Radio galaxy detection in the visibility domain

We explore a new Bayesian method of detecting galaxies from radio interferometric data of the faint sky. Working in the Fourier domain, we fit a single, parameterised galaxy model to simulated visibility data of star-forming galaxies, obtaining a multimodal posterior distribution, which is then sampled using a multimodal nested sampling algorithm such as MultiNest. For each galaxy, we construct parameter estimates for the position, flux, scale-length and ellipticities from the posterior samples. We test our approach on simulated SKA1-MID visibility data of up to 100 galaxies in the field of view considering two signal-to-noise ratio regimes: i) SNR~$\ge 10$ (a typical weak lensing survey threshold), and ii) SNR~$\ge 5$ (a typical threshold for galaxy catalog surveys), where 96% and 74% of the input sources are detected respectively with no spurious source detections in either regime. Comparing inferred galaxy parameter values with their input values to the simulation, we find our approach reliable in galaxy detection providing in particular high accuracy in positional estimates down to SNR$\sim 5$. The presented method does not require transformation of visibilities to the image domain, and requires no prior knowledge of the number of galaxies in the field of view, thus could be a useful tool for constructing accurate radio galaxy catalogs in the future.