The European Centre for Medium-Range Weather Forecasts (ECMWF) is preparing to release Ocean ReAnalysis System 6 (ORAS6), its sixth-generation ocean and sea-ice reanalysis system, representing more than eight years of development over its predecessor, ORAS5.
Ocean reanalyses combine observational data with physics-based numerical models to reconstruct past ocean and sea-ice conditions – a process that addresses the uneven distribution of observations across time and space, particularly in earlier decades of the record. The resulting datasets support climate monitoring and weather forecasting across timescales from medium-range prediction to seasonal outlooks.
A key advance in ORAS6 is the introduction of an ensemble-based data assimilation system, which for the first time allows direct assimilation of sea-surface temperature (SST) observations. Combined with hourly atmospheric data from ERA5, this enables the system to better capture short-term variations including the daily warming and cooling cycle at the ocean surface. Comparisons with drifting ocean observations show significantly closer agreement with observed daily temperature changes than in previous versions.
Other improvements include better simulation of the Gulf Stream, improved sea-surface height, and enhanced representation of sea-ice concentration.
“ORAS6 brings together more than eight years of development and introduces important advances, including improved representation of sea-surface temperature, ocean circulation and sea ice,” said Hao Zuo, senior scientist in ocean data assimilation. “Together, these changes provide a more accurate picture of the evolving ocean state.”
“By improving how we represent upper-ocean conditions, in particular surface temperatures and sea ice, ORAS6 gives us a more realistic starting point for Earth system forecasts and reanalysis,” added Phil Browne, senior scientist in coupled ocean-atmosphere data assimilation.
The improvements have practical forecasting benefits. In medium-range forecasting, ORAS6 means ECMWF no longer requires the previous workaround of restricting full ocean-atmosphere coupling outside the tropics. Gains are also seen in subseasonal and seasonal forecasting, particularly for SST.
ORAS6 will also provide ocean and sea-ice initial conditions for ERA6, the upcoming global reanalysis dataset produced as part of the EU-funded Copernicus Climate Change Service, and has been used to help train ECMWF’s data-driven machine learning models.
When complete, the system will cover the period from 1950 to the present day. Data for the altimeter period from 1993 onward is now being consolidated, with release to member states and external collaborators expected shortly. Production of the earlier record from 1950 onward is continuing, with completion expected in August 2026.
In related news, Colorado State University and Nvidia partner to extend severe storm prediction lead times