The East Asian Observatory SCUBA-2 Survey of the COSMOS Field: Unveiling 1147 Bright Sub-millimeter Sources across 2.6 Square Degrees

Abstract

We present sensitive 850 µm imaging of the COSMOS field using 640 hr of new and archival observations taken with SCUBA-2 at the East Asian Observatory's James Clerk Maxwell Telescope. The SCUBA-2 COSMOS survey (S2COSMOS) achieves a median noise level of σ 850µm = 1.2 mJy beam −1 over an area of 1.6 sq. degree (main; Hubble Space Telescope / Advanced Camera for Surveys footprint), and σ 850µm = 1.7 mJy beam −1 over an additional 1 sq. degree of supplementary (supp) coverage. We present a catalogue of 1020 and 127 sources detected at a significance level of > 4 σ and > 4.3 σ in the main and supp regions, respectively, corresponding to a uniform 2 % false-detection rate. We construct the single-dish 850 µm number counts at S 850 > 2 mJy and show that these S2COSMOS counts are in agreement with previous single-dish surveys, demonstrating that degree-scale fields are sufficient to overcome the effects of cosmic variance in the S 850 = 2-10 mJy population. To investigate the properties of the galaxies identified by S2COSMOS sources we measure the surface density of nearinfrared-selected galaxies around their positions and identify an average excess of 2.0 ± 0.2 galaxies within a 13 radius (∼ 100 kpc at z ∼ 2). The bulk of these galaxies represent near-infrared-selected SMGs and / or spatially-correlated sources and lie at a median photometric redshift of z = 2.0 ± 0.1. Finally, we perform a stacking analysis at sub-millimeter and far-infrared wavelengths of stellar-massselected galaxies (M = 10 10 -10 12 M ) from z = 0-4, obtaining high-significance detections at 850 µm in all subsets (signal-to-noise ratio, SNR = 4-30), and investigate the relation between far-infrared luminosity, stellar mass, and the peak wavelength of the dust SED. The publication of this survey adds a new deep, uniform sub-millimeter layer to the wavelength coverage of this well-studied COSMOS field.

S2COSMOS: Evolution of gas mass with redshift using dust emission

“…to target the COSMOS (Cosmic Evolution Survey) field (Scoville et al 2007), a 2 deg 2 area of sky centred at RA = 10:00:28.60 and Dec = +02:12:21.00 (J2000). We make use of the 850µm map from the SCUBA-2 COSMOS survey (S2COSMOS, Simpson et al 2019), the deepest and most sensitive sub-mm image of the COSMOS field to-date. This map incorporates archival data from the SCUBA-2 Cosmology Legacy Survey (S2CLS, Geach et al 2017;Micha lowski et al 2017) and new data from S2COSMOS, providing a complete and homogeneous map of the COSMOS field at 850µm.…”

“…This analysis shows that our choice of source catalogue is not likely to significantly impact the results of this study, and choosing the CIGALE catalogue over the Driver/magphys catalogue would not change our conclusions. We ultimately choose to use the magphys dataset for the rest of this study, since it is limited to the central regions of the map ( Figure 1) where the noise levels are lowest (Simpson et al 2019).…”

“…Individually, most of the galaxies in our sample have 850µm fluxes below the noise level of the S2COSMOS map (Simpson et al 2019). To circumvent this issue, we make use of well-established stacking methodologies, wherein we co-add the emission from many similar sources to determine an average flux, which is representative of the sub-population in question.…”

“…For each (M * − z) bin, small cutouts of the 850µm S2COSMOS data and error maps (Simpson et al 2019) are made, centred on each individual source. The average 'stamp', S s,850µm , for sources in a bin is calculated using the inverse variance-weighted (IVW) mean of each of the individual cutouts:…”

“…To check the impact of galaxy clustering on our stacking results, we make use of the Python implementation of simstack (Viero et al 2013), a stacking and deblending method that attempts to account for any flux boosting introduced by galaxy clustering. We apply simstack to the non match-filtered S2COSMOS 850µm map (Simpson et al 2019) and our magphys-COSMOS2015 galaxy catalogue, and the sources are split into the same M * and z bins as for the IVW stacking. We use the nonmatched filtered map because SIMSTACK carries out a convolution with a Gaussian beam within its code.…”