Evaluating Optical Proxies of Particulate Organic Carbon across the Surface Atlantic Ocean

Abstract

Empirical relationships between particulate organic carbon (POC) and inherent optical properties (IOPs) are required for estimating POC from ocean-color remote sensing and autonomous platforms. The main relationships studied are those between POC and particulate attenuation (c p) and backscattering (b bp) coefficients. The parameters of these relationships can however differ considerably due to differences in the methodologies applied for measuring IOPs and POC as well as variations in particle characteristics. Therefore, it is important to assess existing relationships and explore new optical proxies of POC. In this study, we evaluated empirical relationships between surface POC and IOPs (c p , b bp and the particulate absorption coefficient, a p) using an extensive dataset collected during two Atlantic Meridional Transect (AMT 19 and 22) cruises spanning a wide range of oceanographic regimes. IOPs and POC were measured during the two cruises using consistent methodologies. To independently assess the accuracy of the POC-IOPs relationships, we predicted surface POC for AMT-22 using relationships developed based on independent data from AMT-19. We found typical biases in predicting POC ranging between 2-3, 4-9, and 6-13% for c p , b bp , and a p , respectively, and typical random uncertainties of 20-30%. We conclude that (1) accurate POC-c p and POC-b bp relationships were obtained due to the consistent methodologies used to estimate POC and IOPs and (2) a p could be considered as an alternative optical proxy for POC in open-ocean waters, only if all physiological variability in the POC:chl ratio could be modeled and used to correct a p .

Approach for Propagating Radiometric Data Uncertainties Through NASA Ocean Color Algorithms

“…We thus wish to demonstrate how with knowledge of model uncertainties one can draw more complete conclusions about biogeochemicallyrelevant data product uncertainties. As such, we present a case study in which we estimate POC measurement uncertainty for two different algorithms: (i) Stramski et al (2008a) and (ii) Rasse et al (2017). Our motivation here is to solely demonstrate how one might develop algorithm uncertainty budgets (data and model uncertainty as per Equation 1) using a FOFM framework.…”

“…In this exercise, we performed rudimentary calculations to estimate measurement uncertainty budgets for two POC algorithms: (i) NASA's standard POC algorithm (Stramski et al, 2008a) and (ii) the IOP-based model of Rasse et al (2017). Conveniently for this exercise, both POC models have a power law formulation:…”

“…For the approach of Rasse et al (2017) X is b bp , 470 and the coefficients a poc and b poc have the values of 141,253 and 1.18, respectively. Note, in this case study we use GIOP-derived estimates of b bp , 470 as inputs to the Rasse et al (2017) model. First, let us consider the model uncertainty component due to imperfect model coefficients.…”

“…In the third term on the right-hand side of Equation 8, we set u model (X) = 0 and u model (X) = u model (b bp,470 ) for Stramski et al (2008a) and Rasse et al (2017), respectively. We have also assumed the covariance of the coefficients a poc and b poc , which are determined by regression fit, is zero.…”

“…We have also assumed the covariance of the coefficients a poc and b poc , which are determined by regression fit, is zero. For the Rasse et al (2017) model, the reported model coefficient uncertainties u model (a poc ) and u model (b poc ) are 45,534 and 0.046, respectively. For the Stramski et al (2008a) model, values of u model (a poc ) and u model (b poc ) were not reported.…”