Solaris: A Foundation Model of the Sun
This addresses solar forecasting for researchers in solar physics, representing a novel application of foundation models in this domain.
The paper tackles the problem of forecasting the Sun's atmosphere by introducing Solaris, the first foundation model for this task, which leverages 13 years of solar imagery and outperforms models trained from scratch on a specific wavelength.
Foundation models have demonstrated remarkable success across various scientific domains, motivating our exploration of their potential in solar physics. In this paper, we present Solaris, the first foundation model for forecasting the Sun's atmosphere. We leverage 13 years of full-disk, multi-wavelength solar imagery from the Solar Dynamics Observatory, spanning a complete solar cycle, to pre-train Solaris for 12-hour interval forecasting. Solaris is built on a large-scale 3D Swin Transformer architecture with 109 million parameters. We demonstrate Solaris' ability to generalize by fine-tuning on a low-data regime using a single wavelength (1700 Å), that was not included in pre-training, outperforming models trained from scratch on this specific wavelength. Our results indicate that Solaris can effectively capture the complex dynamics of the solar atmosphere and transform solar forecasting.