Contamination of the Epoch of Reionization power spectrum in the presence of foregrounds

TL;DR

This paper models and analyzes how foreground emissions contaminate the Epoch of Reionization power spectrum, emphasizing the importance of joint foreground and signal estimation to avoid bias.

Contribution

It introduces a Bayesian framework for modeling foreground contamination in EoR power spectrum estimation, identifying optimal regions in k-space for signal recovery.

Findings

Foreground contamination is significant at low k_perp and k_parallel.

Joint estimation of foregrounds and EoR signal reduces bias.

Simulated observations recover intrinsic power spectra accurately.

Abstract

We construct foreground simulations comprising spatially correlated extragalactic and diffuse Galactic emission components and calculate the `intrinsic' (instrument-free) two-dimensional spatial power spectrum and the cylindrically and spherically averaged three-dimensional k-space power spectra of the Epoch of Reionization (EoR) and our foreground simulations using a Bayesian power spectral estimation framework. This leads us to identify a model dependent region of optimal signal estimation for our foreground and EoR models, within which the spatial power in the EoR signal relative to foregrounds is maximised. We identify a target field dependent region, in k-space, of intrinsic foreground power spectral contamination at low k_perp and k_parallel and a transition to a relatively foreground-free intrinsic EoR window in the complement to this region. The contaminated region of k-space demonstrates that simultaneous estimation of the EoR and foregrounds is important for obtaining statistically robust estimates of the EoR power spectrum; biased results will be obtained from methodologies that ignore their covariance. Using simulated observations with frequency dependent uv-coverage and primary beam, with the former derived for HERA in 37-antenna and 331-antenna configuration, we recover instrumental power spectra consistent with their intrinsic counterparts. We discuss the implications of these results for optimal strategies for unbiased estimation of the EoR power spectrum.