'Deep Learning' Algorithm Reveals Huge Saturn Storm in New Light

These images of a storm in Saturn's atmosphere were obtained with the Cassini spacecraft’s wide-angle camera on March 4, 2008, at a distance of approximately 800,000 miles (1.3 million kilometers) from Saturn. Image scale is 46 miles (74 km) per pixel.
These images of a storm in Saturn's atmosphere were obtained with the Cassini spacecraft’s wide-angle camera on March 4, 2008, at a distance of approximately 800,000 miles (1.3 million kilometers) from Saturn. Image scale is 46 miles (74 km) per pixel.
(Image: © NASA/JPL/Space Science Institute)

A new "deep learning" algorithm that could help scientists better understand planetary atmospheres has passed its first big test, a new study reports.

The software, called PlanetNet, mapped out a monster 2008 Saturn storm system in detail using data gathered by NASA's Cassini spacecraft, which studied the ringed planet up close from 2004 through 2017.

"Missions like Cassini gather enormous amounts of data, but classical techniques for analysis have drawbacks, either in the accuracy of information that can be extracted or in the time they take to perform. Deep learning enables pattern recognition across diverse, multiple data sets," study co-lead author Ingo Waldmann, deputy director of the Center for Space and Exoplanet Data at University College London in England, said in a statement.

Related: Amazing Saturn Photos From NASA's Cassini Orbiter

Cloud distribution as mapped by the PlanetNet algorithm across six overlapping data sets. The stormy region feature (blue) occurs in the vicinity of dark storms (purple/green) in contrast to the unperturbed regions (red/orange). The area covered by the multiple-storm system is equivalent to about 70% of the Earth's surface.

(Image: © Waldmann and Griffith/Nature Astronomy)

"This gives us the potential to analyze atmospheric phenomena over large areas and from different viewing angles, and to make new associations between the shape of features and the chemical and physical properties that create them," Waldmann added.

PlanetNet searches data sets for evidence of "clustering" in cloud structure and atmospheric composition, then uses such information to generate precise maps. Waldmann and study co-leader Caitlin Griffith, of the University of Arizona's Lunar and Planetary Laboratory, trained and tested the algorithm using data gathered by Cassini's Visible and Infrared Mapping Spectrometer (VIMS) instrument.

For the new study, which was published online today (April 29) in the journal Nature Astronomy, the duo chose a data set containing VIMS observations of a multiple-storm system that boiled on Saturn in February 2008. This was meant to be a challenge, for the system was complex and quite large. Combined, its various components covered area equivalent to about 70% of Earth's surface, the researchers said.

PlanetNet took this information and ran with it, providing new insights into the storms. For example, its maps showed that a previously observed S-shaped cloud of ammonia was actually part of a much bigger upwelling that surrounded a dark storm. And PlanetNet detected a similar feature around a different storm, suggesting that ammonia-ice upwellings are common in Saturn's atmosphere, the researchers said. 

"PlanetNet enables us to analyze much bigger volumes of data, and this gives insights into the large-scale dynamics of Saturn," Griffith said in the same statement. "The results reveal atmospheric features that were previously undetected. PlanetNet can easily be adapted to other datasets and planets, making it an invaluable potential tool for many future missions."

Mike Wall's book about the search for alien life, "Out There" (Grand Central Publishing, 2018; illustrated by Karl Tate), is out now. Follow him on Twitter @michaeldwall. Follow us on Twitter @Spacedotcom or Facebook

Have a news tip, correction or comment? Let us know at community@space.com.