade de São Paulo – USP, [email protected]
Poster Session | 1 | 2 | 3 | 4 | Schedule at a Glance
