Assessing the potential of backscattering as a proxy for phytoplankton carbon biomass

Abstract Despite phytoplankton contributing roughly half of the photosynthesis on earth and fueling marine food-webs, field measurements of phytoplankton biomass remain scarce. The particulate backscattering coefficient ( b bp ) has often been used as an optical proxy to estimate phytoplankton carbon biomass ( C phyto ). However, total observed b bp is impacted by phytoplankton size, cell composition, and non-algal particles. The lack of phytoplankton field data has prevented the quantification of uncertainties driven by these factors. Here, we first review and discuss existing b bp algorithms by applying them to b bp data from the BGC-Argo array in surface waters ( < 10m). We find a b bp threshold where estimated C phyto differs by more than an order of magnitude. Next, we use a global ocean circulation model (the MITgcm Biogeochemical and Optical model) that simulates plankton dynamics and associated inherent optical properties to quantify and understand uncertainties from b bp -based algorithms in surface waters. We do so by developing and calibrating an algorithm to the model. Simulated error-estimations show that b bp -based algorithms overestimate/underestimate C phyto between 5% and 100% in surface waters, depending on the location and time. This is achieved in the ideal scenario where C phyto and b bp are known precisely. This is not the case for algorithms derived from observations, where the largest source of uncertainty is the scarcity of phytoplankton biomass data and related methodological inconsistencies. If these other uncertainties are reduced, the model shows that b bp could be a relatively good proxy for phytoplankton carbon biomass, with errors close to 20% in most regions. Plain Language Summary Phytoplankton contribute roughly half of the photosynthesis on earth and fuel fisheries around the globe. Yet, few direct measurements of phytoplankton concentration are available. Frequently, concentrations of phytoplankton are instead estimated using the optical properties of water. Backscattering is one of these optical properties, representing the light being scattered backwards. Previous studies have suggested that backscattering could be a good method to estimate phytoplankton concentration. However, other particles that are present in the ocean also contribute to backscattering. In this paper we examine how well backscattering can be used to estimate phytoplankton. To address this question, we use data from drifting instruments that are spread across the ocean and a computer model that simulates phytoplankton and backscattering over the global oceans. We find that by using backscattering, phytoplankton can be overestimated/underestimated on average by ∼20%. This error differs between regions, and can be larger than 100% at high latitudes. Computer simulations allowed us to quantify spatial and temporal variability in backscattering signal composition, and thereby understand potential errors in inferring phytoplankton with backscattering, which could not have been done before due to the lack of phytoplankton data. Key Points Phytoplankton carbon b bp -based algorithms can differ up to an order of magnitude at low b bp values. An algorithm fitted to a global model output shows biases ranging between 15% and 40% in most regions. Most uncertainties are due to the relative contribution of phytoplankton to total b bp .