Scouting for Clumps in our Cosmic Backyard: Detection and Analysis of the Local Clumpy Galaxy Population

Abstract

A key change in the galaxy population between z ~ 1 and z ~ 0 is that most high-redshift star-forming galaxies host “giant star-forming clumps” (or simply “clumps”), with luminosities ranging an order of magnitude above typical star-forming regions nearby. The origins, lifetimes, and evolutionary roles of these clumps all remain uncertain. To date, large populations of clumpy galaxies have only been observed at low resolution (1 kpc or larger). However, limited observations of strongly-lensed and low-redshift galaxies reveal that most clumps have significant 10-100 pc substructure, which can only be resolved by space telescopes at low redshift. Local clump detections are therefore needed to enable high-resolution follow-up. This thesis presents the first-ever large (N > 1,000), low-redshift (z < 0.1) catalog ofgiant star-forming clumps. It is based upon the results of the citizen science project Galaxy Zoo: Clump Scout, which recruited volunteers to detect clumps in nearly 60,000 Sloan Digital Sky Survey (SDSS) galaxy images. Volunteer detections were aggregated by a novel maximum-likelihood model designed specifically for citizen science. Finally, a Faster-RCNN object detection network was trained on this catalog and applied nearly 240,000 SDSS galaxy images. The final catalog contains thousands of clumps analogous to those in high-redshift galaxies, including over 100 that can be observed at sub-100 parsec resolutions with existing instrumentation. Using this catalog, I compute the “clumpy fraction” – the fraction of star-forming galaxies hosting at least one clump – at z ~ 0. I find an order-of-magnitude drop in this fraction between redshift 1.5 and 0 for a range of galaxy masses. This drop correlates most closely with the drop in star formation rate and galaxy turbulence over that time. Further, I find that the environmental density does not significantly impact the clumpy fraction when controlling for star formation rate. Together, this suggests that most clumps have formed in situ (from turbulent gas collapse) since at least z ∼ 1.5, and that clump formation slowed due to the reduction in gas fractions and accretion rates over that time.