2015MNRAS.449.1856J

We present a comprehensive model to predict the rate of spectroscopic confusion in H i surveys, and demonstrate good agreement with the observable confusion in existing surveys. Generically the action of confusion on the H i mass function was found to be a suppression of the number count of sources below the `knee', and an enhancement above it. This results in a bias, whereby the `knee' mass is increased and the faint end slope is steepened. For ALFALFA and HIPASS, we find that the maximum impact this bias can have on the Schechter fit parameters is similar in magnitude to the published random errors. On the other hand, the impact of confusion on the H i mass functions of upcoming medium depth interferometric surveys, will be below the level of the random errors. In addition, we find that previous estimates of the number of detections for upcoming surveys with Square Kilometre Array-precursor telescopes may have been too optimistic, as the framework implemented here results in number counts between 60 and 75 percent of those previously predicted, while accurately reproducing the counts of existing surveys. Finally, we argue that any future single dish, wide area surveys of H i galaxies would be best suited to focus on deep observations of the local Universe (z < 0.05), as confusion may prevent them from being competitive with interferometric surveys at higher redshift, while their lower angular resolution allows their completeness to be more easily calibrated for nearby extended sources.