Interpolating a Grid of GCM-Simulated Tidally Locked Gaseous Exoplanets Using Machine Learning

With the development of ever-improving telescopes capable of observing exoplanet atmospheres, there is a growing demand for enhanced 3D climate models to support and help interpret observational data. However, the computationally intensive and time-consuming nature of General Circulation Models (GCMs) poses significant challenges for simulating a wide range of exoplanetary atmospheres. These challenges are further amplified by the need to rerun every simulation when altering the inner workings of the GCM, such as updating physical assumptions, which makes exploring new physical scenarios difficult.

Grid studies have been employed to explore parameter spaces, but this approach introduces additional complexity with each varying parameter. To address these limitations, a machine learning approach was applied to interpolate a grid of GCM simulations, done with the ExoRad package, representing hot Jupiters orbiting different host stars at varying distances. The performance of our machine learning frameworks in capturing 3D temperature and wind structures to bridge gaps in the model grid will be discussed. Furthermore, it will be explored how these predictions are reflected in simulated transmission spectra that compare to observational properties of space missions like CHEOPS, JWST and PLATO. This work is part of our science support efforts within the PLATO WPs 116700 and 116800.