Stellar Activity Characterization Through Advanced Clustering of Line Morphologies

Major advancements in the field of exoplanet science have been driven by enabling leaps in technology. With the advent of a new generation of extreme precision radial velocity (EPRV) instruments, stellar activity now presents the biggest obstacle to the discovery and mass measurement of small planets. We leverage the recently commissioned NEID spectrograph to perform a deep exploration of spectral activity manifestation using our best-monitored target — the Sun. NEID observes the disk-integrated Sun (as a star) for about 6 hours daily, through a dedicated solar feed. NEID’s exquisite line-spread function stability, coupled with its broad wavelength range, provide the potential to discriminate between astrophysical, instrumental, and observational variability. We perform an agnostic clustering of various line morphology parameters (e.g. depth, width, skew, integrated flux) using established unsupervised learning techniques (e.g. K-means clustering) to identify families of lines with common responses to activity. These clusters form the basis of a physics-driven classification of line behaviors, to connect individual lines to different types of stellar magnetic activity. An important by-product of this work is the generation of a list of relatively activity insensitive lines, that are best suited for searches of small exoplanets signatures in high-resolution spectra. In combining cutting-edge EPRV data with machine learning techniques, we thus aim to implement scalable, modern practices in the battle against stellar activity.