Counting Stars: Developing Probabilistic Cataloging for Crowded Fields

Abstract

The depth of next generation surveys poses a great data analysis challenge: these surveys will suffer from crowding, making their images difficult to deblend and catalog. Sources in crowded fields are extremely covariant with their neighbors and blending makes even the number of sources ambiguous. Probabilistic cataloging returns an ensemble of catalogs inferred from the image and can address these difficulties. We present the first optical probabilistic catalog, cataloging a crowded Sloan Digital Sky Survey r band image cutout from Messier 2. By comparing to a DAOPHOT catalog of the same image and a Hubble Space Telescope catalog of the same region, we show that our catalog ensemble goes more than a magnitude deeper than DAOPHOT. We also present an algorithm for reducing this catalog ensemble to a condensed catalog that is similar to a traditional catalog, except it explicitly marginalizes over source-source covariances and nuisance parameters. We also study probabilistic cataloging's performance on a toy problem and identify a critical flux threshold that separates the likelihood-dominated and prior-dominated regimes. Finally, we detail efforts to make probabilistic cataloging more computationally efficient and extend it beyond point sources to extended objects. Probabilistic cataloging takes significant computational resources, but its performance compared to existing software in crowded fields make it a enticing method to pursue further.