Machine learning techniques to separate the cosmic from the telluric

In the Roman era, wide-field, deep, visible-to-near infrared images will revolutionize our understanding of galaxy evolution (e.g. environments, morphologies, masses, colors). The legacy value of Roman images and low-resolution spectra (with Roman's prism and grism) will be greatly enhanced by massively multiplexed ground-based observations in the near - future and simultaneously allow us to leverage an impressive bounty of archived spectra from Maunakea facilities. We plan to enhance ground-based NIR spectra of astrophysically interesting objects with ground-sky spectra, atmospheric data, HST spectra and images, and machine learning techniques proven to predict galaxy spectra from images (Wu & Peek, 2020)[1].