Ocean reanalyses that combine ocean models with observations through data assimilation schemes are alternative data resources compared to pure observations or numerical-model-based datasets. The advantage of ocean reanalyses data is that they provide a temporal and spatial ocean state hindcast compared to observation datasets, and they correct model biases of the ocean state using observations as reference data compared to pure numerical model simulations.

These advantages of ocean reanalyses allow a wide range of uses of the datasets, for example, ocean process-based scientific studies, monitoring and understanding ocean variability and changes, reference data for evaluating climate models, boundary conditions for regional models, and initial conditions for ocean forecasts and predictions. In particular, the fast-growing field of artificial intelligence for ocean forecasting requires high-quality ocean reanalyses to train the forecast in order to provide trustworthy AI-based ocean forecasts. Therefore, the performance of ocean reanalyses or, in other words, the quality of ocean reanalyses, is the backbone of the usage of ocean reanalyses in different applications or studies.

This special issue has emerged from the Marine Environment Reanalyses Evaluation Project (MER-EP), an initiative of the Ocean Decade action, but is open to all submissions on the systematic evaluation of marine environment reanalyses to provide guidance to marine reanalysis users (e.g. scientific or private sectors). Papers on all aspects of the evaluation and production of reanalysis products, including physics, biogeochemistry, small-scale processes, and waves, and from polar to equatorial regions are invited.

Review process: all papers of this special issue underwent the regular interactive peer-review process of Ocean Science handled by members of the editorial board as well as guest editors designated by the OS co-editors-in-chief.