radial bone gas temperature maps of the 3D AFGKM exolad planets orbiting various host stars (top to bottom: A, G, M), Tglobal, 800k (left), 1600k (center), and 2400k (right). (TGA, PGAS) Mapgas is displayed as an equatorial slice plot. – Astro-Ph.ep
The development of telescopes that could allow for more detailed observation of Explonet atmospheres has led to an increasing demand for enhanced 3D climate models to support and interpret observational data from space missions such as Chop, Tess, JWST, Plato, and Ariel.
However, the computationally intensive and time-consuming nature of general cyclical models (GCMs) poses important challenges in simulating a wide range of exoplanet atmospheres. The purpose of this study is to determine whether machine learning (ML) algorithms can be used to determine whether 3D temperature and wind structure of any attractive gaseous exoplanets in various planetary parameters can be predicted.
A new 3D GCM grid is being introduced with 60 inflated hot jupiters orbiting A, F, G, K and M host stars modeled in exorad. A dense neural network (DNN) and decision tree algorithm (xgboost) are trained on this grid to predict local gas temperatures along with horizontal and vertical winds. To ensure reliability and quality of predictions in the ML model, the WASP-121 B, HATS-42 B, NGTS-17 B, WASP-23 B, and NGTS-1 B-like planets are selected and modeled in all targets, all Plato observation targets, and two ML methods as test cases. The DNN prediction of gas temperature is the extent to which the spectra calculated within 32 ppm for all but one planet agree on, with only one HCN feature where one HCN feature reaches a difference of 100 ppm.
The developed ML emulator can reliably predict the perfect 3D temperature field of Jupiter, which is very wet, from the warmth that swells around the A-M-type host star. It provides a fast tool to complement and extend traditional GCM grids for exporanet ensemble research. The quality of the prediction is such that it is expected to have or minimal effects on gas phase chemistry and thus cloud formation and transmission spectra.
Alexander Pushuzgu, Amit Leza, Ludmira Karon, Sebastian Garnjack, Christian Herring
Subjects: Earth and Planetary Astrophysics (Astro-Ph.ep); Machine Learning (cs.lg)
Quote: arxiv: 2508.10827 [astro-ph.EP] (Or arxiv: 2508.10827v1 [astro-ph.EP] For this version)
https://doi.org/10.48550/arxiv.2508.10827
Focus to learn more
Submission history
From: Amit Reza
[v1] Thu, August 14, 2025 16:50:38 UTC (8,186 kb)
https://arxiv.org/abs/2508.10827
Astrobiology,
