Dissolved oxygen (O2) is a central observation in oceanography with a long history of relatively high precision measurements and increasing coverage over the 21st century. O2 is a powerful tracer of physical, chemical and biological processes (e.g., photosynthesis and respiration, wave-induced bubbles, mixing, and air-sea diffusion). A commonly used approach to partition the processes controlling the O2 signal relies on concurrent measurements of argon (an inert gas), which has solubility properties similar to O2. However, only a limited fraction of O2 measurements have paired argon measurements.
A recent study published in the Journal of Global Biogeochemical Cycles presents semi-analytical algorithms to separate the biological and physical O2 oxygen signals from O2 observations. Among the approaches, a machine-learning algorithm using ship-based measurements and historical records of physical parameters from reanalysis products as predictors shows encouraging performance. The researchers leveraged this new algorithm to reconstruct regional, inter-annual, and decadal variability of the air-sea flux of biological oxygen (from historical O2 records.The long-term objective of this proof-of-concept effort is to estimate from historical oxygen records and a rapidly growing number of O2 measurements on autonomous platforms. In regions where vertical and horizontal mixing is weak, the projected approximates net community production, providing an independent constraint on the strength of the biological carbon pump.
Authors:
Yibin Huang (Duke University)
Rachel Eveleth (Oberlin College)
David (Roo) Nicholson (Woods Hole Oceanographic Institution)
Nicolas Cassar (Duke University)