New insight into the seasonal carbon dynamics of the global ocean

Researchers at the Max Planck Institute for Meteorology (MPI-M) and colleagues used a machine learning approach to reconstruct a monthly climatology of Mapped Observation-Based Oceanic Dissolved Inorganic Carbon (MOBO-DIC) in the upper ocean. The study, which has just been published in Global Biogeochemical Cycles, was led by Dr Lydia Keppler, from Dr Peter Landschützer’s group “Observations, Analysis and Synthesis” and involved collaborations with the former MPI-M member Dr Irene Stemmler and external collaborators from ETH Zurich (Prof Nicolas Gruber) and the Norwegian Research Center NORCE (Dr Siv K. Lauvset). Using this new data product, the authors were able to describe the phase and amplitude of the contemporary seasonal cycle of dissolved inorganic carbon (DIC) around the globe and estimate how much DIC is removed through biological activity between spring and fall.

Temporal mean DIC from MOBO-DIC. Left: Map of the temporal mean DIC at 10 m. Right: Vertical section of the temporal mean global DIC. Black contour lines are drawn every 50 µmol/kg. Image: L. Keppler, MPI-M

The natural carbon cycle varies substantially on seasonal time scales in the upper ocean. As a result of anthropogenic carbon dioxide emissions, changes in the seasonal DIC cycle have been suggested by theoretical studies and model predictions. These potential changes can exacerbate the impact of ocean acidification on marine organisms, for example, through an earlier onset of critical thresholds. Dr Lydia Keppler says “investigating the present seasonal cycle of DIC forms a crucial basis for investigating future changes in these patterns.”

However, DIC measurements are extremely sparse, rendering it impossible to produce global time-varying DIC fields using traditional statistical interpolation techniques. Therefore, the authors modified a machine learning approach to interpolate the sparse measurements with the aid of more frequently measured proxy data, such as temperature and salinity.

Using this new data product, the authors find the largest seasonal amplitudes in surface DIC in the northern high latitude Pacific. Surface DIC maxima tend to occur in hemispheric spring and minima in fall, driven by the input of DIC into the upper ocean by mixing during winter and the biological draw-down of DIC over summer. Based on the seasonal differences in DIC, the authors estimate the global biological draw-down of inorganic carbon between spring and fall to be 8.2±5.6 Pg C yr-1.

 “This study illustrates the potential of neural networks and how they can help us to maximize the information provided by sparse observations”, says Dr Peter Landschützer.

Original publication:

Keppler, L., Landschützer, P., Gruber, N., Lauvset, S. K., & Stemmler, I. (2020). Seasonal Carbon Dynamics in the Near‐Global Ocean. Global Biogeochemical Cycles, 34, e2020GB006571. Accepted Author Manuscript.

Data set available at


Dr Lydia Keppler

Max Planck Institute for Meteorology

Email: lydia.keppler@we dont want


Dr Peter Landschützer

Max Planck Institute for Meteorology

Email: peter.landschuetzer@we dont want