One step closer to improved tools for studying the past and measuring the future Arctic Ocean

Researchers from the Nansen Center, Spain, and the UK show in a recent study that they can reduce salinity errors in forecasts and reanalyses of the Arctic Ocean by using salinity data measured from a satellite, and a data assimilation technique.


Sea ice along the eastern coast of Greenland. Credit: contains modified Copernicus Sentinel data (2022), processed by ESA, CC BY-SA 3.0 IGOSea ice along the eastern coast of Greenland. Credit: contains modified Copernicus Sentinel data (2022), processed by ESA, CC BY-SA 3.0 IGOAll ocean water is salty, but the amount of salt varies both across the globe and with water depth. The salinity impacts ocean circulation and the marine environment. It also changes over time: Evaporation and freshwater input from river runoff, falling rain or snow, and melting ice influence the salinity.

As the Arctic Ocean is currently undergoing strong warming, more sea ice is melting, and the salt levels at the ocean surface, called sea surface salinity, are strongly affected. To produce forecasts or reanalyses for the Arctic, both for sea ice and ocean, proper information on the sea surface salinity is necessary. Forecasts are produced to make assumptions about likely changes in the future, while reanalyses are used to study the past. A reanalysis is the best possible recreation of conditions somewhere during a certain period, in a computer model, and it resembles a historical record. An example are ocean conditions, for example water temperature, sea surface salinity, and nutrient levels, in the past 20 years in the Arctic.

To make a good reanalysis, all available data on these conditions during that time are gathered and combined in a computer model. In the Arctic, lots of the surface is covered by sea ice, making it difficult to obtain salinity data of the ocean surface. This is where Arctic Ocean reanalyses become tricky to produce: The scarcity of data leads to uncertainties in how true to the past output actually is. The same goes for forecasts: If the data coverage is not optimal, errors are likely, making them less reliable.

Researchers from the Nansen Center, Spain (Institute of Marine Sciences in Barcelona and Barcelona Expert Center), and the UK (ARGANS) decided to tackle this problem. Mathematical methods like data assimilation come in handy for such work, and the team under the lead of Jiping Xie (NERSC), used salinity data taken from space and published their findings recently.

Between 2010 and 2021, the SMOS satellite mission has been investigating soil moisture and salinity across the globe. Recent improvements how to best process that data made it possible for Xie and his colleagues to assimilate as much sea surface salinity data from the Arctic as possible into a computer model for the Arctic Ocean, called TOPAZ. When assimilating data into a model, you both add information on the conditions and give the model the opportunity to learn from the data, making the output more accurate.

Jiping Xie and his colleagues compared their reanalysis to a reanalysis including only salinity data not measured from space, but in the ocean directly, from research vessels. They prove that sea surface salinity data from SMOS can be successfully assimilated and that doing so can significantly decrease errors and uncertainties in the Arctic. Their findings show that it is possible to improve reanalyses and forecasts for the Arctic Ocean with respect to sea surface salinity.



Xie, J., Raj, R. P., Bertino, L., Martínez, J., Gabarró, C., and Catany, R.: Assimilation of sea surface salinities from SMOS in an Arctic coupled ocean and sea ice reanalysis, Ocean Sci., 19, 269–287,, 2023.

Add comment

Login to post comments