Twenty-one centimeter tomography with foregrounds

TL;DR

This paper presents a new blind multifrequency method for extracting 21cm cosmological signals, effectively modeling and removing foreground contamination to enhance the potential of upcoming 21cm tomography experiments.

Contribution

It introduces a novel blind technique leveraging the smooth spectral structure of foregrounds, improving signal extraction in 21cm tomography.

Findings

Foregrounds limit scales k << 1h/Mpc

Noise limits scales k >> 1h/Mpc

Method remains promising despite extreme contamination

Abstract

Twenty-one centimeter tomography is emerging as a powerful tool to explore the end of the cosmic dark ages and the reionization epoch, but it will only be as good as our ability to accurately model and remove astrophysical foreground contamination. Previous treatments of this problem have focused on the angular structure of the signal and foregrounds and what can be achieved with limited spectral resolution (bandwidths in the 1 MHz range). In this paper we introduce and evaluate a ``blind'' method to extract the multifrequency 21cm signal by taking advantage of the smooth frequency structure of the Galactic and extragalactic foregrounds. We find that 21 cm tomography is typically limited by foregrounds on scales $k\ll 1h/$Mpc and limited by noise on scales $k\gg 1h/$Mpc, provided that the experimental bandwidth can be made substantially smaller than 0.1 MHz. Our results show that this approach is quite promising even for scenarios with rather extreme contamination from point sources and diffuse Galactic emission, which bodes well for upcoming experiments such as LOFAR, MWA, PAST, and SKA.