Variations of ocean heat content modulate the Earth's system energy and affect many physical and biogeochemical ocean processes, such as sea level rise, increased hurricane intensity, reduced biodiversity, etc. It is thus important to monitor the ocean heat content change for long, multi-decadal, periods, to understand its low-frequency variations, disentangle external versus internal forcing, and assess the uncertainty associated with the change.

In our article, we use a recently developed ensemble ocean reanalysis system 1961-2022 (http://cigar.ismar.cnr.it) to assess the ocean warming rate, its acceleration, the associated uncertainty, and the main sources of errors. The global warming acceleration is also confirmed by the fact that the most recent year investigated in the study (2022) owns, by far, the largest fraction of ocean points with maximum ocean heat content compared to previous years.

Why the ocean is warming? The ocean is globally warming because of its role in accumulating the majority of the excess Earth’s system energy, caused in turn by the increase of climate-altering gas concentrations. It is estimated that around 90% of this excess energy, the so-called Earth Energy Imbalance (EEI) is absorbed by the oceans. The warming acceleration occurs during the process of ocean heat uptake and corresponds to an increase of net positive energy flux into the oceans across the atmosphere-ocean and ice-ocean interfaces.

Not only the ocean is globally warming and its warming accelerates, but locally there occur even larger accelerations and disparities, due to the redistribution of the heat and appearance of heat convergence and divergences zones because of changes in atmospheric and oceanic circulation.

In our study, it is the first time that an ocean reanalysis – merging observational data, a state-of-the-science numerical ocean-sea-ice model, and boundary conditions coming from the best estimates of the atmospheric state and freshwater inflow from land – is used to investigate the ocean warming acceleration over 60+ years, its acceleration and uncertainty, and rank the main sources of uncertainty. Although reanalyses were often seen in the past to exhibit spurious variability, the main idea behind our dataset is to adopt an ensemble approach (i.e., turning the reconstruction of the ocean climate from a deterministic to a probabilistic picture), which can include all sources of uncertainty, and lead to an accurate assessment of the uncertainty and confidence level of the reconstructed four-dimensional state of the ocean.

This is crucial for studies encompassing several decades. Indeed, while the global Argo program (http://www.argo.net) has revolutionized our knowledge of the ocean warming evolution since about the mid-2000s, before the 2000s the observing networks were not mature enough to allow a high-confidence reconstruction of the ocean heat content variations. Consequently, more sophisticated methods than just observational mapping may be considered, and reanalyses can integrate all the information from atmospheric and oceanic observations with our knowledge of ocean physics.

Our study demonstrates that ensemble ocean reanalyses can be used to further extend back in past our ocean climate reconstruction, which is made possible by similar efforts in the atmospheric community, data rescue and processing programs, and methodological improvements (bias correction, ensemble generation, ocean modeling, etc.). Back-extension and evaluation of other climate change proxies will be the natural evolution of our research.