Can We Estimate Air-Sea Flux of Biological O2 From Total Dissolved Oxygen?

In this study, we compare mechanistic and empirical approaches to reconstruct the air-sea flux of biological oxygen (FO2bio_as ${\mathrm{F}}_{{\left[{\mathrm{O}}_{2}\right]}_{{\text{bio}}<^>{\_\text{as}}}}$) by parameterizing the physical oxygen saturation anomaly (Delta O-2[phy]) in order to separate the biological contribution from total oxygen. The first approach matches Delta O-2[phy] to the monthly climatology of the argon saturation anomaly from a global ocean circulation model's output. The second approach derives Delta O-2[phy] from an iterative mass balance model forced by satellite-based physical drivers of Delta O-2[phy] prior to the sampling day by assuming that air-sea interactions are the dominant factors driving the surface Delta O-2[phy]. The final approach leverages the machine-learning technique of Genetic Programming (GP) to search for the functional relationship between Delta O-2[phy] and biophysicochemical parameters. We compile simultaneous measurements of O-2/Ar and O-2 concentration from 14 cruises to train the GP algorithm and test the validity and applicability of our modeled Delta O-2[phy] and FO2bio_as ${\mathrm{F}}_{{\left[{\mathrm{O}}_{2}\right]}_{{\text{bio}}<^>{\_\text{as}}}}$. Among the approaches, the GP approach, which incorporates ship-based measurements and historical records of physical parameters from the reanalysis products, provides the most robust predictions (R-2 = 0.74 for Delta O-2[phy] and 0.72 for FO2bio_as ${\mathrm{F}}_{{\left[{\mathrm{O}}_{2}\right]}_{{\text{bio}}<^>{\_\text{as}}}}$; RMSE = 1.4% for Delta O-2[phy] and 7.1 mmol O-2 m(-2) d(-1) for FO2bio_as ${\mathrm{F}}_{{\left[{\mathrm{O}}_{2}\right]}_{{\text{bio}}<^>{\_\text{as}}}}$). We use the empirical formulation derived from GP approach to reconstruct regional, inter-annual, and decadal variability of FO2bio_as ${\mathrm{F}}_{{\left[{\mathrm{O}}_{2}\right]}_{{\text{bio}}<^>{\_\text{as}}}}$ based on historical oxygen records. Overall, our study represents a first attempt at deriving FO2bio_as ${\mathrm{F}}_{{\left[{\mathrm{O}}_{2}\right]}_{{\text{bio}}<^>{\_\text{as}}}}$ from snapshot measurements of oxygen, thereby paving the way toward using historical O-2 data and a rapidly growing number of O-2 measurements on autonomous platforms for independent insight into the biological pump.