Across the world’s oceans, a fleet of autonomous instruments known as Argo floats are drifting with the ocean currents.  These floats are part of the Argo program, a network of profiling floats that continuously collect temperature, salinity and biogeochemical variable measurements. The Argo floats move up and down in the water column, usually down to about 2,000 meters, and every ten days, they pop back up to the surface. This process allows them to provide a global map of the state of the ocean, ocean variables and qualities.

Gridded ocean salinity data products, which can be defined as processed data that provide spatially and temporally consistent coverage for ocean salinity and have been widely used in the oceanography and climate communities, have relied heavily on these Argo float measurements since the early 2000s. 

A recent study from the University of Delaware, however, has identified an issue with these salinity products: since 2015, they tend to report false salinity increases in the world’s oceans. Known as “salty drift,” because the inaccurate salinity measurements tend to drift to higher values, the researchers conducted a study to investigate the potential consequences of this instrument drift issue on various gridded salinity products. 

The paper was published in the JGR Oceans scientific journal and was chosen by EOS as an editor’s highlight on the EOS website. 

At UD, the study was led by Xinfeng Liang, associate professor in the School of Marine Science and Policy. 

Liang said to conduct the study, researchers looked at a suite of gridded salinity products that relied on the Argo data to evaluate the products for their consistency. Two major issues arose for the gridded salinity products after 2015. They experienced a significant increase in the global mean salinity and there were also elevated inconsistences between the gridded salinity products. 

For instance, from 2015-2019, the North Indian and North Atlantic Ocean regions displayed large disagreements between gridded products when compared to the prior period of 2010-2014. 

Liang said the researchers were curious about the elevated salinity levels because glaciers and ice sheets are melting into the ocean due to climate change. This is adding more freshwater to the world’s oceans and should lead to a decrease in global salinity, the opposite of what the gridded salinity products showed.

The fact that the products show an increase in ocean salinity and start to deviate from each other more substantially is cause for concern in the scientific community as these gridded Argo products are the cornerstone of many ocean and climate related research studies. 

“People use these data to study the global water cycle, regional water circulation changes and things like that,” said Liang. “Many of those studies are based on specific salinity data and if there’s something wrong with the salinity data and you analyze the data with this salinity drift, you might get some exciting results. But, unfortunately, the results could be simply due to the issue in the data.” 

Liang said the people who work firsthand on the Argo floats are aware of the issue and they try to label the salty drift as such in the raw data directly from the floats, but the problem comes into play when researchers convert the raw data into the gridded products. There are a handful of groups around the world converting the data into gridded products and those products are ones commonly used by many researchers in their studies. 

Liang said the message for the scientific community is simple: the community should be aware of this issue, and when using gridded data products, they need to pay attention to what product they’re using. 

“There needs to be an effort, the groups need to get together to really figure out the reason for this problem and how to reduce this,” said Liang. “That’s for the people who created the data and the people who use this, they need to be aware of this salty drift.”

It also highlights the need for researchers to be more vigilant when using data for research. 

“I always mention that before you use any data, some time should be spent on knowing a little bit more about the data, about how it was produced, possible problems with the data and whether those problems could affect the conclusion of the research people are conducting,” said Liang. “Nowadays, it's so easy to go to the internet, find the data you are looking for, download the data and analyze it. Some effort needs to be spent on knowing more about the data, how the data was created, and any issues associated with the data before actually analyzing them.”