A Multivariable Empirical Algorithm for Estimating Particulate Organic Carbon Concentration in Marine Environments From Optical Backscattering and Chlorophyll-a Measurements

• Published in: Frontiers in Marine Science
• Main Authors: Koestner, Daniel Warren, Stramski, Dariusz, Reynolds, Rick A.
• Format: Article in Journal/Newspaper
• Language: English
• Published: Frontiers 2022
• Subjects: Arctic; Pacific; Pacific Arctic
• Online Access: https://hdl.handle.net/11250/3061095; https://doi.org/10.3389/fmars.2022.941950

Accurate estimates of the oceanic particulate organic carbon concentration (POC) from optical measurements have remained challenging because interactions between light and natural assemblages of marine particles are complex, depending on particle concentration, composition, and size distribution. In particular, the applicability of a single relationship between POC and the spectral particulate backscattering coefficient bbp(λ) across diverse oceanic environments is subject to high uncertainties because of the variable nature of particulate assemblages. These relationships have nevertheless been widely used to estimate oceanic POC using, for example, in situ measurements of bbp from Biogeochemical (BGC)-Argo floats. Despite these challenges, such an in situbased approach to estimate POC remains scientifically attractive in view of the expanding global-scale observations with the BGC-Argo array of profiling floats equipped with optical sensors. In the current study, we describe an improved empirical approach to estimate POC which takes advantage of simultaneous measurements of bbp and chlorophyll-a fluorescence to better account for the effects of variable particle composition on the relationship between POC and bbp. We formulated multivariable regression models using a dataset of field measurements of POC, bbp, and chlorophyll-a concentration (Chla), including surface and subsurface water samples from the Atlantic, Pacific, Arctic, and Southern Oceans. The analysis of this dataset of diverse seawater samples demonstrates that the use of bbp and an additional independent variable related to particle composition involving both bbp and Chla leads to notable improvements in POC estimations compared with a typical univariate regression model based on bbp alone. These multivariable algorithms are expected to be particularly useful for estimating POC with measurements from autonomous BGC-Argo floats operating in diverse oceanic environments. We demonstrate example results from the multivariable algorithm applied to ...