Atmospheric Sciences News

Take Weather Prediction with a Grain of Salt and It Gets Better

Sea surface salinity is starting to rival other methods for seasonal rain forecasting.


A surplus of salt somewhere on the ocean can mean a downpour somewhere on land.

Sea surface salinity is turning into a powerful tool for seasonal weather prediction, researchers claim. It rivals and sometimes surpasses other methods for long-range prediction of precipitation such as taking stock of sea surface temperatures and the El Niño–Southern Oscillation (ENSO).

Using salinity works especially well a few months to even half a year ahead, said Caroline Ummenhofer of the Woods Hole Oceanographic Institution, who developed the method with her colleague Raymond Schmitt. Ummenhofer is presenting a comparison of the skills, or predictive abilities, of sea surface salinity and sea surface temperature at the European Geosciences Union General Assembly in Vienna this week.

“The idea behind it is that a lot of the water that falls on land ultimately comes from the ocean—that’s where most of the water is. As the water evaporates from the ocean, it leaves behind increased salinity. Therefore, before you have a major rainfall event, you think you need to have a change in the salinity somewhere,” Ummenhofer explains.

“Somewhere” in the Midwest and Sahel 

The challenge was to find out where “somewhere” was.

In two case studies, Ummenhofer and Schmitt, with Li Laifang, now at Duke University, applied the method to the record-breaking spring and summer rainfall in 2015 in the southern and midwestern United States and to rainfall patterns in the Sahel region of Africa.

In both cases, the source of moisture was the subtropical North Atlantic. For the United States, researchers found increased salinity in the months prior to the rains in the northwestern part of the ocean. For the Sahel, the salinity of the central to eastern part of the ocean in the spring predicted the intensity of the summer monsoon.

“But these were just case studies, and so we wanted to explore more systematically the relative skill and benefit of using sea surface salinity over sea surface temperature,” Ummenhofer said.

To make the comparison, the researchers divided the United States into regions on the basis of their precipitation characteristics, such as the Southwest and the Midwest, and tried to find connections between those regions and temperature and rainfall anywhere on the world’s oceans.

It turned out that the best links were found through salinity for some regions and through temperature for others.

“Which is interesting,” says Ummenhofer, “because you could imagine that those would be actually co-located: wherever you have warmer temperatures, there you have evaporation loss; you could expect both of them happening in the same place. Also, salinity does much better at longer lead times, 3 to 6 months out, whereas temperature does better at shorter time scales. We’re still figuring out why.”

Weather predictors should be aware of the relative benefits of salinity and temperature, says Ummenhofer. “The best skill you get when you combine the two.”

Important Instrument in the Forecasting Toolbox

Even maximized, this prediction skill is not like the daily forecast. It’s still not possible to tell Americans with any certainty what kind of summer they will be having.

“About a quarter of the variability is coming from these predictors. That doesn’t sound very good, but even for areas in the tropics that are influenced by ENSO, which is one of our best climate predictors, you can only explain about a quarter of the variance for those areas,” Ummenhofer said. “For the Sahel area, sea surface salinity does better than any other climate predictors that we have, including ENSO.”

The work of Ummenhofer, Schmitt, and Li impresses Tony Lee, supervisor of the Ocean Circulation and Air-Sea Interaction Group at NASA’s Jet Propulsion Laboratory, who was not involved with the study. “The method clearly has practical application value,” he wrote in an email. “In addition, their related publications had explained underlying physical processes.”

According to Lee, this research calls for intensified observations, “for example, by increasing the temporal sampling of satellite salinity measurements to daily or a few days—currently we can only get complete coverage of the global ocean from a satellite on a weekly interval or so.”

—Bas den Hond, Science Writer

Citation: den Hond, B. (2019), Take weather prediction with a grain of salt and it gets better, Eos, 100, Published on 12 April 2019.
Text © 2019. The authors. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.