In-depth learning takes Saturn's assault

According to researchers at UCL and the University of Arizona, an "in-depth learning" approach to detect storms on Saturn should transform our understanding of planetary atmospheres.

The new technique, called PlanetNet, identifies and maps the components and features of the turbulent regions of Saturn's atmosphere, giving insight into the processes that drive them.

A study, published today in Nature Astronomy, provides the results of the first demonstration of the PlanetNet algorithm, which clearly shows the vast regions affected by storms and that the dark storm clouds of Saturn contain materials swept from the lower atmosphere by strong vertical winds.

Developed by researchers from UCL and the University of Arizona, PlanetNet has been trained and tested using infrared data from Visible and Infrared Spectrometer (VIMS) on Cassini, joint mission of NASA, the European Space Agency and the Italian Space Agency.

A dataset containing several adjacent storms observed at Saturn in February 2008 was chosen to provide a range of complex atmospheric features to challenge PlanetNet's capabilities.

A previous analysis of the dataset indicated a rare detection of ammonia in the atmosphere of Saturn, in the form of an S-shaped cloud.

The map produced by PlanetNet shows that this feature is an important part of a much larger rise of ammonia ice clouds around a dark, central storm. PlanetNet identifies a similar comeback around another small storm, suggesting that such features are quite common.

The map also shows pronounced differences between the storm center and the surrounding areas, indicating that the eye can clearly see the atmosphere warmer and deeper.

"Missions like Cassini collect huge amounts of data, but conventional analysis techniques have drawbacks, whether in terms of accuracy of information or time extraction." learning "allows the recognition of models in various data sets," said Dr. Ingo Waldmann (UCL Physics & Astronomy), senior author and deputy director of the UCL Center for Spatial Data and Exoplanets.

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

As a first step, PlanetNet searches the data for signs of clustering in the cloud structure and gas composition. For areas of interest, it limits data to eliminate edge uncertainties and performs parallel analysis of spectral and spatial properties. By combining the two streams of data, PlanetNet creates a map that quickly and accurately presents the major components of Saturn's storms with unprecedented accuracy.

PlanetNet's accuracy has been validated on Cassini data not included in the training phase. The complete data set has also been rotated and resampled to create "synthetic" data for later testing. PlanetNet has achieved classification accuracy of more than 90% in both test cases.

"PlanetNet allows us to analyze much larger volumes of data, which allows us to better understand the large-scale dynamics of Saturn," said Professor Caitlin Griffith (University of Arizona), co – author of this article. "The results reveal previously unrecognized atmospheric characteristics, and PlanetNet can easily be adapted to other data sets and planets, making it an invaluable potential tool for many future missions."