Multi-frequency angular power spectrum of the 21~cm signal from the Epoch of Reionisation using the Murchison Widefield Array

TL;DR

This paper applies the Multi-frequency Angular Power Spectrum method to Murchison Widefield Array data from the Epoch of Reionisation, demonstrating its effectiveness in analyzing 21cm signals amidst foreground contamination and systematics.

Contribution

It introduces the application of MAPS to real EoR data, compares it with traditional power spectra, and discusses filtering techniques to mitigate foregrounds and systematics.

Findings

MAPS shows a contrast of 10^2--10^3 to simulated signals after filtering.

Residual spectral structure dominates systematic bias in the measurements.

Single-frequency angular power spectrum performs slightly worse than spherically-averaged power spectrum.

Abstract

The Multi-frequency Angular Power Spectrum (MAPS) is an alternative to spherically-averaged power spectra, and computes local fluctuations in the angular power spectrum without need for line-of-sight spectral transform. To test different approaches to MAPS and treatment of the foreground contamination, and compare with the spherically-averaged power spectrum, and the single-frequency angular power spectrum. We apply the MAPS to 110~hours of data in $z=6.2-7.5$ obtained for the Murchison Widefield Array Epoch of Reionisation experiment to compute the statistical power of 21~cm brightness temperature fluctuations. In the presence of bright foregrounds, a filter is applied to remove large-scale modes prior to MAPS application, significantly reducing MAPS power due to systematics. The MAPS shows a contrast of 10$^2$--10$^3$ to a simulated 21~cm cosmological signal for spectral separations of 0--4~MHz after application of the filter, reflecting results for the spherically-averaged power spectrum. The single-frequency angular power spectrum is also computed. At $z=7.5$ and $l=200$, we find an angular power of 53~mK$^2$, exceeding a simulated cosmological signal power by a factor of one thousand. Residual spectral structure, inherent to the calibrated data, and not spectral leakage from large-scale modes, is the dominant source of systematic power bias. The single-frequency angular power spectrum yields slightly poorer results compared with the spherically-averaged power spectrum, having applied a spectral filter to reduce foregrounds. Exploration of other filters may improve this result, along with consideration of wider bandwidths.