OneArgo Enables New Gap-filled Data Products of Ocean Biogeochemistry

Although ocean biogeochemistry plays an important role in the regulation of Earth’s climate and marine habitats, key questions remain about expected changes to global biogeochemical processes associated with anthropogenic impacts on the Earth system. The OneArgo array is a revolutionary ocean observing system that delivers critical observations in four dimensions and in near-real-time. Recent expansion of the Argo program to include floats carrying a suite of biogeochemical sensors (i.e. the BGC-Argo mission) is providing new opportunities to study critical processes such as primary production and carbon export, ocean acidification and deoxygenation, and air–sea gas fluxes. This work describes an approach to leverage those BGC sensor observations, along with shipboard and core Argo float observations, to construct time-varying data products of dissolved oxygen, nitrate, and pH for the upper 2 km of the ocean. The products are constructed by training empirical machine learning (ML) models that take advantage of relationships between ocean biogeochemical and physical parameters, along with the widespread distribution of BGC training data available from the BGC-Argo array and the broad coverage of ocean physical parameters provided by the core Argo array. Improvements relative to earlier approaches for BGC data product creation include an objective clustering step to identify regions of similar variability prior to model training, a comprehensive evaluation of ML model uncertainty using Earth system model testbeds, and a higher resolution grid over both space and time. Potential opportunities to compute net community production, monitor ocean acidification extreme events, and evaluate ocean deoxygenation are explored using these novel data products.