Stacking of large interferometric data sets in the image- and uv-domain -- a comparative study

We present a new algorithm for stacking radio interferometric data in the uv-domain. The performance of uv-stacking is compared to the stacking of fully imaged data using simulated Atacama Large Millimeter/sub-millimeter Array (ALMA) and the Karl G. Jansky Very Large Array (VLA) deep extragalactic surveys. We find that image- and uv-stacking produce similar results, however, uv-stacking is typically the more robust method. An advantage of the uv-stacking algorithm is the availability of uv-data post stacking, which makes it possible to identify and remove problematic baselines. For deep VLA surveys uv-stacking yields a signal-to-noise ratio that is up to 20 per cent higher than image-stacking. Furthermore, we have investigated stacking of resolved sources with a simulated VLA data set where 1.5" (10-12 kpc at z ~ 1-4) sources are stacked. We find that uv-stacking, where a model is fitted directly to the visibilities, significantly improves the accuracy and robustness of the size estimates. While scientific motivation for this work is studying faint, high-z galaxies, the algorithm analysed here would also be applicable in other fields of astronomy. Stacking of radio interferometric data is also expected to play a big role for future surveys with telescopes such as LOFAR and Square Kilometre Array (SKA).