John R. Weaver
The COSMOS field has proved to be one of the cornerstone surveys in extragalactic Astronomy. Since our last public catalog in 2015, a wealth of new imaging and spectroscopic data has been collected in COSMOS. We describe the collection, processing, and analysis of the best available imaging data in COSMOS over 39 bands and produce a new reference photometric redshift. Powerful new constraints are made possible from the ultra-deep optical imaging with Hyper Suprime-Cam, UltraVISTA near-infrared Ks imaging at least 1 magnitude deeper than in 2015, and the inclusion of all Spitzer IRAC infrared imaging ever taken on COSMOS. We find significant improvement in our astrometric reference from 2015 by utilising Gaia DR1. From this reprocessed data we perform source detection and photometry for 2 million sources across 2 deg2 of the COSMOS field, using both traditional aperture photometric methods and a new profile-fitting photometric measurement tool we have developed. In addition to a detailed comparison of both methods, we report the achievement of sub-percent accuracy in our derived photometric redshifts at i < 22.5 AB and at 25 < i < 27 of less than 5 per-cent. Encouraged by this result, we will soon demonstrate the power of this ultra-deep multi-wavelength survey with a first look at the cosmological evolution of galaxy formation and growth with greater constraints than ever before possible.
This effort is spearheaded by John R. Weaver and Olivier Kauffmann, and is supported by a larger team based at the Cosmic Dawn Center in Copenhagen, the Institute d'Astrophysique de Paris, and the Laboratorie d'Astrophysique de Marseille.
This new catalog opens up a new dawn to the universe. We detect ~1,000,000 galaxies measured in 39 bands detected with ultra-deep near-infrared photometry over the 2deg2 of COSMOS. We perform photometry with two different and independent methods. The Classic approach follows the methods used in the 2015 edition of the catalog (Laigle et al. 2016) by determining fluxes as sums within apertures via the well-known SExtractor software. Additionally, we pursue an innovative approach with the new software called The Farmer (Weaver et al, in prep), which utilises The Tractor (Lang et al. 2016) to measure fluxes as fitted model parameters, hence deriving and fitting a model to each galaxy in the catalogue. Figure 3 illustrates the advantage of model-fitting photometry in recovering the correct flux for two nearby sources for which the aperture would be confused. Shown in Figure 4, comparisons of the derived photometry and colors yields excellent agreement at the 10% level.
We process both photometric catalogs with two independent photometric fitting codes that fit a set of galaxy templates to estimate redshift: Le Phare (Ilbert et al. 2006) and EAZY (Brammer et a. 2008). The resulting measurements of the "photometric" redshift are benchmarked against spectroscopically confirmed redshifts to demonstrate the remarkable accuracy achieved with both methods, finding accuracy <1% at 22.5 mag and <5% at 26 mag as shown in Figure 5. Additionally, we compare the catalogs and fitting codes to confirm extremely similar performances, as shown in Figure 6. The extraordinarily similar performances underscore the robustness of the derived redshifts.
Encouraged by this result, we outline four significant science investigations made possible by the uniquely deep + wide near-infrared photometry in COSMOS2020. This includes a first look at the cosmological evolution of galaxy growth (via the galaxy stellar mass function) with greater constraints than ever before possible, in addition to examing the first sites of galaxy star-formation cessation and "death" at high-redshift. Other significant science investigations are the UV luminosity function to explore the first galaxies and galaxy-galaxy magnification to directly probe the structure of dark matter.
The catalog is expected to be released publically by the end of the year. Details will be found in the coming paper Weaver, Kauffmann et al., in prep.
We are now applying the model fitting techniques pioneered here for ultra-deep fields to 20,000,000 galaxies in the first 20deg2 of the Cosmic Dawn Survey.