Automated Exoplanetary Cartography: Generative Adversarial Networks for the Reconstruction of Speculative Tectonic and Climatic Features

The spatial characterization of terrestrial exoplanets remains severely constrained by current observational limitations, which typically yield only light curves, atmospheric transmission spectra, and basic orbital parameters. To bridge the gap between low-dimensional observational metrics and high-resolution planetary surface visualizations, this study introduces ExoCarto-GAN, a novel deep learning framework utilizing conditional Generative Adversarial Networks (cGANs) to synthesize physically plausible topographic, tectonic, and climatic surface maps of speculative exoplanets. Trained on a multidimensional dataset comprising Earth geomorphology, Mars elevation models, synthesized geophysical fluid dynamics (GFD) simulations, and thermodynamic equilibrium models, ExoCarto-GAN accepts key macro-astronomical inputs—such as stellar irradiance, planetary mass, surface temperature, atmospheric density, and rotation rate—to generate multi-layered raster maps. Our results demonstrate that the model successfully captures non-linear coupling between continental configuration, convective mantle dynamics, precipitation belts, and polar ice coverage. Synthetic validation against held-out numerical climate simulations demonstrates high spatial structural similarity (SSIM = 0.84) and physically consistent thermal gradients. This generative framework provides a powerful computational tool for visual communication, biosignature modeling, and prioritizing targets for future direct-imaging space missions.