2014MNRAS.440.3113H

This is the second of two papers describing the second data release (DR2) of the Australia Telescope Large Area Survey (ATLAS) at 1.4GHz. In Paper I, we detailed our data reduction and analysis procedures, and presented catalogues of components (discrete regions of radio emission) and sources (groups of physically associated radio components). In this paper, we present our key observational results. We find that the 1.4GHz Euclidean normalized differential number-counts for ATLAS components exhibit monotonic declines in both total intensity and linear polarization from millijansky levels down to the survey limit of ∼ 100 µJy. We discuss the parameter space in which component counts may suitably proxy source counts. We do not detect any components or sources with fractional polarization levels greater than 24 percent. The ATLAS data are consistent with a log-normal distribution of fractional polarization with median level 4 percent that is independent of flux density down to total intensity ∼ 10mJy and perhaps even 1mJy. Each of these findings are in contrast to previous studies; we attribute these new results to improved data analysis procedures. We find that polarized emission from 1.4GHz millijansky sources originates from the jets or lobes of extended sources that are powered by an active galactic nucleus, consistent with previous findings in the literature. We provide estimates for the sky density of linearly polarized components and sources in 1.4GHz surveys with ∼ 10arcsec resolution.