Nature of Faint Radio Sources in GOODS-North and GOODS-South Fields. I. Spectral Index and Radio–FIR Correlation

Abstract

We present the first results from the deep and wide 5 GHz radio observations of the Great Observatories Origins Deep Survey (GOODS)-North (σ = 3.5 µJy beam −1 , synthesized beam size θ = 1.47 × 1.42 , and 52 sources over 109 arcmin 2 ) and GOODS-South (σ = 3.0 µJy beam −1 , θ = 0.98 × 0.45 , and 88 sources over 190 arcmin 2 ) fields using the Karl G. Jansky Very Large Array. We derive radio spectral indices α between 1.4 and 5 GHz using the beam-matched images and show that the overall spectral index distribution is broad even when the measured noise and flux bias are considered. We also find a clustering of faint radio sources around α =0.8, but only within S 5GHz < 150 µJy. We demonstrate that the correct radio spectral index is important for deriving accurate rest frame radio power and analyzing the radio-FIR correlation, and adopting a single value of α =0.8 leads to a significant scatter and a strong bias in the analysis of the radio-FIR correlation, resulting from the broad and asymmetric spectral index distribution. When characterized by specific star formation rates, the starburst population (58%) dominates the 5 GHz radio source population, and the quiescent galaxy population (30%) follows a distinct trend in spectral index distribution and the radio-FIR correlation. Lastly, we offer suggestions on sensitivity and angular resolution for future ultra-deep surveys designed to trace the cosmic history of star formation and AGN activity using radio continuum as a probe.

Connecting black holes and galaxies in faint radio populations at cosmic noon

“…This imaging reaches a depth of 0.32 μJy beam −1 with an angular resolution of 0.61 x0.31 , ∼ 10 times deeper and ∼ 2 times higher resolution then comparable surveys (i.e. Delvecchio et al (2017); Gim et al (2019)). The data reduction and source extraction will be presented in Rujopakarn, in prep.…”

The MOSDEF Survey: calibrating the relationship between H α star formation rate and radio continuum luminosity at 1.4 &lt; z &lt; 2.6

“…Gim et al (2019) illustrate that not accounting for this intrinsic variation can result in both significantly higher scatter in the FIRRC and a potential bias. However, our results are unlikely to be significantly affected by this scatter due to the use of stacked samples, while the magnitude bias measured by Gim et al (2019, ∆q FIR = −0.061) is not large enough to affect our conclusions.…”

“…Thirdly, in deriving the radio luminosity measurements we have assumed a constant canonical spectral slope of α = −0.8 (Condon 1992) when converting from flux to luminosity (or vice versa) throughout this work. Although this canonical value has been shown to be a valid assumption for the median spectral slope in SFGs out to high redshift (Calistro Rivera et al 2017;Gim et al 2019), there is still significant scatter around this value. Gim et al (2019) illustrate that not accounting for this intrinsic variation can result in both significantly higher scatter in the FIRRC and a potential bias.…”

“…Although this canonical value has been shown to be a valid assumption for the median spectral slope in SFGs out to high redshift (Calistro Rivera et al 2017;Gim et al 2019), there is still significant scatter around this value. Gim et al (2019) illustrate that not accounting for this intrinsic variation can result in both significantly higher scatter in the FIRRC and a potential bias. However, our results are unlikely to be significantly affected by this scatter due to the use of stacked samples, while the magnitude bias measured by Gim et al (2019, ∆q FIR = −0.061) is not large enough to affect our conclusions.…”

“…Given this close agreement, we choose not to apply any peak-tototal flux corrections to our forced radio flux measurements (the potential effects on our results of this assumption are discussed in Section 3.4). Finally, to allow direct comparison between our radio measurements and literature measurements of the SFR-radio relation at z ∼ 0, we convert our measured fluxes and luminosities from their central frequency of 1.525 GHz to 1.4 GHz assuming a spectral slope of α = −0.8 (a valid assumption for RC emission produced by star-formation, Calistro Rivera et al 2017;Gim et al 2019).…”