Detection of coastal phytoplankton taxonomy, morphology and optical signals with machine learning and time series analysis

Author

Virginie Sonnet, University of Rhode IslandFollow

Date of Award

2023

Degree Type

Dissertation

Degree Name

Doctor of Philosophy in Oceanography

Specialization

Biological Oceanography

Department

Oceanography

First Advisor

Colleen Mouw

Abstract

Phytoplankton are microscopic organisms crucial in the food web, primary production, and carbon export. Their abundance and diversity changes also influence the evolution of marine ecosystems and global biogeochemical cycles. In addition to extensive taxonomic diversity, phytoplankton can harbor different sets of photosynthetic pigments and display a wide range of size, morphologies, and functional traits. These particularities have led phytoplankton monitoring and phytoplankton dynamics studies to develop in many forms. Satellite remote sensing, light microscopy, and more recently particle imaging, are among those. However, they not only have widely different sampling resolution and taxonomic precision, but also target different properties of phytoplankton. Light microscopy, the historical monitoring technique, is sparse, time-intensive, and requires extensive taxonomic knowledge and training, but provides the highest taxonomic resolution. On the other hand, satellite remote sensing has the broadest global spatial coverage but provides data irregularly in coastal areas and focuses on broader phytoplankton groups due to the complexity of parsing out phytoplankton communities from their optical spectra. In the last decade, particle imaging instruments have developed in capabilities and versatility, bringing an additional layer of sampling possibilities to the existing methods. It offers the possibility to go beyond the species level to the individual level, and an opportunity to evaluate, complement and enhance well-established phytoplankton sampling methods like satellite remote sensing and light microscopy.

Recommended Citation

Sonnet, Virginie, "Detection of coastal phytoplankton taxonomy, morphology and optical signals with machine learning and time series analysis" (2023). Open Access Dissertations. Paper 1611.
https://digitalcommons.uri.edu/oa_diss/1611