Biogeosciences (Apr 2023)

Reconstructing ocean carbon storage with CMIP6 Earth system models and synthetic Argo observations

  • K. E. Turner,
  • D. M. Smith,
  • A. Katavouta,
  • A. Katavouta,
  • R. G. Williams

DOI
https://doi.org/10.5194/bg-20-1671-2023
Journal volume & issue
Vol. 20
pp. 1671 – 1690

Abstract

Read online

The ocean carbon store plays a vital role in setting the carbon response to emissions and variability in the carbon cycle. However, due to the ocean's strong regional and temporal variability, sparse carbon observations limit our understanding of historical carbon changes. Ocean temperature and salinity profiles are more widespread and rapidly expanding due to autonomous programmes, and so we explore how temperature and salinity profiles can provide information to reconstruct ocean carbon inventories with ensemble optimal interpolation. Here, ensemble optimal interpolation is used to reconstruct ocean carbon using synthetic Argo temperature and salinity observations, with examples for both the top 100 m and top 2000 m carbon inventories. When considering reconstructions of the top 100 m carbon inventory, coherent relationships between upper-ocean carbon, temperature, salinity, and atmospheric CO2 result in optimal solutions that reflect the controls of undersaturation, solubility, and alkalinity. Out-of-sample reconstructions of the top 100 m show that, in most regions, the trend in ocean carbon and over 60 % of detrended variability can be reconstructed using local temperature and salinity measurements, with only small changes when considering synthetic profiles consistent with irregular Argo sampling. Extending the method to reconstruct the upper 2000 m reveals that model uncertainties at depth limit the reconstruction skill. The impact of these uncertainties on reconstructing the carbon inventory over the upper 2000 m is small, and full reconstructions with historical Argo locations show that the method can reconstruct regional inter-annual and decadal variability. Hence, optimal interpolation based on model relationships combined with hydrographic measurements can provide valuable information about global ocean carbon inventory changes.