Robust Photometry for Roman High-Latitude Imaging Survey Cosmology Using Roman and Rubin Imaging

Accurate and precise photometry is essential for Roman cosmology because it affects photometric-redshift performance, galaxy and cluster selection, tomographic binning, and redshift-distribution characterization. Achieving robust photometry is challenging because Roman's depth leads to substantial source blending, particularly in joint Roman--Rubin analyses, where the two surveys have different spatial resolutions. In this work, we develop and validate slimfarmer, a model-fitting photometry pipeline designed for Roman High-Latitude Imaging Survey (HLIS) cosmological analyses. Building on The Farmer, slimfarmer introduces treatments of the correlated noise present in Roman images and of astronomical shot noise, together with model-fitting configurations tuned for Roman imaging. We validate the pipeline using Roman coadded simulated images and perform forced photometry on matched Rubin image simulations. We find that slimfarmer recovers galaxy colors with the mode of the color residuals within 20 millimag for Roman-based colors and within 70 millimag for Rubin-based colors. Properly accounting for correlated noise is essential for uncertainty quantification, as photometric uncertainties are otherwise underestimated by up to a factor of ~3. We also find that applying multi-object fitting consistently to both Roman and Rubin imaging substantially mitigates blending-induced color systematics. In crowded regions, single-object fitting produces environment-dependent color biases of up to ~0.5 mag, leading to environment-dependent degradation of photometric-redshift performance. Together, these results establish slimfarmer as a prototype photometric pipeline for Roman HLIS cosmology applications and identify the correlated-noise corrections, astronomical shot-noise treatment, and joint Roman--Rubin multi-object fitting required for robust photometric-redshift characterization.