Date of Award


Level of Access Assigned by Author

Campus-Only Dissertation

Degree Name

Doctor of Philosophy (PhD)




Mary Jane Perry

Second Committee Member

Collin Roesler

Third Committee Member

Andrew Barnard


Thirty years of ocean color remote sensing has provided unparalleled capabilities for quantifying global phytoplankton biomass and productivity. However, coastal margins, such as the Gulf of Maine (GOM) are more optically and biologically complex. In these regions, ocean color is confounded by contamination of satellite-estimated chlorophyll a concentration ([Chlsat]) by non-covariant colored dissolved organic matter (CDOM) and variability in the accuracy of [Chlsat] as a consequence of a shifting phytoplankton composition. Through analysis of bio-optical and discrete surveys of the GOM's upper water column, the seasonal variability in CDOM and phytoplankton absorption, the character and diversity of particulate and dissolved matter, and the sources, sinks, and forcing mechanisms that act on this material are described. In addition, we assessed the makeup of empirical [Chlsat] algorithms to determine sources of absorption and scattering bias leading to [Chlsat] estimation error. Our goal is to understand the sources of ocean color and provide more accurate [Chlsat] for primary production and ecosystem models that assess the broader health of the ocean. The GOM is a seasonally and spatially diverse and complex optical environment. Variations in component absorption were conserved properties of GOM hydrographic provinces even when influenced by winter mixing events. While diverse CDOM magnitude and character existed regionally due to stratification, broad basin scale uniformity developed with respect to CDOM spectral slope in fall and CDOM absorption in spring. Phytoplankton absorption and particle type were found to be linked to mixing resulting in dramatic seasonal and regional differences. The gradient in the OC3v5 [Chlsat] algorithm resulted from a balance between phytoplankton (aph) and detrital (adm) absorption in the algorithm training dataset. Underestimates of [Chlsat] occurred where aph dominated and overestimates where adm dominated. Detrital absorption variance dominated the variance in OC3v5 while only a narrow order of magnitude variability in [Chl]-specific absorption cross-section was found. These results suggest that regional (GOM) or global historical records of [Chlsat] lack or will lack the coincidence in IOP variability found in the mean global MBR [Chlsat] algorithms. Inevitably, this will lead to [Chlsat] estimate errors and the potential for implausible conclusions regarding trends in phytoplankton productivity.