A comparative study of the Argo-era ocean heat content among four different types of datasets

Fanglou Liao, Ibrahim Hoteit

Research output: Contribution to journalArticlepeer-review


We conducted a comparative study of ocean heat content (OHC) in the top 2000 m during the Argo-era using 12 latest and representative global ocean datasets. The differences in the global and basins-wide OHC trends were minor among the observation-based datasets, and remarkable among the ocean reanalyzes (RAs). Some RAs might exhibit much higher or lower basins-wide warming rates than the observation-based datasets. In the top 700 m, RAs suggested similar large-scale warming and cooling patterns, in agreement with the observation-based datasets. Below 700 m, the major warming and cooling features were however significantly different between RAs and observation-based datasets. All datasets suffered from relatively larger uncertainties in the highly dynamic regions. Special caution is suggested when estimating the OHC using only a single dataset, especially a RA. Differences of RAs’ OHC from observation-based datasets were significantly reduced when considering their ensemble mean, to be further confirmed with a larger sample of datasets.
Original languageEnglish (US)
JournalEarth's Future
StatePublished - Aug 19 2022

Bibliographical note

KAUST Repository Item: Exported on 2022-09-14
Acknowledgements: We thank the two reviewers for providing their helpful comments and suggestions. The Argo-related data were collected and made freely available by the International Argo Program and the national programs that contribute to it (https://argo.ucsd.edu,https://www.ocean-ops.org). The Argo Program is part of the Global Ocean Observing System. Efforts taken to collect, process and deliver other observations (such as XBT) are also greatly appreciated. We are also grateful for the publicly available datasets used in this work.


Dive into the research topics of 'A comparative study of the Argo-era ocean heat content among four different types of datasets'. Together they form a unique fingerprint.

Cite this