Citizen Science and AI Tools for Roman Slitless Spectroscopy

The Nancy Grace Roman Space Telescope will measure weak gravitational lensing and galaxy clustering to study dark energy and cosmic acceleration. Accurate redshifts for millions of galaxies are required and will come from slitless spectroscopy. But slitless spectra across the field face overlapping spectra, contamination from nearby sources, and galaxy morphology that complicate redshift measurements; traditional one-dimensional extraction falls short, and forward modeling is too computationally costly for the expected data volumes.

This project develops an automated quality-validation pipeline that pairs citizen science with deep learning. Using the Zooniverse platform, volunteers will visually inspect and annotate two-dimensional slitless spectroscopic images, marking artifacts, contamination and real emission-line features.

Those annotations will form the training set for deep learning models.

roman telescope, slitless spectroscopy, weak lensing, galaxy clustering, redshifts, citizen science, zooniverse, deep learning, emission lines, contamination