Extracting 21cm signal by frequency and angular filtering

TL;DR

This paper presents a method using Wiener filters in frequency and angular space to extract the faint 21cm hydrogen signal from dominant foregrounds and noise, demonstrating robustness despite beam uncertainties.

Contribution

The paper introduces a novel application of Wiener filtering in both frequency and angular domains for 21cm signal extraction, showing its effectiveness and robustness.

Findings

Method effectively separates 21cm signal from foregrounds and noise.

Robustness of the method despite beam knowledge inaccuracies.

Good reconstruction quality achieved in simulations.

Abstract

Extracting the neutral hydrogen (HI) signal is a great challenge for cosmological 21cm experiments, both the astrophysical foregrounds and the receiver noise are typically several orders of magnitude greater than the 21cm signal. However, the different properties of the 21cm signal, foreground, and noise can be exploited to separate these components. The foregrounds are generally smooth or correlated over the frequency space along a line of sight (l.o.s.), while both the 21cm signal and the noise varies stochastically along the same l.o.s. The foreground can be removed by filtering out the smooth component in the frequency space. The receiver noise is basically uncorrelated for observations at different times, hence for surveys they are also uncorrelated in the different directions, while the 21cm signal which traces the large scale structure are correlated up to certain scales. In this exercise, we apply Wiener filters in frequency and angular space, to extract the 21cm signals. We found that the method works well. The inaccurate knowledge about beam could degrade the reconstruction, but the overall result is still good, showing that the method is fairly robust.