The Properties of Hitherto Unidentified Radio Sources in the MIGHTEE Survey

Radio astronomy is a pioneering branch of astronomy research - new radio surveys have gathered a lot of attention in the last few years, due to the huge expansion in capabilities of new radio facilities, with new observatories such as MeerKAT and LOFAR surveying wide areas to large depths at great speed. But radio sources alone are not especially informative, e.g. no redshift information can be obtained from radio continuum data, and therefore no clear route to obtaining information about the physics in the sources. This work investigates the properties of the no-ID radio sources taken from the MIGHTEE catalogue produced by Prescott et al. (submitted), specifically in the COSMOS field. The catalogue we use for the NIR data is described by McCracken et al. (2012) the UltraVISTA survey with deep and ultra-deep coverage of the COSMOS field. Combining the data from these two catalogues we are able to determine that the 114 no-ID sources have 141.8±31.4 excess galaxies when the separation is up to 6”. We find that these sources have a flux density distribution of 0.007 - 1.840 mJy, KS magnitude distribution of mainly 20 < mK < 24 and a redshift distribution of 0 < z < 2. The magnitude distribution includes only the sources we can detect, thus the rest must be fainter. We therefore suspect a bimodal magnitude distribution for the no-ID counterparts. Some fraction of the sources also have unknown redshifts because we have not been able to detect any counterparts even in the UltraVISTA data reaching mK < 26. Our results suggest that data such as from VIDEO are as good for cross-identifying MIGHTEE sources as UltraVISTA data, considering there is no significant difference between the source population when mK < 24 and mK < 26, which may benefit MIGHTEE investigations in the XMM-LSS, ELAIS-S1 and ECDFS fields. To further our investigation, we look at how many of the 114 sources would have at least one possible multi-wavelength counterpart using part of the Likelihood Ratio technique. We find that 60.3±16.1 per cent or around 68 of the no-ID sources have at least one counterpart detected in the UltraVISTA data. We also find that at search radius of 10”, there are still three sources with no possible counterparts. We agree with other works in the literature that combining different methods of cross-identification may be the most efficient and rewarding approach for radio-continuum surveys.